re: [CinCVS] Motion plugin slightly extended..

[Pierre Marc Dumuid]

Hi again,

I'm not on the mailing list.. (still not finished my PhD), but I saw
your emails in regards to the one I posted in the archives.

> When i compile after svn update i have this error:
> motionwindow.C: In member function 'int MotionWindow::create_objects()':
> motionwindow.C:175: error: 'xpmd' was not declared in this scope

oops!! fixed!  (my bad, from erraneous usage of git :S


> How is the second region represented in the GUI?
> (yes, we want pictures! :-)

Hmm, sorry not sure what you're saying.. some demo's of the source and
outputs are shown in the files at:

http://cvs.cinelerra.org/~pmdumuid/index.php?dir=motiondemo%2F

motdemo_nostab_noaddprev.ogg  - the result without any "addprevious"
checkbox set or stabilization
motdemo_stab_addprev.ogg  - the result with "addprevious" and
stabilization
motionDemo2007_12_22.cine.xml - the cinelerra file
p1010110.mov  - the source (from my camera: Panasonic
Lumix DMC-FX12 [no zoom / focusing while shooting :S])
rotatemot_trial.mov   - another example.

BTW: My main use was to actually stabilize and crop footage that was
supposed to be steady when recorded (as opposed to the above, which was
just me mucking around..)

>Since Christmas is nigh, I won't mind a merry debate on how to
>enhance the motion tracker's usability. Pierre just enhanced its
>power and utility.

Hmm, I'm not big on debates, particularly when they involve lots of
thought, with little time left for
implementation. However, my main gripes are:

* Using thousands of small files such as: /tmp/motion??
 - results in an inability to have two tracks with a motion effect being
used..
 - Ideally a single file could be used, (and specified)

* rotation compensation uses area round point of rotation..
 - for maximum accuracy, the angle should be determined using
examination of a region as far away from the point of rotation.
 - I was actually considering replicating the motion plugin, and
actually tracking 2 points... and then using those two points to
determine translation and rotation..

However, my procrastination stint is over..

:P

Pierre

BTW: please CC' me your responses.. (as I'm not on the mailing list !!)

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

[CinCVS] Motion plugin slightly extended..

[Pierre Marc Dumuid]

Hi All (long time no contact),

just lettting you know about my commit to SVN:

PLUGIN:Motion - Ability to add offsets from tracked frame


To stabilize video, the motion plugin uses a "tracking frame" to which
to track to and a region within that frame to track (generally an
object in the background) in the current frame. When the region is
obscured, (often by something in the foreground, or by leaving the
screen), then the motion compensation would fail, and the video jump's
all over the place.

This patch helps work around the problem by allowing one to set a
second region to track and then add offset the (previously calculated and
saved to the respective file in /tmp/???/)

Here's how I use it:

--k---k-
  ^  ^^
  A  BC

A - object1 is visible in the background up until C
B - (the frame before C) has both object1 and object2 visible
C - has only object2 visible

1. Make a keyframe and set to track object1
2. Make a keyframe at C and track frame at B, set to track object2
3. Set keyframe at C to add offsets that were calcuated at B.

Done! And it works REALLY well!!! (I am really happy with this! :P)

--

Also, I recently pushed some other patches onto my pipapo branch that I
didn't think were right for the main branch:

http://pipapo.org/gitweb?p=cinelerra/pmdumuid;a=shortlog;h=recent2007Dec

As before, I'm still not finished with my PhD and had a bit of a
procrastination week... (well, it's christmas...)

Pierre

--
"What a charming amusement for young people this is, Mr. Darcy!
There is nothing like dancing after all. I consider it as one of
the first refinements of polished society."

"Certainly, sir; and it has the advantage also of being in vogue
amongst the less polished societies of the world. Every savage can
dance."

Pride and Prejudice


___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

[CinCVS] most of the php files are all static!

[Pierre Marc Dumuid]

Hi,

I was wondering if anyone know's why (a number of) the php files are all 
static at the website?


Some time ago I moved the html files to php files, and moved the 
sidemenu index to a single php file that was included (so editing of the 
side menu only needed to be performed once.)  For some reason this has 
been undone?


Pierre

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

[CinCVS] Re: [CVS] SVN Commit History

[Pierre Marc Dumuid]

Hi,

rafael2k wrote:
> hi all,
> I think the svn commit history stopped at r1005...
>
> bye!
> rafael diniz

FIXED.

I am still not on the mailing list  (soo please reply to my address, not 
the list address) but I browse archives from time to time and noticed 
this message.  I just logged onto skolelinux, and found that the php 
file that I spent a little while writing has been overwritten with 
static content!


Luckily, at one stage I had gitified that directory so have managed to 
restore it.


j6t: BTW: the git binaries in ~j6t/bin/ aren't working anymore, I had to 
scp the .git dir home to extract the file... if you recompile, I'd be 
most appreciative..


ta,
Pierre

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

[CinCVS] Bugzilla Image

[Pierre Marc Dumuid]

Hi Richard,

I just modified the picture at http://bugs.cinelerra.org/  to show all 
three themes, (rather than just bluedot)
I always found it funny that bluedot was the theme used at bugzilla, and 
till yesterday hasn't been usable since 2.0...


Pierre

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

[CinCVS] Bluedot theme

[Pierre Marc Dumuid]

Hi Miha / cinelerra-cv,

I have just committed Miha's bluedot theme branch into SVN (after a few 
more commits)


Please email me / cc me in any problems that result from the commit. (as 
I am still not on the mailing list)


Regards,
Pierre


___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

[CinCVS] no-longer on the mailing list.

[Pierre Marc Dumuid]

Hi All,

Just in case anyone poses any queries on the mailing list for me, I 
thought I'd advise that I'm unsubscribing myself from the mailing list.


I am suffering from sever procrastination in not wishing to finish my 
PhD, so have thought it best to un-subscribe myself.

If you need to contact me, please email me directly, (or cc' me).

I know there are a few tasks that I were working on, in particular, 
neatening and merging Hermann Vosseler beizer patch, but I feel that I 
really need to close off all distractions and focus on my work.


Hope all goes well, and will join you again at a later stage,

Regards,
Pierre Dumuid 




___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] Thanks! [motion tracking is very cool...]

[Pierre Marc Dumuid]

In my case the rendering (with farm) and then putting pices together
took more time than the method explained above...
  

I had the same experience..

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] setup changes for git users on pipapo.org

[Pierre Marc Dumuid]

Hi Christian,

I'm happy for you to delete the old cinelerra-USERNAME directories 
whenever you desire... It'll force me to update the respective references..


Pierre

Christian Thaeter wrote:

I made some changes in the server setup, which should ease mamagement
and allows anyone to setup repos on his own. For now it's all done in a
way that it is compatible with the old setup, when something gets
broken, notify me, that is a bug.

I create a page on my wiki describing the new setup:
 http://www.pipapo.org/pipawiki/Cinelerra/ServerSetup

Please use that when you create a new repository.
Existing anonymous repository access which is available as
cinlerella-USERNAME are now deprecated in favor of cinelerra/USERNAME.
I'll fix the wiki pages explaining that, explain that in new documents
and change your registered archives. Anyways the old names are still
reachable, maybe I remove them sometime in future (in a few months or
so) but nothing going to be broken soon.

Christian

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra
  


___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] r975 issues

[Pierre Marc Dumuid]

Hi Miha,

Miha Kitič wrote:

Hi All!

I have two points...

1. Yesterday I tried to compile the latest SVN.

After running ./configure the strange error appeared - I don't remember
now and I can not repeat it... but it was something involving
mjpegtools...
something like 'syntakse error' or unrecognized keyword  (MJPEG or
sth.).

It must be r975 issue, because when I checked r960 and run ./configure,
everything went through smoothly...

What may that be??? An error or a new feature?
  


"can not repeat" --> I cannot help you solve the problem.
BTW: the 975 commit doesn't even touch the configure script, so that 
can't be the culprit.
It could the that you re-ran ./autogen.sh and that fixed the problem and 
made the problem un-repeatable.


If you manage to get it to display the error again, please cut'n paste 
the error message.



2. This is more a joke than a seriouse thing...
  

:)

BTW: did you get my message about bluedot theme? (about re-basing and 
fixing the icon's that are missing)


PIerre

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

[CinCVS] cinelerra/fileyuv.C not translated..

[Pierre Marc Dumuid]

For those managing the translation files, I just noticed (whilst 
increasing the extent of what widgetgrid covers) that fileyuv.C isn't 
covered by the POTFILES.in, and hence untranslated.


Pierre

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] http://www.pipapo.org/.cinelerra-svn_sync/authors

[Pierre Marc Dumuid]

Hi Heroine,

No, I don't think it was a jab at  Jack Crossfire/Heroine Virtual Ltd.

This list is not in the official release and is simply the list of 
author's that commit to the SVN repository, and is not to be a 
representation of the author's of cinelerra, (to whom we acknowledge 
that you are the main one!)


This file is used only for a script that fetches each commit into the 
SVN and makes a similar commit on the git repository at pipapo.org.


This file is the relationships file between the SVN committer's and 
their name is real-life and is used in the script that automatically 
commits to the repository.


(some people consider that git may be a way to develop ideas prior to 
pushing them into the SVN repository, but most people are used to the 
SVN...)


Please also be aware that the commits during the merger's (see 
http://cvs.cinelerra.org/svn_log.php?rev=2_0merge2_1 for example) are 
really all your work (broken down into feature-wise pieces), and the 
committer's name in this instance, (j6t) is merely because we didn't 
make an author up as a user on cvs.cinelerra.org.


If you have further concerns, or feel jabbed in any other way, please 
let us know as there is seriously no jabbing meant. 


As a friend once told me:

If you took afence, please put it back, as the fence was never meant to 
be taken!


Pierre

Heroine Virtual Ltd. wrote:



Johannes Sixt <[EMAIL PROTECTED]>
minmax=Andraz Tori <[EMAIL PROTECTED]>
herman=Herman Robak <[EMAIL PROTECTED]>
baver=Richard Baverstock <[EMAIL PROTECTED]>
pere=Petter Reinholdtsen <[EMAIL PROTECTED]>
tfheen=Tollef Fog Heen <[EMAIL PROTECTED]>
andreask=Andreas Kielb <[EMAIL PROTECTED]>
theraz=Tyler Geddes <[EMAIL PROTECTED]>
dyce=Gergely Erdelyi <[EMAIL PROTECTED]>
dreamlx=David Arendt <[EMAIL PROTECTED]>
ga=Gustavo Iniguez <[EMAIL PROTECTED]>
memenk=Michael Eric Menk <[EMAIL PROTECTED]>
benjif=Benjamin Flaming <[EMAIL PROTECTED]>
cobra=Kevin Brosius <[EMAIL PROTECTED]>
taraba=Mark Taraba <[EMAIL PROTECTED]>
nate=Nathan Kurz <[EMAIL PROTECTED]>
mammique=Camille Harang <[EMAIL PROTECTED]>
kbielefe=Karl Bielefeldt <[EMAIL PROTECTED]>
alexf=Alex Ferrer <[EMAIL PROTECTED]>
pmdumuid=Pierre Dumuid <[EMAIL PROTECTED]>
giskard=Riccardo Setti <[EMAIL PROTECTED]>
jstewart=Joe Stewart <[EMAIL PROTECTED]>
doudou=Sylvain Joyeux <[EMAIL PROTECTED]>
rafael2k=Rafael Diniz <[EMAIL PROTECTED]>



Was that some kind of jab at Jack Crossfire/Heroine Virtual Ltd?
Whether you agree with it or not, you should credit original authors if
you copy their work.





___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] Cinelerra build problems

[Pierre Marc Dumuid]

Thanks also for the reference to Fedora, I will take a look at the 
packages there and see if I can figure out what the difference is.
The problem was fixed during the merger, I recall it being an incorrect 
type of escape character.  But bug's we find don't automatically create 
a new release upstream, and this is actually a build-system bug, which 
we have completely replaced and used automake, autoconf etc..




Bob

Pierre Marc Dumuid wrote:
That bug is know but this mail list is dedicated to cinelerraCV not 
cinelerra from HW.  (P.S. it made fine on Fedora which is what HW uses)


You must be behind a firewall as I could checkout today myself using 
the command described at cvs.cinelerra.org.


Pierre

Bob McGowan wrote:

I've downloaded the Cinelerra source from:

http://heroinewarrior.com/download.php3

I run configure without any issues.  When I run make, I get the 
following error:


$ make
make -f build/Makefile.cinelerra
sh: -c: line 1: syntax error: unexpected end of file
make[1]: Entering directory 
`/home/rmcgowan/src/multimedia/cinelerra-2.0'
gcc -c -O2 -fomit-frame-pointer -falign-loops=2 -falign-jumps=2 
-falign-functions=2 -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE 
-D_FILE_OFFSET_BITS=64 -I../../freetype-2.1.4/include -I../../ 
-DHAVE_OSS -DHAVE_FIREWIRE  soundtest.c -o i686/soundtest.o

Assembler messages:
FATAL: can't create i686/soundtest.o: No such file or directory
make[1]: *** [i686/soundtest.o] Error 1
make[1]: Leaving directory 
`/home/rmcgowan/src/multimedia/cinelerra-2.0'

make: *** [all] Error 2

The reason is that the directory 'i686' does not exist.  On 
investigation, I find this line in the file 'global_config':


  $(shell sh -c 'if ! test -d $(OBJDIR)\; then mkdir $(OBJDIR)\; fi' )

Which, as far as I can tell, is supposed to create directory.  The 
value of OBJDIR may be 'i686', based on an 'if' test near the 
beginning of the file that sets the value of PLUGINDIR.


Note that the first error is in regard to 'sh: -c: ...', so I've 
experimented with this type of line in a test makefile.  If I remove 
the backslashes, and actually assign the non-existent output to a 
variable, and so on, I can eventually get it to create a directory.


It would seem to me that this is such an obvious problem that it 
would have been found and fixed quickly.  Since it hasn't, I wonder 
if the build setup is designed for use with some other 'make' 
program than the standard Gnu make?


As it happens in a half dozen different places and I've not been 
able to grep out the locations of all of them, I'm stuck.


Any suggestions welcome.

FYI, I thought if I could get the source from SVN, I might be able 
to get around this, but my connection to the svn server is 
rejected.  This may be a firewall problem so I'll be trying to do 
the same checkout of source later from a different system that 
hopefully has no firewall problems.  Meaning this may become moot 
for me.  ;-)


Thanks,

Bob



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] Cinelerra build problems

[Pierre Marc Dumuid]

That bug is know but this mail list is dedicated to cinelerraCV not 
cinelerra from HW.  (P.S. it made fine on Fedora which is what HW uses)


You must be behind a firewall as I could checkout today myself using the 
command described at cvs.cinelerra.org.


Pierre

Bob McGowan wrote:

I've downloaded the Cinelerra source from:

http://heroinewarrior.com/download.php3

I run configure without any issues.  When I run make, I get the 
following error:


$ make
make -f build/Makefile.cinelerra
sh: -c: line 1: syntax error: unexpected end of file
make[1]: Entering directory `/home/rmcgowan/src/multimedia/cinelerra-2.0'
gcc -c -O2 -fomit-frame-pointer -falign-loops=2 -falign-jumps=2 
-falign-functions=2 -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE 
-D_FILE_OFFSET_BITS=64 -I../../freetype-2.1.4/include -I../../ 
-DHAVE_OSS -DHAVE_FIREWIRE  soundtest.c -o i686/soundtest.o

Assembler messages:
FATAL: can't create i686/soundtest.o: No such file or directory
make[1]: *** [i686/soundtest.o] Error 1
make[1]: Leaving directory `/home/rmcgowan/src/multimedia/cinelerra-2.0'
make: *** [all] Error 2

The reason is that the directory 'i686' does not exist.  On 
investigation, I find this line in the file 'global_config':


  $(shell sh -c 'if ! test -d $(OBJDIR)\; then mkdir $(OBJDIR)\; fi' )

Which, as far as I can tell, is supposed to create directory.  The 
value of OBJDIR may be 'i686', based on an 'if' test near the 
beginning of the file that sets the value of PLUGINDIR.


Note that the first error is in regard to 'sh: -c: ...', so I've 
experimented with this type of line in a test makefile.  If I remove 
the backslashes, and actually assign the non-existent output to a 
variable, and so on, I can eventually get it to create a directory.


It would seem to me that this is such an obvious problem that it would 
have been found and fixed quickly.  Since it hasn't, I wonder if the 
build setup is designed for use with some other 'make' program than 
the standard Gnu make?


As it happens in a half dozen different places and I've not been able 
to grep out the locations of all of them, I'm stuck.


Any suggestions welcome.

FYI, I thought if I could get the source from SVN, I might be able to 
get around this, but my connection to the svn server is rejected.  
This may be a firewall problem so I'll be trying to do the same 
checkout of source later from a different system that hopefully has no 
firewall problems.  Meaning this may become moot for me.  ;-)


Thanks,

Bob



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] AUTHORS file in svn

[Pierre Marc Dumuid]

Hi Christian,

We could make a file called, "SVN-COMMITTERS".  Though you need to fetch 
the file before you run git-svn ...


Also, I seem to be the quickest to note that the sync isn't working, so 
if you make the author file on pipapo cinelerra-group writable, and give 
access to the output logs, then I could fix it as soon as I see a 
problem also...


PIerre


Yes, the Syntax needs to be conform what git wants, while the GNU coding
standards mandate another syntax. Currently the AUTHORS file is empty
which has no use for either git nor gnu coding standards. So far I dont
know any other tool which parses the AUTHORS file which is in moderate
frequent use (correct me if I am wrong). So having a readable but
slightly unexpected syntax is not problematic.

The authors file I need has to contain at least all svn committers, but
there is in no way a restriction to to add more than that. I would
strongly opt for honor anyone ever worked on cinelerra there, also from
the HV/upstream list.

I think this would be far better than the current no-solution, maybe not
perfect but I see absolutely no harm with it.

Finally, this thing isn't really worth much discusion. I can keep the
'authors' file and fix it whenever someone new appears as committer.

Christian


___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra
  



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] Annoyance: changing media type changes file name

[Pierre Marc Dumuid]

I've found it annoying as well.
I personally would prefer cinelerra to remember the name up until the 
last dot (i.e. the filename w/out the extension).


The makes sense for:
* when rendering audio then video seperately
*when rendering a single project to multiple formats

The filename should not change, but the extension should...

Pierre

Johannes Sixt wrote:
In the Render dialogs the media type (.mov, .avi, RAW DV etc) can of course be 
changed. Cinelerra CV remembers a different file name for each media format.


I find it most annoying that changing the media type also switches the file 
name, in particular, because when you work through the dialog top to bottom, 
the first thing to choose is the file name, then comes the media - but when 
you change it, it kills the file name that you have just selected.


I'd like to kill the feature. Nate? You've added this 2 years ago.

Has someone else been annoyed by this behavior?
Has someone found this feature extremely useful?

-- Hannes

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra
  



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] What Does This Mean ??

[Pierre Marc Dumuid]

Nicolas et al.,
Rumours have it that somebody here 
already worked on ways to link documentation pagers into the 
GUI.  Is that workable?
  

yep, it is still being developed,
see the #helpsystem on my git repository at pipapo.org 
(http://pipapo.org/gitweb?p=cinelerra-pmdumuid;a=shortlog;h=helpsystem)


You'll need to copy tar extract 
http://cvs.cinelerra.org/~pmdumuid/usr_share_gnome_help_cinelerra.tar.bz2
into your gnome/help directory on your system for it to work , (and have 
yelp installed)


This is by far not the final solution, but is a means for nicolas to 
play around with (he'll likely use numbers as references instead of text 
references so that it is language independant.


I've also worked a dcop script somewhere that somehow managed to use 
push kpdf to jump to the specific page, (in the demo above, a new help 
window appears for each ...)


And before people complain, I am a kde person, and would rather a kde 
document system... The intention is to later a variety of possibilities 
for getting the documentation, (web browsers etc..)  That is why it 
isn't in the main branch...


afternoon all...

Pierre

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] Exporting to mpeg video..

[Pierre Marc Dumuid]

This still seems the wrong way to go to me.  Currently I think packagers
should install whatever is required on their distro to make this work. 
That's what the SUSE packages do.
  

I was aware of what SUSE did (the symlink), I though that was a bit silly.

Or cinelerra should use the default binary name.  Making an in program
path location is working around it in the wrong place, in my opinion.
  
Placing a "binary location" seemed a way of letting the end user know 
that they needed to have the binary and provide the location to it and 
that the mpeg2enc.plugin wasn't anything but the binary itself, (.plugin 
could make it seem like it's a library object or something.)


Should just go with what is returned by `which mpeg2enc`  (i.e. search 
default path) and remove the need for a special " "mpeg2enc.plugin" file?





___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] Exporting to mpeg video..

[Pierre Marc Dumuid]

I was just writing a patch to save a "mpeg2enc" binary location.
I was going to make the location settable in the Render->Video Settings 
window, but realised that that dialog, and the code surrounding it seems 
to only handle modification to the asset, and I didn't really want the 
binary location to vary between assets for different projects.


I'm guessing that it should go in preferences dialog, but non of the 
tabs make sence, (it's not a recording or playback option)  Any one have 
an idea were it should go, or if it is a good idea.


I was also going to commit with the same patch, and little note saying 
people should try using yuv4mpeg pipe to have a finer control.


Pierre

Herman Robak wrote:

On Mon, 2006-12-18 at 23:15 -0700, Joe Friedrichsen wrote:
  

On 12/18/06, Wesley T Allen <[EMAIL PROTECTED]> wrote:


Hi all...

I'm having difficulty exporting to mpeg video using the yuv 4:2:0  I get
errors no matter what and, while the other yuv setting allows me to render
mpeg2 - it crashes everything but mplayer...


Here's my problem:
"/usr/local/lib/cinelerra/mpeg2enc.plugin: not found"
  

I feel like I have much better control when I use the yuv4mpeg pipe. I
can specify the exact options that I want, and even include other
parts of mjpegtools when necessary.



 Thanks for pointing out the workaround; it is good for users to know.


However, "mpeg2enc.plugin: not found" is a _bug_.  If it is not in
bugs.cinelerra.org already, it is time to file a bug report on it.


  



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] a bug: clip tail drag as long as you want

[Pierre Marc Dumuid]

I confirm it, please file a bug report..

Pierre

xes garcia wrote:

Hi
i've found a bug
If you load a clip in the timeline and move it a little with the arrow 
of Drag and drop editing mode then you can inmediatly scale the end of 
the clip as long as you want no matters how much long the clip is.

so:
load a clip in timeline
move it a little
drag the tail of the clip as long as you want

It's in last svn, revision 959
Ubuntu edgy 32bit

Can somebody tellme if this happens them too?

bye

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] hd test files - or bringing cincvs to the knees

[Pierre Marc Dumuid]

I had trouble downloading the scene_1_shot10 clip, but downloading 
scene_1_shot2 clip I got
between 24->28 fps when compositor was at a scale of 50 % on a Intel(R) 
Pentium(R) D  CPU 2.66GHz


When at 100% I get 19 fps with the picture going over both screens (dual 
monitor) and 24-25 fps when compositor on 1 screen, (but whole picture 
not visible)


Video Driver: X11-XV
Play every frame is selected
Decode frames asyncronously unselected (REALLY slow when selected!)
Video Card: nVidia Corporation GeForce 6200
Dual monitor, both @ 1280x1024
nvidia module version = 1.0.9629

/sbin/hdparm -t -T /dev/hda
/dev/hda:
Timing cached reads:   1884 MB in  2.00 seconds = 941.43 MB/sec
Timing buffered disk reads:  214 MB in  3.02 seconds =  70.90 MB/sec

Using Fedora Core 6, and livna repository to satisfy required libraries.

Compiling cinelerra with the following script:

#!/usr/bin/bash
echo configureing for pentium4
# --with-external-ffmpeg
../hvirtual/configure --prefix=/usr --enable-x86 --enable-mmx 
--enable-freetype2 \

 --with-buildinfo=svn \
   CFLAGS='-O3 -pipe -fomit-frame-pointer -funroll-all-loops 
-falign-loops=2 -falign-jumps=2 -falign-functions=2 -ffast-math 
-march=nocona -mfpmath=sse,387 -mmmx -msse' \
   CPPFLAGS='-O3 -pipe -fomit-frame-pointer -funroll-all-loops 
-falign-loops=2 -falign-jumps=2 -falign-functions=2 -ffast-math 
-march=nocona -mfpmath=sse,387 -mmmx -msse' \
   CXXFLAGS='-O3 -pipe -fomit-frame-pointer -funroll-all-loops 
-falign-loops=2 -falign-jumps=2 -falign-functions=2 -ffast-math 
-march=nocona -mfpmath=sse,387 -mmmx -msse'



Though /proc/cpuinfo shows me that I don't have sse3 (which nocona 
requires!) so I am not too sure of the stability..


PIerre

Jan Luo wrote:
o.k. is there any way to understand what the bottlenecks on the 
hardware side

are. like cpu, hd-speed or ram and so on?

cheers

jan

Nicolas Maufrais wrote:

On Sat, Nov 18, 2006 at 02:24:29AM +0100, Jan Luo wrote:
 
found the following link in the hdforindies blog with ( in my 
understanding ) hd clips:


http://www.siliconimaging.com/DigitalCinema/gallery_footage.html

when i play the scene_1_shot10 clip / MPEG-2 19Mb/s 1080/24P (89MB)
i get frame rates around 17 fps. my box is a dual amd64 4800. can 
anyone play these

at 30 fps? on which hardware?

thanks

jan



I got 13fps on my AMD64 X2 4600.

Nicolas.

  


___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra


___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] [Bug 360] New: consistent list of items in View & Show Overlays

[Pierre Marc Dumuid]

Hey Andraz,

Attached is a "bit more work" patch that makes it match now..  I'm 
wondering if some of the things I did are a non-standard way of doing 
things, so could you review the patch please?



Modifying the order of the menubar results in the numbers not being in 
order!  Do people mind if we change the keyboard shortcuts so the 
numbering is in order?



That's a bit nasty, couldn't we put a bit of extra more work and change
the way overlay window works?

  

done..

and while we are at it: i think in overlay window, each checkbox should
also have min and max range values there which should be editable... :)
  
If I was to do this, I'd want to use my widgetgrid class.. though that 
the same time, my desktop real estate isn't that large as it is, let 
alone adding textboxes!


Though if I used widgetgrid I'd probably make something so you could 
hide the ranges... I'll look into it sometime..
diff --git a/cinelerra/gwindowgui.C b/cinelerra/gwindowgui.C
index ebe4bed..d9b3961 100644
--- a/cinelerra/gwindowgui.C
+++ b/cinelerra/gwindowgui.C
@@ -31,7 +31,7 @@ GWindowGUI::GWindowGUI(MWindow *mwindow,
 	this->mwindow = mwindow;
 }
 
-static char *other_text[OTHER_TOGGLES] =
+static char *other_text[NONAUTOTOGGLES_COUNT] =
 {
 	"Assets",
 	"Titles",
@@ -55,13 +55,36 @@ static char *auto_text[] = 
 	"Nudge"
 };
 
+#define ISAUTO  0x1000
+#define NOTAUTO 0x2000
+
+static int auto_reorder[] = 
+{
+
+	NOTAUTO + NONAUTOTOGGLES_ASSETS,
+	NOTAUTO + NONAUTOTOGGLES_TITLES,
+	NOTAUTO + NONAUTOTOGGLES_TRANSITIONS,
+	ISAUTO  + AUTOMATION_FADE,
+	ISAUTO  + AUTOMATION_MUTE,
+	ISAUTO  + AUTOMATION_MODE,
+	ISAUTO  + AUTOMATION_PAN,
+	NOTAUTO + NONAUTOTOGGLES_PLUGIN_AUTOS,
+	ISAUTO  + AUTOMATION_MASK,
+	ISAUTO  + AUTOMATION_CAMERA_X,
+	ISAUTO  + AUTOMATION_CAMERA_Y,
+	ISAUTO  + AUTOMATION_CAMERA_Z,
+	ISAUTO  + AUTOMATION_PROJECTOR_X,
+	ISAUTO  + AUTOMATION_PROJECTOR_Y,
+	ISAUTO  + AUTOMATION_PROJECTOR_Z,
+};
+
 void GWindowGUI::calculate_extents(BC_WindowBase *gui, int *w, int *h)
 {
 	int temp1, temp2, temp3, temp4, temp5, temp6, temp7;
 	int current_w, current_h;
 	*w = 10;
 	*h = 10;
-	for(int i = 0; i < OTHER_TOGGLES; i++)
+	for(int i = 0; i < NONAUTOTOGGLES_COUNT + AUTOMATION_TOTAL; i++)
 	{
 		BC_Toggle::calculate_extents(gui, 
 			BC_WindowBase::get_resources()->checkbox_images,
@@ -75,29 +98,12 @@ void GWindowGUI::calculate_extents(BC_Wi
 			&temp5, 
 			&temp6,
 			&temp7, 
-			other_text[i]);
+			auto_reorder[i] & ISAUTO ? 
+auto_text[auto_reorder[i] - ISAUTO] : other_text[auto_reorder[i] - NOTAUTO]);
 		*w = MAX(current_w, *w);
 		*h += current_h + 5;
 	}
 
-	for(int i = 0; i < AUTOMATION_TOTAL; i++)
-	{
-		BC_Toggle::calculate_extents(gui, 
-			BC_WindowBase::get_resources()->checkbox_images,
-			0,
-			&temp1,
-			¤t_w,
-			¤t_h,
-			&temp2,
-			&temp3,
-			&temp4,
-			&temp5, 
-			&temp6,
-			&temp7, 
-			auto_text[i]);
-		*w = MAX(current_w, *w);
-		*h += current_h + 5;
-	}
 	*h += 10;
 	*w += 20;
 }
@@ -109,28 +115,16 @@ void GWindowGUI::create_objects()
 	int x = 10, y = 10;
 
 
-	for(int i = 0; i < OTHER_TOGGLES; i++)
+	for(int i = 0; i < NONAUTOTOGGLES_COUNT + AUTOMATION_TOTAL; i++)
 	{
-		add_tool(other[i] = new GWindowToggle(mwindow, 
+		add_tool(toggles[i] = new GWindowToggle(mwindow, 
 			this, 
 			x, 
 			y, 
-			-1,
-			i, 
-			other_text[i]));
-		y += other[i]->get_h() + 5;
-	}
-
-	for(int i = 0; i < AUTOMATION_TOTAL; i++)
-	{
-		add_tool(auto_toggle[i] = new GWindowToggle(mwindow, 
-			this, 
-			x, 
-			y, 
-			i,
-			-1, 
-			auto_text[i]));
-		y += auto_toggle[i]->get_h() + 5;
+			auto_reorder[i],
+			auto_reorder[i] & ISAUTO ? 
+auto_text[auto_reorder[i] - ISAUTO] : other_text[auto_reorder[i] - NOTAUTO]));
+		y += toggles[i]->get_h() + 5;
 	}
 }
 
@@ -145,14 +139,9 @@ void GWindowGUI::update_toggles(int use_
 {
 	if(use_lock) lock_window("GWindowGUI::update_toggles");
 
-	for(int i = 0; i < OTHER_TOGGLES; i++)
-	{
-		other[i]->update();
-	}
-
-	for(int i = 0; i < AUTOMATION_TOTAL; i++)
+	for(int i = 0; i < NONAUTOTOGGLES_COUNT + AUTOMATION_TOTAL; i++)
 	{
-		auto_toggle[i]->update();
+		toggles[i]->update();
 	}
 
 	if(use_lock) unlock_window();
@@ -205,40 +194,38 @@ GWindowToggle::GWindowToggle(MWindow *mw
 	GWindowGUI *gui, 
 	int x, 
 	int y, 
-	int subscript, 
-	int other,
+	int toggleref, 
 	char *text)
  : BC_CheckBox(x, 
  	y, 
-	*get_main_value(mwindow, subscript, other), 
+	*get_main_value(mwindow, toggleref), 
 	text)
 {
 	this->mwindow = mwindow;
 	this->gui = gui;
-	this->subscript = subscript;
-	this->other = other;
+	this->toggleref = toggleref;
 }
 
 int GWindowToggle::handle_event()
 {
-	*get_main_value(mwindow, subscript, other) = get_value();
+	*get_main_value(mwindow, toggleref) = get_value();
 	gui->update_mwindow();
 
 
 // Update stuff in MWindow
 	unlock_window();
 	mwindow->gui->lock_window("GWindowToggle::handle_event");
-	if(subscript >= 0)
+	if(toggleref & ISAUTO)
 	{
 		mwindow->gui->canvas->draw_overlays();
 		mwindow->gui->canvas->flash();
 	}
 	e

Re: [CinCVS] [Bug 360] New: consistent list of items in View & Show Overlays

[Pierre Marc Dumuid]

I was having trouble making an attachment to bugzilla.
I made a patch for this issue, but the ordering is kinda outa wack in 
the menu:


Here is a patch that fixes the issue, but because of the way the overlay 
window is created, it can't easily be made to match the order of the 
menubar. 

Modifying the order of the menubar results in the numbers not being in 
order!  Do people mind if we change the keyboard shortcuts so the 
numbering is in order?



Pierre


[EMAIL PROTECTED] wrote:

http://bugs.cinelerra.org/show_bug.cgi?id=360

   Summary: consistent list of items in View & Show Overlays
   Product: Cinelerra
   Version: 2.1
  Platform: PC
OS/Version: Linux
Status: NEW
  Severity: enhancement
  Priority: Medium
 Component: User Interface
AssignedTo: cinelerra@skolelinux.no
ReportedBy: [EMAIL PROTECTED]


Hi!

The Program --> View show the list of different menu items (Show assets, Show
titles) which is not in the same order as the items in Program --> Window -->
Show Overlays, so the feature request is to put the two in the same order to
improve consistency in the user-interface.

Sincerely,
Gour



--- You are receiving this mail because: ---
You are the assignee for the bug, or are watching the assignee.

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra
  


Index: cinelerra/mainmenu.C
===
--- cinelerra/mainmenu.C	(revision 957)
+++ cinelerra/mainmenu.C	(working copy)
@@ -171,11 +171,11 @@
 	viewmenu->add_item(show_assets = new ShowAssets(mwindow, "0"));
 	viewmenu->add_item(show_titles = new ShowTitles(mwindow, "1"));
 	viewmenu->add_item(show_transitions = new ShowTransitions(mwindow, "2"));
+	viewmenu->add_item(plugin_automation = new PluginAutomation(mwindow, "7"));
 	viewmenu->add_item(fade_automation = new ShowAutomation(mwindow, _("Fade"), "3", AUTOMATION_FADE));
 	viewmenu->add_item(mute_automation = new ShowAutomation(mwindow, _("Mute"), "4", AUTOMATION_MUTE));
 	viewmenu->add_item(mode_automation = new ShowAutomation(mwindow, _("Mode"), "5", AUTOMATION_MODE));
 	viewmenu->add_item(pan_automation = new ShowAutomation(mwindow, _("Pan"), "6", AUTOMATION_PAN));
-	viewmenu->add_item(plugin_automation = new PluginAutomation(mwindow, "7"));
 	viewmenu->add_item(mask_automation = new ShowAutomation(mwindow, _("Mask"), "8", AUTOMATION_MASK));
 	viewmenu->add_item(camera_x = new ShowAutomation(mwindow, _("Camera X"), "", AUTOMATION_CAMERA_X));
 	viewmenu->add_item(camera_y = new ShowAutomation(mwindow, _("Camera Y"), "", AUTOMATION_CAMERA_Y));
Index: cinelerra/viewmenu.C
===
--- cinelerra/viewmenu.C	(revision 952)
+++ cinelerra/viewmenu.C	(working copy)
@@ -122,7 +122,7 @@
 
 
 PluginAutomation::PluginAutomation(MWindow *mwindow, char *hotkey)
- : BC_MenuItem(_("Plugin keyframes"), hotkey, hotkey[0]) 
+ : BC_MenuItem(_("Plugin Autos"), hotkey, hotkey[0]) 
 { 
 	this->mwindow = mwindow; 
 }
Index: cinelerra/gwindowgui.C
===
--- cinelerra/gwindowgui.C	(revision 952)
+++ cinelerra/gwindowgui.C	(working copy)
@@ -55,6 +55,22 @@
 	"Nudge"
 };
 
+static int auto_reorder[] = 
+{
+	AUTOMATION_FADE,
+	AUTOMATION_MUTE,
+	AUTOMATION_MODE,
+	AUTOMATION_PAN,
+	AUTOMATION_MASK,
+	AUTOMATION_CAMERA_X,
+	AUTOMATION_CAMERA_Y,
+	AUTOMATION_CAMERA_Z,
+	AUTOMATION_PROJECTOR_X,
+	AUTOMATION_PROJECTOR_Y,
+	AUTOMATION_PROJECTOR_Z
+};
+
+
 void GWindowGUI::calculate_extents(BC_WindowBase *gui, int *w, int *h)
 {
 	int temp1, temp2, temp3, temp4, temp5, temp6, temp7;
@@ -127,9 +143,9 @@
 			this, 
 			x, 
 			y, 
-			i,
+			auto_reorder[i],
 			-1, 
-			auto_text[i]));
+			auto_text[auto_reorder[i]]));
 		y += auto_toggle[i]->get_h() + 5;
 	}
 }

Re: [CinCVS] [CINELERRA-CV MANUAL] Point of the situation

[Pierre Marc Dumuid]

Hi,
Andraz sorry for the extra noise.

could you please try getting the docbook working?

makeinfo --docbook cinelerra.texi

once it works, try:

yelp file://$PWD/cinelerra.xml

and if you move the cinelerra.xml to 
/usr/share/gnome/help/cinelerra/C/cinelerra.xml one can do:


yelp ghelp:cinelerra?THE-PATCHBAY

If this gets implemented I'd like to set up help buttons throughout 
cinelerra's gui.. that execute the appropriate command line..


Pierre

Nicolas wrote:

Hello,

Here are the opinions about "British English vs. American English":
Kurt Georg Hooss: prefers British English
Gour: has no preference
Pierre Dumuid: prefers American English
Stefan de Konink: prefers British English
Mack Allison: prefers American English (doesn't he?)
Julian Oliver: prefers British English
Nicolas Maufrais: prefers British English
- British English: 4 votes
- American English: 2 votes
- No preference: 1 votes
So, we'll use British English for the documentation... unless more
people give their opinion!

Mack Allison wants to participate to the documentation.
So, here is the list of people interested:
- Nicolas Maufrais
- gour
- Mack Allison

On Thu, Nov 02, 2006 at 09:08:51AM +0100, Gour wrote:
  

Do we want to add more deps to Cinelerra build process or we can just
ship produced pdf manual along with the html one from the doc branch?

I am thinking that we won't immediately update manual with the every
revision number, but, let's say, periodically when tagging some bigger
release like e.g. 2.1-merge, quarterly...and therefore it is, imho,
better to just have ready manual in doc/ which clearly states for which
revision it is up to date.

In such situation, there would not be need for adding additional
dependencies.

What do you think?



IMO, we shouldn't ship pdf or html manuals with the source. Those
manuals can be generated from the cinelerra.texi texinfo files, and the
png images. Generating the manuals takes only a few seconds. IMO, the
same principle than source distribution applies. I mean, does
cinelerra-cv ships with binaries? It ships with sources, which can be
compiled. We should use the same method for the documentation.
We will upload the manual to the SVN official branch 1 time per month I
think.

On Thu, Nov 02, 2006 at 08:48:59PM +1030, Pierre Marc Dumuid wrote:
  

I'd prefer to have a configure option
configure --makedocumentation
so dependencies depend on your configure arguments.



That configure option is a good idea. We should use that.

  

I could create one shell script to do all of these tasks. But I don't
think that's the way to go. I assume make should handle the process. Can
someone help me with that issue? That would avoid me to RTFM about
makefiles, and I could concentrate my efforts on working on the manual's
contents.
  

This can be done with autoconf tools. Can you give me url where I can
get manual's tarball?



Here's the tarball:

http://www.europephoto.com/info/cinelerra/200611/CV_Manual_20061102.tgz

Instructions to generate the manual are in the README_MANUAL.txt file.

Right now, I RTFM about the GPL licence, and how to make reference to it
in the manual (precise rules apply).

I ask Herman for an SVN account, in order to be able to upload the
manual, 1 time per month approximately.

Now, how can we work together on the manual?
There're some possibilities:
1 - use SVN or git or another VCS
2 - set up a wiki, where people can edit the cinelerra.texi file
directly (one chapter per page)
3 - set up a wiki, where people can edit the manual without formatting.
I would regularly select the new parts added by people, and put them
into the cinelerra.texi file.

Solution 1 is the most rigorous.
Solution 2 is fine but editing a texinfo file isn't as easy and
"beautiful" than editing the manual in another form.
Solution 3 is easy for people, but I (or someone else) will have to
regularly update the cinelerra.texi file. That solution allows a better
control over what's written in the texinfo file. I'd add typesetting
commands, and correct possible texinfo errors.

You can have a look at the cinelerra.texi file by downloading the .tgz
(see URL below in this email). Do you think that's easy to edit? Do you
prefer to edit a plain text page without formatting?

Gour suggested on the ML to use DocBookWiki:
http://doc-book.sourceforge.net/homepage/

However I'm not satisfied at all with the PDF manual quality it
produces. And we use texinfo as Heroine Virtual Ltd does.

I prefer solution 2. People can easily edit the manual, and me and other
people who want to do it will select and format the contents into the
cinelerra.texi file. However, if some of you think the texinfo file can
easily be edited by *any* people (I mean people not really wishing to
RTFM a bit about texinfo), we can edit the texinfo file directly.

OK, that's a wiki. And there's already 

Re: [CinCVS] [CINELERRA-CV MANUAL] Licensing issues solved

[Pierre Marc Dumuid]

Do we want to add more deps to Cinelerra build process or we can just
ship produced pdf manual along with the html one from the doc branch?

I am thinking that we won't immediately update manual with the every
revision number, but, let's say, periodically when tagging some bigger
release like e.g. 2.1-merge, quarterly...and therefore it is, imho,
better to just have ready manual in doc/ which clearly states for which
revision it is up to date.

In such situation, there would not be need for adding additional
dependencies.

What do you think?
  

I'd prefer to have a configure option
configure --makedocumentation

so dependencies depend on your configure arguments.



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] [PATCH] Cinelerra paste and load with "align cursor to frame" will do paste so..

[Pierre Marc Dumuid]

the only time "autotrimming" should come into play is when dragging
_media_files_ from asset window (or in asset window rightclick+paste) or
when loading _media_files_ directly into timeline. this way you have no
problems with loading XMLs whatsoever.
  

done and now in SVN..
Please test, and make packages after possibl bugs have settled..

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] [PATCH] Cinelerra paste and load with "align cursor to frame" will do paste so..

[Pierre Marc Dumuid]

Andraž Tori wrote:

On Tue, 2006-10-31 at 18:06 +1030, Pierre Dumuid wrote:
  

This is micromanagement.

All you need is simple rule - if align cursor on track is set when
putting assets to tracks, they are always 'rounded' to frame boundaries.
  
  

yes but rounding to which frame boundaries?

I have mpeg material that in units of video frames has 200 for video, 
and 209.3 for audio. do we round each to the closest frame, [200,209], 
the minimum of the two [200,200], the maximmum of the two [209,209].


I personally on loading an mpeg want it clipped to the minimum of the two.



yes, i agree

  
But when pasting a xml project, I'd want it pasted as the maximum to 
preserve the insertion of space!



i don't quite get this. isn't this much more rare scenario in which you
would simply turn off "autotrimming" and do what you want manually
  

not quite..
consider you have a set of tracks:

i1 i2 i3 i4
V xx||
A1xx|x|
A2 xx|


where A1 has been done away with at the end, (bad audio) and a 
replacement has been added (A2)


At the moment with my patch:
1. If I save this project, and then load this project with "align 
cursor../ clip to minimum of audo/video" It'll be reduced to a length of 
i3, being the minimum of (maxlength_audio,maxlength_video)


2. If I select from i1 to i4 and copy, then paste at i4, then it'll 
paste in V only up until i3.


you may then say that when copy / pasting then something idfferent 
should be done.. may be this is an idea (i.e. if source is an xml file 
then past to ceil(maxlength_audio,maxlength_video)...


Does this seem reasonable?

Pierre


bye
andraz


___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra
  



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] [PATCH] Cinelerra paste and load with "align cursor to frame" will do paste so..

[Pierre Marc Dumuid]

It introduces an annoyance (that is resolve by turning off "align  
cursor to frame" ) where a clip having tracks of intentionally 
different frame length (that is already cut properly) will be pasted 
incorrectly..





This problem is more annoying that I initially first thought!  Realise 
that the states of pasting / loading are needed:


 1. Load / Paste to max of all track lengths (non-frame aligned)
 2. Load / Paste to min of all track lengths (frame aligned)  --
needed when loading a video file for first time, we want audio and
video to be the same length.
 3. Load / Paste to max of all track lengths (frame aligned)  --
needed when loading a project that has multiple tracks, and some
tracks are intentionally shorter (quite commin)

Given this, i can't simply use the "align_cursor_on_frame" setting, and 
would need a 3 state toggle somehow.. what do people think? (should 
there be an icon on the toolbar? - would need to re-work the bc_toggle 
to include multi-state buttons to do this!)  Should there be an option 
on loading?



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] FC6 and Ubuntu 6.10

[Pierre Marc Dumuid]

I upgraded to FC6 w/out problems, (I had to recompile entire source 
after yum upgrading -- I use livna repository).


Aaron Newcomb wrote:

I know these distros just came out, but I am just wondering who has
tried running Cinelerra on them yet? I am a Fedora user primarily
(although I run Ubuntu on my laptop) and I am excited about some of
the upgrades in FC6, but I am also tied to Cinelerra. Does anyone
anticipate any problems other than having to compile a few packages by
hand?




___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

[CinCVS] [PATCH] Cinelerra paste and load with "align cursor to frame" will do paste so..

[Pierre Marc Dumuid]

The following is a patch so that when loading and pasting  / dragging 
onto timeline, etc.. the length of the inserted material is the minimum 
of all the  the tracks rounded to the nearest frame.  This fixes issues  
I (and others) have when working with mpeg files that audio and video of 
different lengths.


It introduces an annoyance (that is resolve by turning off "align  
cursor to frame" ) where a clip having tracks of intentionally different 
frame length (that is already cut properly) will be pasted incorrectly..


Can people test before committing?

pierre

diff --git a/cinelerra/asset.C b/cinelerra/asset.C
index 1610374..2fc4772 100644
--- a/cinelerra/asset.C
+++ b/cinelerra/asset.C
@@ -3,6 +3,7 @@ #include "asset.h"
 #include "assets.h"
 #include "bchash.h"
 #include "bcsignals.h"
+#include "clip.h"
 #include "edl.h"
 #include "file.h"
 #include "filesystem.h"
@@ -1200,6 +1201,21 @@ int Asset::update_path(char *new_path)
 	return 0;
 }
 
+double Asset::total_length_framealigned(double fps) 
+{
+	if (video_data && audio_data) {
+		double aud = floor(( (double)audio_length / sample_rate) * fps) / fps;
+		double vid = floor(( (double)video_length / frame_rate) * fps) / fps;
+		return MIN(aud,vid);
+	}
+	
+	if (audio_data)
+		return (double)audio_length / sample_rate;
+	
+	if (video_data)
+		return (double)video_length / frame_rate;
+}
+
 void Asset::update_index(Asset *asset)
 {
 //printf("Asset::update_index 1 %d\n", index_status);
diff --git a/cinelerra/asset.h b/cinelerra/asset.h
index ee289fa..8e3fd49 100644
--- a/cinelerra/asset.h
+++ b/cinelerra/asset.h
@@ -101,6 +101,8 @@ public:
 	int write_index(FileXML *xml);
 	int update_path(char *new_path);
 
+	double total_length_framealigned(double fps);
+
 // Path to file
 	char path[BCTEXTLEN];
 
diff --git a/cinelerra/edits.C b/cinelerra/edits.C
index 35956ac..f9d6fb7 100644
--- a/cinelerra/edits.C
+++ b/cinelerra/edits.C
@@ -701,6 +701,8 @@ void Edits::clear(int64_t start, int64_t
 	}
 
 	optimize();
+// Reset the loaded_length value. (the last edit should always be a silence going to infinityish)
+	loaded_length = last->startproject;
 }
 
 // Used by edit handle and plugin handle movement but plugin handle movement
diff --git a/cinelerra/mwindowedit.C b/cinelerra/mwindowedit.C
index bf5e2e0..c5b341a 100644
--- a/cinelerra/mwindowedit.C
+++ b/cinelerra/mwindowedit.C
@@ -1421,6 +1421,12 @@ SET_TRACE
 			edl->session->frame_rate, 
 			TRACK_VIDEO);
 
+// Align cursor on frames:: clip the new edl to paste to the last joint frame.
+		if (edl->session->cursor_on_frames)
+		{
+			edl_length = new_edl->tracks->total_length_framealigned(edl->session->frame_rate);
+			new_edl->tracks->clear(edl_length, new_edl->tracks->total_length() + 100, 1);
+		}
 SET_TRACE
 
 
diff --git a/cinelerra/track.C b/cinelerra/track.C
index 00e893f..4d8333b 100644
--- a/cinelerra/track.C
+++ b/cinelerra/track.C
@@ -1148,6 +1148,7 @@ int Track::clear(double start, 
 
 	if(edit_edits)
 		edits->clear((int64_t)start, (int64_t)end);
+
 	return 0;
 }
 
diff --git a/cinelerra/trackcanvas.C b/cinelerra/trackcanvas.C
index 04dad9a..56277df 100644
--- a/cinelerra/trackcanvas.C
+++ b/cinelerra/trackcanvas.C
@@ -1097,6 +1097,67 @@ void TrackCanvas::draw_paste_destination
 			mwindow->session->drag_clips->total)
 			clip = mwindow->session->drag_clips->values[0];
 
+// 'Align cursor of frame' lengths calculations
+		double paste_audio_length, paste_video_length;
+		int64_t asset_length;
+		double desta_position = 0;
+		double destv_position = 0;
+
+		if (asset)
+		{
+			double asset_length_ac = asset->total_length_framealigned(mwindow->edl->session->frame_rate);
+			if (mwindow->edl->session->cursor_on_frames)
+			{
+paste_video_length = paste_audio_length = asset_length_ac;
+			} 
+			else 
+			{
+paste_audio_length = (double)asset->audio_length / asset->sample_rate;
+paste_video_length = (double)asset->video_length / asset->frame_rate;
+			}
+
+			// Images have length -1 (they are a single image!!)
+			if (asset->video_data && asset->video_length < 0)
+			{
+if(mwindow->edl->session->si_useduration)
+	paste_video_length = mwindow->edl->session->si_duration / asset->frame_rate;
+else	
+	paste_video_length = 1.0 / asset->frame_rate;  // bit confused!! (this is 1 frame)
+			}
+
+			int64_t asset_length = 0;
+
+			if(asset->audio_data)
+			{
+// we use video if we are over video and audio if we are over audio
+if(asset->video_data && mwindow->session->track_highlighted->data_type == TRACK_VIDEO)
+	asset_length = mwindow->session->track_highlighted->to_units(paste_video_length, 1);
+else
+	asset_length = mwindow->session->track_highlighted->to_units(paste_audio_length, 1);
+
+desta_position = mwindow->session->track_highlighted->from_units(get_drop_position(&insertion, NULL, asset_length));
+			}
+
+			if(asset->video_data)
+			{
+asset_length = mwindow->session->track_highlighted->to_units((double)paste_video_length, 1);
+destv_

Re: [CinCVS] bclistbox.h:365: error: extra qualification 'BC_ListBox::' on member 'calculate_last_coords_recursive'

[Pierre Marc Dumuid]

It seems you checked out an incorrect branch in the repository since 
this error was fixed ages ago.  The reason the version in 950 is because 
svn increments the version across all branches (which I think is stupid!).

make sure you're checking out from:

svn://svn.skolelinux.org/cinelerra/trunk/hvirtual

Execute:
$ svn info 


to see where the repository is from... It should say:
URL: svn://svn.skolelinux.org/cinelerra/trunk/hvirtual
Repository Root: svn://svn.skolelinux.org/cinelerra/trunk/hvirtual

Pierre

mack allison wrote:
I'm building cinelerra (r950) on ubuntu 6.10-amd64, and this is what 
I've come to.  A search of lists and google for info related to 
guicast and bclistbox.h only yield svn links to the code itself (a 
little over my head) .


This one has got me stumped.  I've met all the dependancies looked for 
in configure. All of the dependancies were met from the 6.10 package 
tree except libquicktime, which is  libquicktime-0.9.10, compiled from 
thier latest sourceforge release tarball, and x264, which is r596 
--enable-pic.  I've run into this wall:


Making all in guicast
make[2]: Entering directory `/usr/src/hvirtual/guicast'
if /bin/bash ../libtool --tag=CXX --mode=compile g++ -DHAVE_CONFIG_H 
-I. -I. -I.. -I../quicktime   -D_LARGEFILE_SOURCE 
-D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64  -g -O2 -MT bcbitmap.lo 
-MD -MP -MF ".deps/bcbitmap.Tpo" \
  -c -o bcbitmap.lo `test -f 'bcbitmap.C' || echo 
'./'`bcbitmap.C; \

then mv -f ".deps/bcbitmap.Tpo" ".deps/bcbitmap.Plo"; \
else rm -f ".deps/bcbitmap.Tpo"; exit 1; \
fi
mkdir .libs
 g++ -DHAVE_CONFIG_H -I. -I. -I.. -I../quicktime -D_LARGEFILE_SOURCE 
-D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -g -O2 -MT bcbitmap.lo 
-MD -MP -MF .deps/bcbitmap.Tpo -c bcbitmap.C  -fPIC -DPIC -o 
.libs/bcbitmap.o
bclistbox.h:365: error: extra qualification 'BC_ListBox::' on member 
'calculate_last_coords_recursive'

make[2]: *** [bcbitmap.lo ] Error 1
make[2]: Leaving directory `/usr/src/hvirtual/guicast'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/usr/src/hvirtual'
make: *** [all] Error 2





___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] [Patches] Possibility to not build again ffmpeg in quicktime4linux

[Pierre Marc Dumuid]

committed (with adjustments)

Note, people developing versioninfo.h outside of the ./configure script 
will need to follow info in the commit message!


Alexis,
I just realised after commiting that I forgot to credit you with the 
creation of this patch!! I'm so sorry!


Pierre

Alexis Ballier wrote:

Hi,

regarding the problems on building cinelerra on ppc arches, I
remembered the crazy idea I had some time ago : Leave the possibility
to not build ffmpeg again and link to it instead.

So, I wrote two patches that are attached, the externalffmpeg.patch
adds the configure check to pass --with-external-ffmpeg, if nothing is
passed to the ./configure, it is just as it was before.

The typo.patch is a small fix, a typo was introduced for a #define in
some old ffmpeg versions and then they corrected it, this patch fixes
this typo in cinelerra. This is needed because, otherwise, when
including ffmpeg headers from the system, compilation will fail because
it cannot find the defined constant.


I'd really like to have the possibility to not build again ffmpeg, in
the future this might make possible to build cinelerra on other arches
than x86, amd64 & ppc.

So far this works well for me on amd64.


Regards,


Alexis.
  



--- cinelerra-svn-checkout/configure.in 2006-10-27 07:49:05.482831480 +0200
+++ cinelerra-external-ffmpeg/configure.in  2006-10-27 08:48:00.339987883 
+0200
@@ -14,6 +14,8 @@
 AC_LANG_C
 AC_PROG_CXX
 
+AM_PROG_AS

+
 AC_ENABLE_SHARED
 AC_DISABLE_STATIC
 AC_PROG_LIBTOOL
@@ -394,6 +378,26 @@
 AC_SUBST(OPENGL_LIBS)
 # END OpenGL
 
+ External ffmpeg

+
+
+AC_ARG_WITH([external-ffmpeg], AC_HELP_STRING([--with-external-ffmpeg], [use 
external ffmpeg library]))
+
+if test "x$with_external_ffmpeg" = "xyes"; then
+ PKG_CHECK_MODULES([FFMPEG], [libavcodec libpostproc])
+ FFMPEG_FOLDER=""
+else
+ FFMPEG_FOLDER=ffmpeg
+ FFMPEG_CFLAGS="-I$(srcdir)/ffmpeg/libavcodec"
+ FFMPEG_LIBS="ffmpeg/libavcodec/libavcodec.la"
+fi
+
+AC_SUBST(FFMPEG_FOLDER)
+AC_SUBST(FFMPEG_CFLAGS)
+AC_SUBST(FFMPEG_LIBS)
+
+# END External ffmpeg
+
 ## .png TO .o CONVERSION
 
 AC_CHECK_TOOL(OBJCOPY, objcopy)

--- cinelerra-svn-checkout/quicktime/Makefile.am2006-09-10 
11:41:23.089983933 +0200
+++ cinelerra-external-ffmpeg/quicktime/Makefile.am 2006-10-27 
08:43:46.535909148 +0200
@@ -1,4 +1,4 @@
-SUBDIRS = encore50 ffmpeg
+SUBDIRS = encore50 @FFMPEG_FOLDER@
 
 
 lib_LTLIBRARIES = libquicktimehv.la

@@ -10,12 +10,12 @@
$(MJPEG_CFLAGS) \
$(LIBDV_CFLAGS) \
-DENCORE_INCLUDE=\"encore50/encore.h\" \
-   -I$(srcdir)/ffmpeg/libavcodec
+   @FFMPEG_CFLAGS@
 
 libquicktimehv_la_LDFLAGS = -version-info 1:0:0 -release 1.6.0 -Wl,--no-undefined 
 
 libquicktimehv_la_LIBADD = \

-   ffmpeg/libavcodec/libavcodec.la \
+   @FFMPEG_LIBS@ \
$(XIPH_LIBS) \
-lmp3lame \
$(FAAD_LIBS) \
  



--- cinelerra-svn-checkout/quicktime/mpeg4.c2006-10-20 21:27:40.181569638 
+0200
+++ cinelerra-external-ffmpeg/quicktime/mpeg4.c 2006-10-27 08:51:33.593001333 
+0200
@@ -671,7 +671,7 @@
 
 	context->b_quant_factor = 1.25;

context->b_quant_offset = 1.25;
-   context->error_resilience = FF_ER_CAREFULL;
+   context->error_resilience = FF_ER_CAREFUL;
context->error_concealment = 3;
context->frame_skip_cmp = FF_CMP_DCTMAX;
context->ildct_cmp = FF_CMP_VSAD;
--- cinelerra-svn-checkout/quicktime/ffmpeg/libavcodec/avcodec.h
2006-09-10 11:41:18.918902957 +0200
+++ cinelerra-external-ffmpeg/quicktime/ffmpeg/libavcodec/avcodec.h 
2006-10-27 08:50:54.561601215 +0200
@@ -1000,7 +1000,7 @@
  * - decoding: set by user
  */
 int error_resilience;
-#define FF_ER_CAREFULL1
+#define FF_ER_CAREFUL 1
 #define FF_ER_COMPLIANT   2
 #define FF_ER_AGGRESSIVE  3
 #define FF_ER_VERY_AGGRESSIVE 4
--- cinelerra-svn-checkout/quicktime/ffmpeg/ffmpeg.c2006-09-10 
11:41:20.606531118 +0200
+++ cinelerra-external-ffmpeg/quicktime/ffmpeg/ffmpeg.c 2006-10-27 
08:51:05.539182498 +0200
@@ -179,7 +179,7 @@
 static int do_interlace_dct = 0;
 static int do_interlace_me = 0;
 static int workaround_bugs = FF_BUG_AUTODETECT;
-static int error_resilience = FF_ER_CAREFULL;
+static int error_resilience = FF_ER_CAREFUL;
 static int error_concealment = 3;
 static int dct_algo = 0;
 static int idct_algo = 0;
--- cinelerra-svn-checkout/quicktime/ffmpeg/ffplay.c2006-09-10 
11:41:20.646522306 +0200
+++ cinelerra-external-ffmpeg/quicktime/ffmpeg/ffplay.c 2006-10-27 
08:51:13.497429039 +0200
@@ -197,7 +197,7 @@
 static enum AVDiscard skip_frame= AVDISCARD_DEFAULT;
 static enum AVDiscard skip_idct= AVDISCARD_DEFAULT;
 static enum AVDiscard skip_loop_filter= AVDISCARD_DEFAULT;
-

Re: [CinCVS] A short extract before/after motion effect

[Pierre Marc Dumuid]

Nicolas,
di you mind leaving those clips there for a while? I was thinking of 
linking to them from cvs.cinelerra.org as demonstration clips... Is that 
o.k?


Nicolas wrote:

On Wed, Oct 18, 2006 at 06:27:07PM +, [EMAIL PROTECTED] wrote:
  

Nicolas,
Great analysis.  This will help me.  I will try to get to this tonight/tomorrow.

Do you have before and after vids?
scott



Scott,

Here is a short extract...

Before: deinterlaced "Raw" DV footage
http://www.europephoto.com/info/cinelerra/200610/20061018_motion_effect_BEFORE.m2v

After: motion effect and zoom-in applied:
http://www.europephoto.com/info/cinelerra/200610/20061018_motion_effect_AFTER.m2v

Nicolas.

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra
  



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] build problem undefined reference to `mm_flags'

[Pierre Marc Dumuid]

I fixed this for gour and Jean-Luc (see discussion in IRC log).. Could 
you confirm that it is fixed for yourself? (I assume you don't configure 
with --enable-mmx)


Pierre

Graham Evans wrote:

On Fri, 2006-10-27 at 17:39 -0200, rafael2k wrote:
  
after the ./configure, place the attached file inside 
quicktime/ffmpeg

and see if cinelerra compiles.




failed again:

...
libtool: link: warning: library `/usr/lib64/libdv.la' was moved.
gcc
-shared  .libs/atom.o .libs/avcc.o .libs/avi_hdrl.o .libs/avi_idx1.o .libs/a 
vi_movi.o .libs/avi_strl.o .libs/avi_odml.o .libs/avi_ix.o .libs/avi_indx.o 
.lib s/avi_riff.o .libs/cmodel_default.o .libs/cmodel_float.o 
.libs/cmodel_yuv420p.o .libs/cmodel_yuv422.o .libs/codecs.o .libs/colormodels.o 
.libs/ctab.o .libs/dinf .o .libs/dref.o .libs/edts.o .libs/elst.o .libs/esds.o 
.libs/graphics.o .libs/hd lr.o .libs/ima4.o .libs/interlacemodes.o .libs/jpeg.o 
.libs/libdv.o .libs/libmjp eg.o .libs/matrix.o .libs/mdat.o .libs/mdhd.o 
.libs/mdia.o .libs/minf.o .libs/mo ov.o .libs/mp4a.o .libs/mvhd.o 
.libs/plugin.o .libs/qtcache.o .libs/qtdv.o .libs /qtffmpeg.o .libs/qth264.o 
.libs/qtpng.o .libs/qtmp3.o .libs/quicktime.o .libs/r aw.o .libs/rawaudio.o 
.libs/rle.o .libs/smhd.o .libs/stbl.o .libs/stco.o .libs/s tsc.o .libs/stsd.o 
.libs/stsdtable.o .libs/stss.o .libs/stsz.o .libs/stts.o .lib s/tkhd.o 
.libs/trak.o .libs/twos.o .libs/udta.o .libs/ulaw.o .libs/util.o .libs/ v308.o 
.libs/v408.o .libs/v410.o .libs/vmhd.o .libs/vbraudio.o .libs/vorbis.o .l 
ibs/workarounds.o .libs/yuv2.o .libs/yuv4.o .libs/yv12.o .libs/wmx2.o 
.libs/wma. o .libs/mpeg4.o -Wl,--whole-archive 
ffmpeg/libavcodec/.libs/libavcodec.a encore5 0/.libs/libencore.a 
-Wl,--no-whole-archive  -Wl,--rpath -Wl,/home/gray/hvirtual/ libmpeg3/.libs 
-la52 -lvorbisenc -lvorbisfile -lvorbis -ltheora -logg -lmp3lame -lfaad -lfaac 
../libmpeg3/.libs/libmpeg3hv.so -lx264 -L/lib /usr/lib64/libdv.so 
/usr/lib64/libjpeg.so -lpng -lz -lm -ldl -lpthread  -Wl,--no-undefined 
-Wl,-sona me -Wl,libquicktimehv-1.6.0.so.1 -o 
.libs/libquicktimehv-1.6.0.so.1.0.0
ffmpeg/libavcodec/.libs/libavcodec.a(utils.o)(.text+0x18f6): In function
`avcode c_decode_video':
/home/gray/hvirtual/quicktime/ffmpeg/libavcodec/utils.c:635: undefined
reference  to `mm_flags'
ffmpeg/libavcodec/.libs/libavcodec.a(utils.o)(.text+0x1e55): In function
`avcode c_encode_video':
/home/gray/hvirtual/quicktime/ffmpeg/libavcodec/utils.c:597: undefined
reference  to `mm_flags'
ffmpeg/libavcodec/.libs/libavcodec.a(mpegvideo.o)(.text+0x3c8): In
function `DCT _common_init':
/home/gray/hvirtual/quicktime/ffmpeg/libavcodec/mpegvideo.c:247:
undefined refer ence to `MPV_common_init_mmx'
ffmpeg/libavcodec/.libs/libavcodec.a(mpegvideo.o)(.text+0x28b0): In
function `ff _draw_horiz_band':
/home/gray/hvirtual/quicktime/ffmpeg/libavcodec/mpegvideo.c:4006:
undefined refe rence to `mm_flags'
ffmpeg/libavcodec/.libs/libavcodec.a(mpegvideo.o)(.text+0x6748): In
function `MP V_frame_end':
/home/gray/hvirtual/quicktime/ffmpeg/libavcodec/mpegvideo.c:1593:
undefined refe rence to `mm_flags'
ffmpeg/libavcodec/.libs/libavcodec.a(mpegvideo.o)(.text+0x72fb): In
function `MP V_encode_picture':
/home/gray/hvirtual/quicktime/ffmpeg/libavcodec/mpegvideo.c:5299:
undefined refe rence to `mm_flags'
ffmpeg/libavcodec/.libs/libavcodec.a(mpegvideo.o)(.text
+0x79a6):/home/gray/hvirt
ual/quicktime/ffmpeg/libavcodec/mpegvideo.c:5457: undefined reference to
`mm_fla gs'
ffmpeg/libavcodec/.libs/libavcodec.a(mpegvideo.o)(.text
+0x8322):/home/gray/hvirt
ual/quicktime/ffmpeg/libavcodec/mpegvideo.c:2219: undefined reference to
`mm_fla gs'
ffmpeg/libavcodec/.libs/libavcodec.a(mpegvideo.o)(.text
+0x108d5):/home/gray/hvir
tual/quicktime/ffmpeg/libavcodec/mpegvideo.c:5781: more undefined
references to `mm_flags' follow
ffmpeg/libavcodec/.libs/libavcodec.a(dsputil.o)(.text+0xd1de): In
function `dspu til_init':
/home/gray/hvirtual/quicktime/ffmpeg/libavcodec/dsputil.c:3911:
undefined refere nce to `dsputil_init_mmx'
ffmpeg/libavcodec/.libs/libavcodec.a(imgresample.o)(.text+0x553): In
function `i mg_resample':
/home/gray/hvirtual/quicktime/ffmpeg/libavcodec/imgresample.c:464:
undefined ref erence to `mm_flags'
ffmpeg/libavcodec/.libs/libavcodec.a(mpeg12.o)(.text+0x2a62): In
function `decod e_frame':
/home/gray/hvirtual/quicktime/ffmpeg/libavcodec/mdec.c:210: undefined
reference to `mm_flags'
ffmpeg/libavcodec/.libs/libavcodec.a(mpeg12.o)(.text+0x79ee): In
function `mpeg_ decode_frame':
/home/gray/hvirtual/quicktime/ffmpeg/libavcodec/mpeg12.c:3186: undefined
referen ce to `mm_flags'
ffmpeg/libavcodec/.libs/libavcodec.a(mpeg12.o)(.text+0x7ba1): In
function `slice _decode_thread':
/home/gray/hvirtual/quicktime/ffmpeg/libavcodec/mpeg12.c:2706: undefined
referen ce to `mm_flags'
ffmpeg/libavcodec/.libs/libavcodec.a(ratecontrol.o)(.text+0xac3): In
function `f f_rate_control_uninit':
/home/gray/hvirtual/quicktime/ffmpeg/libavcodec/ratecontrol.c:181:
undefined ref erence to `mm_flags'
ffmpeg/libavcodec/.libs/libavcodec.a(r

Re: [CinCVS] build problem undefined reference to `mm_flags'

[Pierre Marc Dumuid]

Jean-Luc Coulon (f5ibh) wrote:

Le 27.10.2006 17:45:33, Graham Evans a écrit :

I just ran into problems trying to upgrade from svn 940 to svn 947 on
fc4_64.  I ran a make clean and autogen then ./configure
--enable-opengl

Make aborted as follows:
...



Got the same.
Was not present in 946

Are you on 64 bit machine as well?


Regards

Jean-Luc



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] build problem undefined reference to `mm_flags'

[Pierre Marc Dumuid]

Could you please post the avconfig.h file that is auto-generated?
pierre

Jean-Luc Coulon (f5ibh) wrote:

Le 27.10.2006 17:45:33, Graham Evans a écrit :

I just ran into problems trying to upgrade from svn 940 to svn 947 on
fc4_64.  I ran a make clean and autogen then ./configure
--enable-opengl

Make aborted as follows:
...



Got the same.
Was not present in 946

Regards

Jean-Luc



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] small fix for entering timecodes

[Pierre Marc Dumuid]

Andraž Tori wrote:

Yes, this is nice. One thing though, horizontal dialogs seem a bit less
disturbing to me.. if everything is vertical i get intimidated by the
sheer number of things to set.
So.. even with widgetgrid, things could go into width instead hight a
little bit more.
  
I agree, (It was just a proof of concept). When I was playing around, 
with this some time ago, I actually made a dialog box flip from 
horizonal to vertical whilst the user dragged the borders of the dialog 
box. This seemed cool and handy for small desktops (not sure if GTK / QT 
would do that! :) )


Pierre

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] proposed patched for powerpc compilation fix

[Pierre Marc Dumuid]

Applying this now.

(Note, I personally prefer 1 big patch than 4 smaller patches .. svn 
diff > /tmp/foo.diff works well for me!)


Pierre

rafael2k wrote:

hi all,
with the help of Pierre Dumuid, i wrote the following patches to make 
cinelerra compatible, by now, with i586 and powerpc. (other archs may come 
easily, as sparc, alpha, powerpc64 and mips).


there are four patches:

- avcmakefile-fix.diff: fix the Makefile.am of quicktime/ffmpeg/libavcodec/ to 
include conditional for powerpc processor. TODO - put altivec conditional


- configure-fix.diff: fix configure.in - added a call to 
quicktime/ffmpeg/configure to generate architecture coerent 
quicktime/ffmpeg/avconfig.h (exatly the same of the old static avconfig.h w/ 
arch things!). TODO - just a warning: do not use --enable-altivec yet!


- ffconfigure-fix.diff: fix quicktime/ffmpeg/configure - small fixes and now 
it generates avconfig.h directly


- ppcmakefileam-fix.diff: added a Makefile.am file in 
quicktime/ffmpeg/libavcodec/ppc. TODO - put altivec files to compile when 
altivec is enabled


here it generated correct code for both i386 and ppc.

altivec is to come.

bye,
rafael diniz


  



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] problems when compiling

[Pierre Marc Dumuid]

Well, the fact there's a 2 there means it's not the same, and likely a newer 
function.
You may want to check to see if you can find a later  libavc1394-0 and 
libavc1394-dev packages...


[EMAIL PROTECTED] wrote:

i checked for the file
/usr/include/libavc1394/avc1394_vcr.h
and he is there... ok

although there is no such function (avc1394_vcr_get_timecode2) in this file
i could find one named:
avc1394_vcr_get_timecode(raw1394handle_t handle, nodeid_t node);

these functions are supposed to do the same aren't they?

Pierre Marc Dumuid disse:
  

do you get the following?
/usr/include/libavc1394/avc1394_vcr.h

Also you could try:
grep avc1394_vcr_get_timecode2 /usr/include/* -r

I personally get:
[EMAIL PROTECTED] ~]$ grep avc1394_vcr_get_timecode2 /usr/include/* -r
/usr/include/libavc1394/avc1394_vcr.h:avc1394_vcr_get_timecode2(raw1394handle_t
handle, nodeid_t node, char *output);


[EMAIL PROTECTED] wrote:


hi,
i've recently downloaded cinelerra source code from svn.
the configure script ran just fine... but when compiling i get the
following error:

avc1394control.C: In member function `char* AVC1394Control::timecode()':
avc1394control.C:166: error: `avc1394_vcr_get_timecode2' undeclared
(first
use
   this function)
avc1394control.C:166: error: (Each undeclared identifier is reported
only
once
   for each function it appears in.)

i have libavc1394-0 and libavc1394-dev packages installed... so...

thanks for any tips

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

  

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra







___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra
  



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] problems when compiling

[Pierre Marc Dumuid]

do you get the following?
/usr/include/libavc1394/avc1394_vcr.h

Also you could try:
grep avc1394_vcr_get_timecode2 /usr/include/* -r

I personally get:
[EMAIL PROTECTED] ~]$ grep avc1394_vcr_get_timecode2 /usr/include/* -r
/usr/include/libavc1394/avc1394_vcr.h:avc1394_vcr_get_timecode2(raw1394handle_t 
handle, nodeid_t node, char *output);



[EMAIL PROTECTED] wrote:

hi,
i've recently downloaded cinelerra source code from svn.
the configure script ran just fine... but when compiling i get the
following error:

avc1394control.C: In member function `char* AVC1394Control::timecode()':
avc1394control.C:166: error: `avc1394_vcr_get_timecode2' undeclared (first
use
   this function)
avc1394control.C:166: error: (Each undeclared identifier is reported only
once
   for each function it appears in.)

i have libavc1394-0 and libavc1394-dev packages installed... so...

thanks for any tips

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra
  



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] powerpc and other archs problem for ffmpeg inside quicktime dir

[Pierre Marc Dumuid]

Hi Rafael,

I asked you on the IRC to try use the configure script.  I forgot to 
mention that the script needs a bit of hacking.. (see the attached patch)


The idea I had has to add something like the following to the configure 
script:

# BEGIN QUICKTIME COMPILE SETTING FIX
mkdir -p quicktime/ffmpeg/
cd quicktime/ffmpeg
$(source_dir)/quicktime/ffmpeg/configure
mv config.h avconfig.h
cd ../../
# END QUICKTIME COMPILE SETTING FIX

The first question I had was if the config.h was similar to the 
avconfig.h and if cinelerra would compile with this..


or something like that..

Cheers,
Pierre

rafael2k wrote:
I realized that in the compilation of cinelerra the file 
quicktime/ffmpeg/avconfig.h is not regenerated in compilation time.

And it contains defines like:
#define ARCH_X86 1
#define __CPU__ 586


i don't exacly know how to fix it properly, but using the ./configure inside 
the quicktime/ffmpeg directory, it generates a config.h file that contains 
the correct arch information, and the rest is exatly the same of the 
avconfig.h.


thats not the only problem, another problem now is that linking 
libquicktimehv-1.6.0.so.1.0.0 gives me undefined reference for 
dsputil_init_ppc

i'll look in the makefile.

bye,
rafael diniz


  


diff --git a/quicktime/ffmpeg/configure b/quicktime/ffmpeg/configure
index cf77f23..d5b23a4 100755
--- a/quicktime/ffmpeg/configure
+++ b/quicktime/ffmpeg/configure
@@ -3,6 +3,9 @@ #
 # ffmpeg configure script (c) 2000, 2001, 2002 Fabrice Bellard
 #
 
+# This configure script has been hacked for cinelerra's own purposes!
+# search for the 'cinelerra' strings to observe these hacks.
+
 if test x"$1" = x"-h" -o x"$1" = x"--help" ; then
 cat << EOF
 
@@ -1621,12 +1624,13 @@ if test "$source_path_used" = "yes" ; th
 FILES="Makefile libavformat/Makefile libavcodec/Makefile \
   libavcodec/libpostproc/Makefile tests/Makefile vhook/Makefile \
   doc/Makefile doc/texi2pod.pl"
-for dir in $DIRS ; do
-mkdir -p $dir
-done
-for f in $FILES ; do
-ln -sf "$source_path/$f" $f
-done
+# CINELERRA - commented to avoid creating directories...
+#for dir in $DIRS ; do
+#mkdir -p $dir
+#done
+#for f in $FILES ; do
+#ln -sf "$source_path/$f" $f
+#done
 echo "SRC_PATH=$source_path" >> config.mak
 else
 echo "SRC_PATH='$source_path'" >> config.mak
@@ -1690,13 +1694,18 @@ rm -f $TMPO $TMPC $TMPE $TMPS $TMPH
 # build pkg-config files libavcodec.pc, libavformat.pc and libpostproc.pc 
 
 lavc_build=`grep '#define LIBAVCODEC_BUILD' "$source_path/libavcodec/avcodec.h" | sed 's/[^0-9]//g'`
-lavf_build=`grep '#define LIBAVFORMAT_BUILD' "$source_path/libavformat/avformat.h" | sed 's/[^0-9]//g'`
+# CINELERRA: libavformat directory not actually in the Cinelerra repo
+#lavf_build=`grep '#define LIBAVFORMAT_BUILD' "$source_path/libavformat/avformat.h" | sed 's/[^0-9]//g'`
 
 requires=
 test "$libogg" = "yes" && requires="$requires ogg >= 1.1"
 test "$vorbis" = "yes" && requires="$requires vorbis"
 test "$theora" = "yes" && requires="$requires theora"
 
+# CINELERRA: 
+# BEGIN COMMENTING OUT THE .pc FILES
+if test 1 -eq 0 ; then
+
 # libavcodec.pc
 cat 

Re: [CinCVS] merge editing modes

[Pierre Marc Dumuid]

Oops, I didn't read johannes email above!  didn't need to propose git! 
(was browsing email via web interface rathe than thunderbird...)

Pierre
Pierre Dumuid wrote:


Thanks Andraz,

BTW:

Johannes mentioned this idea on IRC and I second it,

To save on commits to the SVN these ideas can be placed in a git repo set up on
pipapo for git users.  This can provide a place for others to try your idea
before commit.  The branching system also I find really cool to in being able to
keep commits which didn't in the end make it, and everyone can still see / find
it easiliy.

Pierre


Quoting Andraž Tori <[EMAIL PROTECTED]>:

 


done

bye
andraz

On Fri, 2006-10-20 at 23:45 +0200, Johannes Sixt wrote:
   


On Friday 20 October 2006 20:49, Andraž Tori wrote:
 


Yes, since everyone is so overhelmingly against this change, i'll find
out some other solution ...

Though i still think editing modes are something that confuses users...
it's the same as emacs ... powerful but only handfull of people will
master them.

i'll try to code up something tomorrow
   


Please, pretty please, make the reversal of r942 a *separate* commit, not
intermingled with whatever your new solution will be.

Thank you,
-- Hannes

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra
 


___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

   




___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra
 




___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] merge editing modes

[Pierre Marc Dumuid]

Ok.. I really don't like this modification. I hate having to hold the 
shift down.

I work with selection more than anything else...

Andraz,
Isn't your patch making the default mode "drag and drop" and modify the 
functionality of dragging on the time-bar, or pressing the shift key to 
turn selectionabilty mode on. (i.e. creating extra effort for every 
attempt to use selection style editing)


Why not bring the modes back, but make the default "drag and drop" if 
you like, and add keep the modifiers there. That way when you teach 
newbies, you can place it in the "drag and drop" mode and leave it there 
for them. You could even lock it to not be switchable in the preferences..


Failing that, I'll have to use a personal branch of cinelerra at all 
times for my own use..


Pierre

Andraž Tori wrote:

On Fri, 2006-10-20 at 09:59 +0200, Kurt Georg Hooss wrote:
  

folks, once i understood i found it makes sense to have the two modes.
but having the issue on the agenda anyway...
there is one closely related thing that really annoys me.

in either mode, i found it problematic that cutting out some part
(be it a complete clip or a drag-&-drop selection)
automatically moves the rest of the track from the right, leaving no gap.

i mean, would it not be more intuitive if cutting would just cut
and moving would just move?

sure, combining those two functionalities may be desirable in cases...
but how can i just remove some part from some (but not all) tracks,
leaving the rest synchronous with the remaining tracks?



what you really are trying to do is "copy'n'mute"

just press 'c' and then 'm' afterwards. 


as of when dragging clips by mouse, ... if you see >< as insertion point
the clip will be cutout and inserted at the insertion point, if you
don't see it, clip will be "copy&mute" and inserted at insertion point

bye
andraz



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra
  



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] merge editing modes

[Pierre Marc Dumuid]

Hi Andraz,

I find this patch REALLY difficult to use. I really don't know how to 
work with this patch anymore..
I personally have previewed a project via clicking-and-dragging. in a 
scrubbing manner. I know one can do this on the time-line, but these 
selection doesn't grow with the view. Also holding down the shift key 
sucks!


I really don't like the default of clicking on the time-line to be 
dragging clips around! (especially
I am also used to looking at a feature, ( be it in the audio, or the 
video) and clicking just before it / after it to set the cursor at that 
location to play back from, or be the "out" of my new selection.


With your patch, the default is "do nothing" and if you move a pixel, 
sucked in, you've move your edits around (and possibly loosing sync)! I 
like the fact that to move edit's around took extra work of going into a 
different mode.


Just my thought.

Pierre

Andraž Tori wrote:

a patch that completely merges both editing modes of cinelerra into a single 
one, with shift key being the modifier ...

editing modes are one of the hardest things for new learners of cinelerra to 
comprehend (by my experience) and there is really no reason not to merge them, 
since most of people expect to use shift for selecting.

bye
andraz  
  



diff -ru --exclude-from exclude hvirtual-svn/cinelerra/cwindowgui.C 
hvirtual-2.1/cinelerra/cwindowgui.C
--- hvirtual-svn/cinelerra/cwindowgui.C 2006-10-12 16:59:13.0 +0200
+++ hvirtual-2.1/cinelerra/cwindowgui.C 2006-10-19 14:22:43.0 +0200
@@ -612,8 +612,6 @@
 		cwindow->gui, 
 		mwindow->theme->cedit_x, 
 		mwindow->theme->cedit_y,
-		mwindow->edl->session->editing_mode, 
-		0,

1,
 		0, 
 		0,

diff -ru --exclude-from exclude hvirtual-svn/cinelerra/editpanel.C 
hvirtual-2.1/cinelerra/editpanel.C
--- hvirtual-svn/cinelerra/editpanel.C  2006-10-12 12:09:09.0 +0200
+++ hvirtual-2.1/cinelerra/editpanel.C  2006-10-19 14:22:16.0 +0200
@@ -31,8 +31,6 @@
BC_WindowBase *subwindow,
 	int x, 
 	int y, 
-	int editing_mode, 
-	int use_editing_mode,
 	int use_keyframe, 
 	int use_splice,   // Extra buttons

int use_overwrite,
@@ -48,8 +46,6 @@
int is_mwindow,
int use_cut)
 {
-   this->editing_mode = editing_mode;
-   this->use_editing_mode = use_editing_mode;
this->use_keyframe = use_keyframe;
this->use_splice = use_splice;
this->use_overwrite = use_overwrite;
@@ -70,8 +66,6 @@
this->x = x;
this->y = y;
this->meter_panel = 0;
-   arrow = 0;
-   ibeam = 0;
keyframe = 0;
fit = 0;
fit_autos = 0;
@@ -89,21 +83,12 @@
 
 void EditPanel::update()

 {
-   int new_editing_mode = mwindow->edl->session->editing_mode;
-   if(arrow) arrow->update(new_editing_mode == EDITING_ARROW);
-   if(ibeam) ibeam->update(new_editing_mode == EDITING_IBEAM);
if(keyframe) keyframe->update(mwindow->edl->session->auto_keyframes);
subwindow->flush();
 }
 
 void EditPanel::delete_buttons()

 {
-   if(use_editing_mode)
-   {
-   if(arrow) delete arrow;
-   if(ibeam) delete ibeam;
-   }
-   
if(use_keyframe)
delete keyframe;
 
@@ -144,19 +129,11 @@
 
 
 SET_TRACE

-   if(use_editing_mode)
-   {
-   subwindow->add_subwindow(arrow = new ArrowButton(mwindow, this, 
x1, y1));
-   x1 += arrow->get_w();
-   subwindow->add_subwindow(ibeam = new IBeamButton(mwindow, this, 
x1, y1));
-   x1 += ibeam->get_w();
-   }
-
if(use_keyframe)
{
subwindow->add_subwindow(keyframe = new KeyFrameButton(mwindow, 
x1, y1));
x1 += keyframe->get_w();
-   x1 += mwindow->theme->toggle_margin;
+// x1 += mwindow->theme->toggle_margin;
}
 
 // Mandatory

@@ -320,19 +297,11 @@
this->y = y;
x1 = x, y1 = y;
 
-	if(use_editing_mode)

-   {
-   arrow->reposition_window(x1, y1);
-   x1 += arrow->get_w();
-   ibeam->reposition_window(x1, y1);
-   x1 += ibeam->get_w();
-   }
-
if(use_keyframe)
{
keyframe->reposition_window(x1, y1);
x1 += keyframe->get_w();
-   x1 += mwindow->theme->toggle_margin;
+// x1 += mwindow->theme->toggle_margin;
}
 
 	inpoint->reposition_window(x1, y1);

@@ -1004,55 +973,6 @@
 
 
 
-ArrowButton::ArrowButton(MWindow *mwindow, EditPanel *panel, int x, int y)
- : BC_Toggle(x, 
- 	y, 
-	mwindow->theme->get_image_set("arrow"),

-   mwindow->edl->session->editing_mode == EDITING_ARROW,
-   "",
-   0,
-   0,
-   0)
-{
-   this->mwindow = mwindow;
-   this->panel = panel;
-   set_tooltip(_("Drag and drop editing mode"));
-}
-
-int ArrowButton::handle_event()
-{
-

Re: [CinCVS] installing from SVN on Fedora Core 5

[Pierre Marc Dumuid]

Hi Claude,

Here's a spec file that I made up a couple of weeks ago:

It's still got a little bit to go..

The Requires  lines within the spec file is a list of the current 
libraries I had when I compiled.. probably get away with less.


All the libraries are available from livna, which is available from 
dries (but this is on the livna bugzilla

http://bugzilla.livna.org/show_bug.cgi?id=1179)

The idea of the first few %defines is to assign some optimisation 
flags.  the genopt_flags are general and are typically good for all 
architectures. and the rpmcflags contains specific architecture info.


You'll need to make an archive of the svn checkout (with the .svn stuff 
included.. this could change later by modifying the %build section to 
make a custom versioninfo.h file) and name it as cinelerra-${ver0} and 
place it in /usr/src/redhat/SOURCES/  (my file name was 
cinelerra-2.1.CV.svn.r909M.tar.gz)


Don't know if this helps as a starter for anything..

Pierre


Claude Jones wrote:

On Wed October 18 2006 11:10 am, [EMAIL PROTECTED] wrote:
  

Hi Claude,
I've had a good deal of experience with FC4.  I assume the
deps for FC5 are more or less the same.  Here is a list of the
dependent packages for that system:
http://crazedmuleproductions.blogspot.com/2006/08/fc4-dependen
cies-for-cinelerra-cvs.html

After you make sure you ...snip...



Scott: Thanks for your detailed response. I'm going to experiment 
this week with this. I'm actually planning to wipe the slate 
clean next week and install FC6, but this will be good practice. 
Hope folks don't mind my asking questions as I get into this...


  


# This is not relocatable
%define prefix /usr
%define ver0 2.1.CV.svn.r909M
%define genopt_flags -O3 -pipe -fomit-frame-pointer -fprefetch-loop-arrays 
-funroll-loops -minline-all-stringops
%define ver1 athlonxp
%define rpmcflags %{genopt_flags} -march=athlon-xp -mmmx -msse -mfpmath=sse,387 
-ffast-math

Summary: A video editing software for linux.
Name: cinelerra
Version: %{ver0}
Release: %(date '+%Y%m%d').%{ver1}
License: GPL
Group: Applications/Multimedia
URL: http://cvs.cinelerra.org
Source0: %{name}-%{ver0}.tar.gz
BuildRoot: %{_tmppath}/%{name}-%{version}-%{release}-root
Requires: a52dec>= 0.7.4  
Requires: faac  >= 1.24 
Requires: faad2 >= 2.0 
Requires: libavc1394>= 0.5.1
Requires: libiec61883   >= 1.0.0
Requires: libraw1394>= 1.2.1
Requires: libsndfile>= 1.0.15
Requires: OpenEXR   >= 1.2.2
Requires: mjpegtools>= 1.8.0
Requires: ffmpeg>= 0.4.9
Requires: lame  >= 3.96.1
Requires: fftw  >= 3.1.1
# Not sure if this is REALLY needed
#Requires: fftw2 >= 2.1.5
Requires: libdv >= 0.104
Requires: libtheora >= 1.0
Requires: x264

BuildRequires: a52dec-devel>= 0.7.4  
BuildRequires: faac-devel  >= 1.24 
BuildRequires: faad2-devel >= 2.0 
BuildRequires: libavc1394-devel>= 0.5.1
BuildRequires: libiec61883-devel   >= 1.0.0
BuildRequires: libraw1394-devel>= 1.2.1
BuildRequires: libsndfile-devel>= 1.0.15
BuildRequires: OpenEXR-devel   >= 1.2.2
BuildRequires: mjpegtools-devel>= 1.8.0
BuildRequires: ffmpeg-devel>= 0.4.9
BuildRequires: lame-devel  >= 3.96.1
BuildRequires: fftw-devel  >= 3.1.1
# Not sure if this is REALLY needed
#BuildRequires: fftw2-devel >= 2.1.5
BuildRequires: libdv-devel >= 0.104
BuildRequires: libtheora-devel >= 1.0
BuildRequires: x264-devel


%description

CinelerraCV is a video editing software that is a fork of the official
version developed at http://heroinewarrior.com/cinelerra.php3. The
"CV" stands for community version, and this version exists due to the
the lack of support provided by heroinewarrior.

%prep
%setup -q -c

%build
./configure --prefix=%{prefix} --enable-x86 --enable-mmx --enable-freetype2 
--with-buildinfo=svn/recompile --with-plugindir=/usr/lib/cinelerra/ \
CFLAGS='%{rpmcflags}' CXXFLAGS='%{rpmcflags} -fno-check-new'

%install
rm -rf $RPM_BUILD_ROOT
make prefix=$RPM_BUILD_ROOT%{prefix} install
( ln -s /usr/bin/mpeg2enc 
$RPM_BUILD_ROOT%{prefix}/lib/cinelerra/mpeg2enc.plugin )


%clean
rm -rf $RPM_BUILD_ROOT


%files
%defattr(-,root,root,-)
%doc
%{prefix}/bin/cinelerra
%{prefix}/bin/mpeg3dump
%{prefix}/bin/mpeg3toc
%{prefix}/bin/mpeg3cat
%{prefix}/bin/mplexlo

%{prefix}/share/locale/*
%exclude %{prefix}/include/*

%{prefix}/lib/cinelerra/*.so
%{prefix}/lib/cinelerra/mpeg2enc.plugin
%{prefix}/lib/cinelerra/shapewipe/*

%exclude %{prefix}/lib/cinelerra/*.la

%{prefix}/lib/cinelerra/fonts/*

%{prefix}/lib/libguicast*
%{prefix}/lib/libmpeg3hv*
%{prefix}/lib/libquicktimehv*

%{prefix}/share/applications/cinelerra.desktop
%{prefix}/share/pixmaps/cinelerra.xpm




%changelog
* Thu Oct  5 2006 Pierre Marc Dumuid <[EMAIL PROTECTED]> 
- First package with merged Cinelerra 2.1

[CinCVS] vobview...

[Pierre Marc Dumuid]

Hi All,

On the IRC the other day, 
(http://cvs.cinelerra.org/irclog/index.php?file=2006-10-13.html) I 
mentioned that I suggested using vobview to see if it could split the 
mpeg file he was feeding to his camera.


I immediately backtracked my recommendation remembering that I had made 
a fair few changes to my version of vobview... 

I thought people may be interested in the hacks that I made, (they are 
in no way very well, and I am not really proud of the patches, and would 
re-do them myself.)


To get access to my patched version, I have included it in git.  To get 
my git archive, run:

cg-clone http://cvs.cinelerra.org/~pmdumuid/mygits/vobview/

BTW: I'm not really recommending vobview.  I just like it cause it's 
really fast way of seeing the rough content of my vob files, and doing 
rough splits... (Note: one of the commits removes the ability to set a 
left edge, this is because if the left edge is not zero, the audio and 
video can get be out of sync)


Regards,
Pierre Dumuid

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] limiting keyframe values

[Pierre Marc Dumuid]

muzzol,
Please just try it... then ask your questions.../ give your comments.

It's a bit hard to respond to this question because you might likely 
don't understand what the patches have implemented and they may have 
implemented what you are thinking anyhow.


If you really don't  revert back to your current SVN version if you 
really don't like it!


pierre

muzzol wrote:

i dont really get it. why do you need to fit keyframes if can't go out
of boundaries?

2006/10/14, Ichthyostega <[EMAIL PROTECTED]>:

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1

muzzol schrieb:
> will this fix make disappear "Fit keyframes" (ALT+F) action or i im
> missing something?
>
No, rather it makes it work as always expected (in my opinion)...
Same for the Fades, which are now forced to be within the sensible 
range.

I think, this implementation of "automation ranges" as of r908 was a
big improvement.

Hermann

-BEGIN PGP SIGNATURE-
Version: GnuPG v1.4.5 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFFMT8MZbZrB6HelLIRAl7NAJwNgNhoN+odi5GuEdJyPTOAm6if7QCdHKBY
EnNYkQWnY54fHKwQV54zWSU=
=FW6j
-END PGP SIGNATURE-

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra







___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] limiting keyframe values

[Pierre Marc Dumuid]

This should have been fixed in r908..

Pierre

muzzol wrote:

hi!

i think this has been discused before but i can't find proper threads.

is there a way to limit values for a keyframe for specific settings?
for example, i can apply more than 100% oppacity and less than 0% and
that doesn't make any sense.

is this a known bug?

regards,

muzzol




___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] 720p camera, editing machine, and cinelerra

[Pierre Marc Dumuid]

I had similar questions when I bought my new computer with no responses.

It may be good to declare a test suite, that we could publish our 
performances so we can all start building up an idea of what is good for 
new PC's..



Ed Colmar wrote:

Greetings Cinelerra users!

I have been experimenting around with Cinelerra for about a year now 
on a wimpy athlon 1.3 ghz machine.  I have grown very fond of the 
software, and I feel like I have a good enough grasp of it to justify 
some new equipment purchases.  Though I am still very much a newbie to 
video, I do have experience with underlying concepts from audio 
editing, and graphic design.


I read through the archives and noticed a few HD threads that gave me 
some hints, but I wanted to start a quick thread about the setup that 
I am considering.  Hopefully, I can get some insight and avoid any 
pitfalls with the advice of you all.


I am really interested in the Sanyo Xacti VPC-HD1A.  It fits my 
budget, and some of the extra features seem like they would be of 
great use.  This would put 720p HD (as well as 640x480) source in my 
hands in mpeg4 format.
I don't have a huge budget for an editing machine, so I am aiming to 
get a reasonable performance quality of out the cheapest solution.  I 
was considering a dual core athlon64.  Is this adequate?  I do have 
several athlon boxes, and one athlon 64 lying around for render farm 
use once edits are complete.


Questions for the group...  Do you see any problems with connectivity 
between this setup and cinelerra?  Is there a better price/performance 
solution for my editing machine?  Any other advice you can offer me?


Thanks!

-ed

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] a bug fixed

[Pierre Marc Dumuid]

Oops!  That was supposed to be replied to a personal email address.

Please .. non-devel's  and non-tester's don't use the git unless you are 
planning on bug-reporting / testing... and contact Christian (cehteh on 
IRC) or myself if you plan to do so.


The machine that is hosting the git is on a bandwidth limited machine, 
and we don't want everyone using up all his bandwidth!


Pierre

Pierre Marc Dumuid wrote:

Hey Andraz!

Not me. BTW... If you're interested in working with git... see the 
pages at


http://www.pipapo.org/pipawiki/Cinelerra
and
http://www.pipapo.org/gitweb

Christian has set up a auto-syncing git for us who like working with 
git..


Regards,
Pierre

Andraž Tori wrote:

by the way... in svn today i also fixed a annoying bug which came up
when dragging assets or clips from asset window to main window and there
were more armed tracks than used in dragged asset/clip.

The bug was that part of unused tracks were cut out ... now they are not
anymore.

i also have a question to other developers: is anyone working on briging
native cinelerra-to-mpeg-ready-for-dvd export? using ffmpeg or
something?

bye
andraz


___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra
  



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] a bug fixed

[Pierre Marc Dumuid]

Hey Andraz!

Not me. BTW... If you're interested in working with git... see the pages at

http://www.pipapo.org/pipawiki/Cinelerra
and
http://www.pipapo.org/gitweb

Christian has set up a auto-syncing git for us who like working with git..

Regards,
Pierre

Andraž Tori wrote:

by the way... in svn today i also fixed a annoying bug which came up
when dragging assets or clips from asset window to main window and there
were more armed tracks than used in dragged asset/clip.

The bug was that part of unused tracks were cut out ... now they are not
anymore.

i also have a question to other developers: is anyone working on briging
native cinelerra-to-mpeg-ready-for-dvd export? using ffmpeg or
something?

bye
andraz


___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra
  



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] Blue Dot

[Pierre Marc Dumuid]

private message sent.
If anyone wants to know where a git repository is that this patch can be 
checked out, let me (or Christian) know.. (there is a new git-repository 
that is hosted at a bandwidth limited place... but is freely available 
for developers and reviewers... just talk to us in person first...)


Pierre

Miha Kitic( wrote:

I am not sure I can help you with coding.
I want to learn very much, but I am still very new
to Linux. 


As for examining it...
I guess it won't hurt to have a look.
So if you have any good reference, please let me know.

All the best!

Miha 


On sre, 2006-10-11 at 22:22 +0930, Pierre Marc Dumuid wrote:
  

It's not in a compatible state to work with 2.1..
sorry!
If you know how to code and want to fix it, I have a branch that you 
could examine..


Miha Kitic( wrote:


Hi all!

Now I have one question.

How do you compile Cinelerra with Blue Dot theme.
I tried with:

./configure --with-bluedottheme

But nothing really happened :(
I can not select it under preferences.

Someone can tell me where did I go wrong?

Miha


___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra
  
  

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra





___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra
  



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] Audio Question...

[Pierre Marc Dumuid]

you can shift the wav-file in time, (zoom right in, and get the 
waveforms to match by inserting space into either track.


Once it's in sync, then you'll need to work out the ratio of the 
sampling rates and get them to match... If in one recording the a second 
event, (say 40 minutes into the recording) occurs at 40:20 minutes, and 
40:40 in the other channel, then you'll need to adjust make the ratios 
of the sample rates (40*60+20)/(40*60+40) ..

(if they are the same).

I've had issues with cinelerra' resampler at times, so you may need to 
use sox to do a better quality version..

Pierre

Miha Kitic( wrote:

 This assumes that the camera records no sound, which is not the case
for video cameras.  It is much easier to synchronise two sounds with
each other, very accurately:




That's very good idea I haven't thought of...
I will keep it in mind.

Thanks!

Miha


___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra
  



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] Blue Dot

[Pierre Marc Dumuid]

It's not in a compatible state to work with 2.1..
sorry!
If you know how to code and want to fix it, I have a branch that you 
could examine..


Miha Kitic( wrote:

Hi all!

Now I have one question.

How do you compile Cinelerra with Blue Dot theme.
I tried with:

./configure --with-bluedottheme

But nothing really happened :(
I can not select it under preferences.

Someone can tell me where did I go wrong?

Miha


___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra
  



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

[CinCVS] new feature, smoothautos insert and a git repository..

[Pierre Marc Dumuid]

Hi all,

Whilst considering the implementing of the "audio minimum auto to -96 
dB" patch, I considered that it was possible to make a patch that can 
automatically convert your old projects automation curves into new 
range.  The means of doing this was as follows:


As soon as the audio curves transition over -40 dB, a keyframe could be 
inserted and then a second keyframe 1 frame / sample after the -40 dB 
transiton could be inserted that sits at -96 dB.


This would in effect keep the curve the same for everything above -40 dB 
(so people could convert to the new range without any concerns) but also 
ensure that what was silent in the past is still silent (very important!).


In order to implement this idea, it was required that a new keyframe 
inserted at any point didn't modify the existing beizer shape. I 
expected this code to already exist within the cinelerra codebase, but 
it didn't. Thus I started makeing another patch:  smoothautos.


The smoothauto's patch is described as:

This branch adds the ability to insert autos without disrupting the 
existing beizer shape. It also adds a toggle button that toggles between 
non-disruptive, and zero slope when adding auto's.  Zero slope is 
convenient for fade curves when you insert a keyframe to force the sound 
/ video to zero.  you don't really want the keyframe smooth as you would 
be dragging it to zero anyhow..


I have recently implemented a smoothauto patch, and it is ready to be 
tested.  Rather than pushing it into svn for testing, Christian Thäter 
(cehteh on IRC) has set up a git repository on his server that allows 
developer's to push their branches to the server.  His server maintains 
a 12 hour sync branch with the cinelerra svn also.


The server that Christian Thäter has set up is on limited bandwidth so 
instructions on how to obtain the git are not included in this email.  
But if people are interested in testing this patch and having a git 
access to the cinelerraCV development, please contact [EMAIL PROTECTED] and 
myself to find out how to do so.


TESTERS ARE VERY WELCOME!

I will wait a few days before comiting the patch so if you are 
interested in testing, (or developing using git in general) please 
contact us..


Regards,

Pierre Dumuid


___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] Exporting to Quicktime DV

[Pierre Marc Dumuid]

YUV is raw data, so that means massive file size.
The idea is that you can pipe from cinelerra using a YUV pipe directly 
into encoders (such as ffmpeg etc.) I definitely don't suggest it as a 
format to save to!! (unless you're living in 2200 with massive HDD space..)


Matt Pfingsten wrote:
On 10/8/06, *Pierre Marc Dumuid* <[EMAIL PROTECTED] 
<mailto:[EMAIL PROTECTED]>> wrote:


I use yuvpipe, and let an external program do it.. (ffmpeg /
mencoder /
mpeg2enc)
... DV is actually a lossy compression so I personally think it's
better
(and quicker) whilst YUV is lossless.. so I think it is better.

 
Cool. I was not previously aware that YUV is lossless. Would it be 
acceptable for editing?


--
Matt Pfingsten
www.GotWookiee.com <http://www.GotWookiee.com>



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] Exporting to Quicktime DV

[Pierre Marc Dumuid]

I use yuvpipe, and let an external program do it.. (ffmpeg / mencoder / 
mpeg2enc)
... DV is actually a lossy compression so I personally think it's better 
(and quicker) whilst YUV is lossless.. so I think it is better.




Matt Pfingsten wrote:
Unfortunately, that requires recompression. I'm trying to export to a 
Quicktime DV (or directly to MPEG2 so I can mplex it and burn it to 
DVD, but that doesn't work either). The problem that I am having is 
that I have various clips that all claim to be Quicktime's ocompressed 
with the DVC codec but each one is somehow how different.


However, when I go to export the video for Internet playback, I will 
definately find your table to be most helpful. Thanks.


On 10/8/06, *Scott C. Frase * <[EMAIL PROTECTED] 
> wrote:


Matt,
I started creating a doc on QT compatibility by exporting video from
Cinelerra using different compression schemes with the QT container
format:
http://content.serveftp.net/video/qtcompatibility.ods.html
http://content.serveftp.net/video/qtcompatibility.ods (Open Office
Calc)

For now, I've found MPEG4/JPEG photo compression schemes most
compatible
among the major media players.  Note that I've used twos complement as
my audio stream.  You'll probably get even better results from MP3/4,
but I just haven't had time to test these.

hope this helps,
scott

On Sat, 2006-10-07 at 22:43 -0500, Matt Pfingsten wrote:
> A while back I was having issues importing QuickTime DV files
that had
> been captured on a Mac with Final Cut Pro. I learned that,
apparently,
> in spite of the fact that Cinelerra both claimed they were
compressed
> with the DV codec and were quicktime files, someohow the FCP
quicktime
> was different. Cinelerra, for whatever reason, wouldn't read all of
> the frames. Playback appeared choppy, but even when advancing a
frame
> at a time only evey fourth of fith frame would actually display. I
> figured that Apple, being the creators of Quicktime, had done
> something different to their Quicktimes that the Linux efforts
hadn't
> caught up to just yet. MPlayer/MEncoder could read it fine, so I
> managed to export it to an AVI DV, although Cinelerra proved
unable to
> play anything pass the first GB of data.
>
> Well, in the last few days I've learned more. It's not just FCP that
> creates "different" DV compressed Quicktimes. There appear to be
many
> different DV Quicktimes, each claiming to be the same thing but
> obviously different in some way. Here is a comparison between three
> different clips:
>
> 1) A file called fhmp_footage.mov. This first file is some footage
> captured in Final Cut Pro.
> -In Cinelerra, playback is choppy and only every fourth or fifth
frame
> is displayed.
> -In MPlayer it plays fine. "mplayer -identify fhmp_footage.mov"
yields
> this: (relevant portions only): ID_VIDEO_FORMAT=dvc, and later
> ID_VIDEO_CODEC=qdv
> -FFMpeg cannot play it, complaining "could not open codecs."
>
> 2) A file called 16A-1.mov. This is a clip I shot two weeks ago and
> captured using dvgrab.
> -Cinelerra plays it back fine. No issues.
> -In MPlayer it plays fine. "mplayer -identify 16A-1.mov" yields
this:
> (relevant portions only): ID_VIDEO_FORMAT=dvc, and later
> ID_VIDEO_CODEC=qdv
> -FFMpeg can play it. "ffmpeg -i 16A-1.mov" identifies the video as
> dvvideo. FFPlay plays it fine.
>
> 3) A file called titles_end.mov. This clip was created in Cinelerra
> from a bunch of PNG files. It is a one minute title sequence.
> -Cinelerra can play it back fine. No issues.
> -MPlayer/MEncoder have no problems. "mplayer -identify
titles_end.mov"
> yields this: (relevant portions only): ID_VIDEO_FORMAT=dvc, and
later
> ID_VIDEO_CODEC=qdv
> -FFMpeg can't play it, complaining "could not open codecs."
>
> Cinelerra identifies all three as DV Codec compressed Quicktimes. It
> should also be noted that Kino 0.9.1 can't open any of these files
> without complaining that the file isn't a DV file and wanting to
> import it via MEncoder. Kino 0.7.6 can open all of these files,
> although it takes about five minutes to open the first file
> (fhmp_footage.mov) probably because it is 13 GB.
>
> At the present time this is preventing me from burning my dailies or
> cuts to DVD which I could like to do before Tuesday. The biggest
issue
> is why Cinelerra is exporting Quicktimes that can't be read by
> others.
>
> I remember reading something in the archives about dvc, dvsd and
> multiple headers or something being written into one file. Of
course,
> two hours of searching has failed to produce this message now that I
> needed it.
>
> Anyone have any idea what 

Re: [CinCVS] audio fade line bug in r909

[Pierre Marc Dumuid]

In the latest release of SVN, you'll find the slider range is the same 
as the automation range for audio curve.


[EMAIL PROTECTED] wrote:

Guys,
FYI - for my projects, I fade audio down to -60db, as I still hear noise around 
-40 to -50db.  It would be nice if the fader to the left of the track went down 
to this level.

scott
 -- Original message --
From: rafael2k <[EMAIL PROTECTED]>



  




Subject:
Re: [CinCVS] audio fade line bug in r909
From:
rafael2k <[EMAIL PROTECTED]>
Date:
Fri, 6 Oct 2006 06:41:57 +
To:
cinelerra@skolelinux.no

To:
cinelerra@skolelinux.no


i agree.
lets choose a way to do and use that way.


bye,
rafael diniz

Em Quinta 05 Outubro 2006 22:16, Pierre Dumuid escreveu:
  

Rafael2k,

I could "quickly submit a commit" that makes the default range -40 dB,
but I was gonna wait until the backward compatibility patch was
implemented. (the patch consists of looking for a project-setting for
the NEGINFINTYGAIN, and if it wasn't set, set it to -40 dB, whilst all
new projects should have a default of -96 dB being the maximum
attenuation possible for 16 bit audio)

Looking at the svn history, I was a bit weary that there are so many
patches for this auto-curves stuff, so I wanted to limit the commits for
this topic.

Pierre

rafael2k wrote:


thanx all,
now, with the right range, its ok.
btw, the default range of audio can be changed to 'what work' :-)

after the r836, i'm uploading the new cinelerra version to
i'm uploading the cinelerra packages for slackware 11.0 to:
http://slack.sarava.org/packages/slackware/slackware-11.0/multimedia/
i'll write a readme for instalation.

to compile yourself, use the slackbuilds:
http://slack.sarava.org/slackbuilds/

bye,
rafael diniz

Em Quinta 05 Outubro 2006 20:13, Pierre Marc Dumuid escreveu:
  

At the moment, the default range I set for the audio faders -80 to 6
dB.  A few days later I discovered that internally anything less than
-40 dB cinelerra forces to -Inf dB, hence I created a patch that makes
anything less than 40 dB to be clipped at 40 dB to indicate to the user
that this is a low the sound needs to go.  ( The -40 dB was to be
changed to -96 dB at some point later with a backward compatibilty
patch.)

Anyhow, this is the reason that the fade curve doesn't go below -40 dB
(being halfway down the track).   Please change the -80 dB in the audio
range to -40 dB if that helps. You will notice you can place a keyframe
below -40 dB, and this is valid, it causes the auto curve to have a
steeper shape than if the keyframe sat at -40 dB.

Pierre

rafael2k wrote:


hi all,
i'm facing a bug in the audio fade line in my fresh slack 11 w/ the
r909.

the range is correct, the keyframes can be placed ok, but the white
line do not go down the center of the track.

anyone noticed that?

bye,
rafael diniz
  


  



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] audio fade line bug in r909

[Pierre Marc Dumuid]

At the moment, the default range I set for the audio faders -80 to 6 
dB.  A few days later I discovered that internally anything less than 
-40 dB cinelerra forces to -Inf dB, hence I created a patch that makes 
anything less than 40 dB to be clipped at 40 dB to indicate to the user 
that this is a low the sound needs to go.  ( The -40 dB was to be 
changed to -96 dB at some point later with a backward compatibilty patch.)


Anyhow, this is the reason that the fade curve doesn't go below -40 dB 
(being halfway down the track).   Please change the -80 dB in the audio 
range to -40 dB if that helps. You will notice you can place a keyframe 
below -40 dB, and this is valid, it causes the auto curve to have a 
steeper shape than if the keyframe sat at -40 dB.


Pierre

rafael2k wrote:

hi all,
i'm facing a bug in the audio fade line in my fresh slack 11 w/ the r909.

the range is correct, the keyframes can be placed ok, but the white line do 
not go down the center of the track.


anyone noticed that?

bye,
rafael diniz

  



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] audio levels

[Pierre Marc Dumuid]

16bit -> 2^16 = 65536

Which means that there is exactly 65536 possible levels
that sound signal can occupy. But because 1 bit is used
for the sign (+ or -) we can actually only use one half,
32768 levels. 


To put it in simple words - to make the loudest noise
completely silent, you must divide it with more than 32768
in order to make it 0.
  
hmm, thinking about it now, you need to divide it by 96 dB to get [32768 
-32768] to go to [0.5 -0.5] which would round down to zero.  Dividing by 
90 dB would get to [1 -1].  Going with 96 dB is probably the go (and 
people recognise that number).

But if this will be applied, your keyframes set to -40dB may be heard.
(-40dB = 100  --->  32768 / 100 = 327,68)
Of course things now work in such a way (according to Pierre), that any
sound divided by 100 or more is automatically muted (how this is
realized, I don't know).
  

It is realize by lines:

   if(fade_value <= INFINITYGAIN)
   value = 0;
   else
   value = DB::fromdb(fade_value);

in cinelerra/virtualanode.C  where INFINITYGAIN is defined in units.h as 
-40.



It least this is how I understand things. But if I am wrong, someonen
please correct me.
  

Seems rightish to me.

Again, please use the latest SVN if you want sane methods of 
controlability of the automation curves! 



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] The term "Insertion Point" is used throughout the manual..

[Pierre Marc Dumuid]

There is already docs distributed with the SVN sources (look in the doc 
directory).

Would be better if non-coder's could tackle the documentation area.

Pierre


Gour wrote:

On Sat, 2006-09-30 at 09:22 +0930, Pierre Marc Dumuid wrote:


  
I propose replacing the occurence of "insertion point" throughout the 
manual with "timeline cursor", or even just "cursor"



I agree. Having good documentation is as important as program's
features.


  
I understand there are 3 manual locations, (HV's, SVN's, and wiki).  Do 
other's agree with the change in terminology.



Is is possible that CV of Cinelerra ships with some complete manual -
(X)HTML or even PDF?


Wiki is nice tool for getting input from many users, but it would be,
imho, very nice to have program shipped with fine user's manual instead
of forcing users to look at different places which are (sometimes) not
in sync with the latest released version?

I do not know whether Twiki has such ability, but eg. when I was taking
look Drupal some months ago, I noticed ability to export content in
DocBook and then get (X)HTML and/or PDF.

Sincerely,
Gour

  



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] audio levels

[Pierre Marc Dumuid]

20*log10(2^24) = 144.5 dB

though if you consider that 1 bit is taken up for the sign (i.e. + or - 
) then

20*log10(2^23) =  138.5 dB

( Note how 6 dB = 2x )

Pierre


Herman Robak wrote:
On Sat, 30 Sep 2006 17:45:23 +0200, Pierre Marc Dumuid 
<[EMAIL PROTECTED]> wrote:


I discoved that this is for the meter only, and the minimum that 
cinelerra forces to -Inf dB is -40 dB!! See cinelerra/virtualanode.C 
which contains:

if(fade_value <= INFINITYGAIN) fade_value = 0;

and INFINITYGAIN is defined in guicast/units.h as:
#define INFINITYGAIN -40

...
Prior to discussions with j6t on IRC, I had been considering setting 
the "INFINITYGAIN" as a property of the current project (thus 24 bit 
audio freaks could specify huge attenuations).


 I would like to know, with reference to authoritative sources, what the
canonical dynamic range for 24 bits sampling is.  Is it linear, like 
16bit?


--Herman Robak

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] audio levels

[Pierre Marc Dumuid]

Current SVN should now be more workable (see thread as I actually 
described the issue as being a fact of shared range), since automation 
ranges are more sane.


At the moment if you drag to -40 dB you force to -Inf dB. I recently 
(last night for me in Aussie) applied a patch that limits' the 
automation range to this -40 dB in the time-line to indicate that this 
is as far as needed to go (see r908).


Pierre

Miha Kitič wrote:

But there may be a workaround for you...

How about using the combination of faders and mute function.
For instance... you fade the audio level to very low level and at the
end of it set a mute. That should kill all audio for sure.

Of course, I assume that mute will actually switch
of the audio channel, but I am not sure. 
Perhaps someone who is very familiar with the sources can answer that.


Miha

On sob, 2006-09-30 at 11:34 -0700, Robert Persson wrote:
  

On Sat, 2006-30-09 at 16:51 +0200, Herman Robak wrote:


Someone mentioned the fader did not completely mute the sound at
the lowest level; that would be a bug! 
  

Forgive me if I say something that has already been said before, since I
have only just joined this list, but I have encountered this problem
with the faders (I'm assuming we're talking about the audio level graphs
in the timeline rather than sliders or knobs) and it is an absolute
blocker for me -- inability to fade to silence makes a video editor
absolutely useless for what I need to do. As a result I have wasted a
lot of time trying to deal with windows video editors and ended up in
all kinds of trouble as a result. I filed a bug report about it a couple
of days ago, but I am actually really surprised that no one else has
done this already. I wondered for a long time whether I had missed some
obvious workaround.

What has made me want to try cinelerra again is that I accidentally
discovered on an old live cd distro that a much older version of
cinelerra (from around 2004) doesn't have this bug. It looks very much
like a regression in the 2.x series.

Does anyone have any idea how long it will take to get this ironed out?
If not, am I likely to be able to get an older version cinelerra working
reliably (and without too much pain) on an up-to-date Linux distribution
with recent versions of all its libraries?

Best
Robert




___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra






___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra
  



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] Doubtful fade keyframe

[Pierre Marc Dumuid]

Oops, I just realized that the video fade actually controls the opacity, 
(alpha) and isn't really a gain as such.  When the alpha goes to 101, it 
wraps around and perfectly explains what the Raffaella saw (and I can 
replicate).  I have just applied a patch to fix this.


Pierre.

Pierre Marc Dumuid wrote:

Huh,
depends on how you look at it.  I personally like to be able to gain 
above 100% (which results in the user needing to be wary of 
wrap-around), but at the same time, I don't want it to saturate 
because I know that for saturation, the following code:


value = value > 255 ? 255 : value;

needs to be executed on every pixel on every frame, which I thought 
may slow things (not sure how much though).  Wrap-around can results 
in some wierd effects as well.. (which some people might call 
"artistic" :) )


Pierre

Herman Robak wrote:
On Sat, 30 Sep 2006 01:07:28 +0200, Pierre Marc Dumuid 
<[EMAIL PROTECTED]> wrote:


Which version of cinelerra are you using (community or HV,  and if 
community, which SVN revision number).


In the latest version of cinelerra-CV, the automation ranges have 
recently been fixed (within the last few weeks)


if you are using an older version of cinelerra-CV ensure that your 
automation range, (the textbox to the left of the "Selection Start" 
box) is 0-100 when editing video.
if it is say 0-700, the fade curve could possibly be cycling through 
7 times,


 Just getting this straight: Wraparound is a bug, not a feature, right?

If Cinelerra does a wraparound on brightness or colour values, that
behaviour should always be filed as a bug.  Cinelerra should saturate,
not wrap around.

--Herman Robak

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] audio levels

[Pierre Marc Dumuid]

In
Cinelerra, there is a configuration setting for how low you want the level
meters to go.
  
I discoved that this is for the meter only, and the minimum that 
cinelerra forces to -Inf dB is -40 dB!! See cinelerra/virtualanode.C 
which contains:

if(fade_value <= INFINITYGAIN) fade_value = 0;

and INFINITYGAIN is defined in guicast/units.h as:
#define INFINITYGAIN -40

I proposed on the IRC to change this to -96 to account for 16 bit audio, 
(the most people are likely to work with).  j6t and i agreed to this 
idea, but more recently, I have realised that people loading old 
projects with levels of -40 dB would suddenly find that the audio 
signals will not be forced to -Inf dB any more, and thus they will hear 
sounds. 

Prior to discussions with j6t on IRC, I had been considering setting the 
"INFINITYGAIN" as a property of the current project (thus 24 bit audio 
freaks could specify huge attenuations).  j6t advised against this 
(micromanagement), but on considering the problem of old projects 
issues, I wish for this to be re-considered.  What do people think?  I 
personally see the idea of having a per-project minimum dB as good, both 
for backward compatibilty, as well as for professionals..


I was going to also make the default for new projects independent of the 
last project, (i.e. the default for a new project would always be -96 dB.)


Pierre


What numbers have I to write if I want silence when the white fade line
is at the bottom margin of the track and a slightly too loud sound when
the line is at the top margin?



The defaults should be like that already.
  



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] Doubtful fade keyframe

[Pierre Marc Dumuid]

Raffaella,
I advise upgrading to the lastest SVN version..  In the older versions, 
(prior to r898 , see http://cvs.cinelerra.org/svn_log.php) the 
automation range was used the same for all automations, which didn't 
make sense, (cause audio typical range is -40 - 6 dB, whilst video 
generally has a range of 0 - 100, translations -100 - 100, and zoom 
0.01  - 8.  So depending on which curves you wanted to work with, you'd 
have to select the range each time.  After the 898 patch, the ranges 
have been seperated according to the above.


For the audio track's I'd have to have a look at the code to see if 
saturation is in effect.  Note that at the moment, a gain of -40 dB 
(which really mean a gain of .0001) is forced to -Inf dB (which really 
means a gain of 0).


Pierre

Raffaella Traniello wrote:

Thanks Pierre!

I've the pleasure of understanding in all my body!!

I'm using Cinelerra 2.0 CV (no idea of which SVN revision number).
It took me a while to get to the automation range, but at the end I got
there.
It was set to -1264.00 to 1392.00.

Now that I set it 0-100 everything in the video track is logical, clear,
obedient.
Fader range and keyframe line vertical range go toghether.

If I set the automation range to 0-200 the keyframe line vertical range
will cover 2 dark-bright cycle. Only the first one will be shown on the
fader. 
What can be the use of settings that differ from 0-100? 


The brightness cycles go from dark to correct brightness. Is it possible
to set it from dark to white (in order to fade out to white or get an
overexposed effect)?

When I launch Cinelerra, before importing my project, it shows the value
I last used. Is 0-100 the basic value, the default one?

Where can I learn more about wrap-around, colour range, colour gain?

I can see that for audio tracks things look less clear and logical.
But that's another chapter.

Many many thanks
Raffella


___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra
  



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] Doubtful fade keyframe

[Pierre Marc Dumuid]

Huh,
depends on how you look at it.  I personally like to be able to gain 
above 100% (which results in the user needing to be wary of 
wrap-around), but at the same time, I don't want it to saturate because 
I know that for saturation, the following code:


value = value > 255 ? 255 : value;

needs to be executed on every pixel on every frame, which I thought may 
slow things (not sure how much though).  Wrap-around can results in some 
wierd effects as well.. (which some people might call "artistic" :) )


Pierre

Herman Robak wrote:
On Sat, 30 Sep 2006 01:07:28 +0200, Pierre Marc Dumuid 
<[EMAIL PROTECTED]> wrote:


Which version of cinelerra are you using (community or HV,  and if 
community, which SVN revision number).


In the latest version of cinelerra-CV, the automation ranges have 
recently been fixed (within the last few weeks)


if you are using an older version of cinelerra-CV ensure that your 
automation range, (the textbox to the left of the "Selection Start" 
box) is 0-100 when editing video.
if it is say 0-700, the fade curve could possibly be cycling through 
7 times,


 Just getting this straight: Wraparound is a bug, not a feature, right?

If Cinelerra does a wraparound on brightness or colour values, that
behaviour should always be filed as a bug.  Cinelerra should saturate,
not wrap around.

--Herman Robak

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

[CinCVS] The term "Insertion Point" is used throughout the manual..

[Pierre Marc Dumuid]

... and I think this is incorrect.
In my discussions with Nicolas on IRC, I got confused at his use of the 
term "insertion point" for what I'd call in my discussions with people 
the "cursor".


He then directed me to the manual, and I saw the definition of INSERTION 
POINT was the "flashing cursor".


This I find REALLY confusing, as:

To demonstrate:
Press the "[" key, then move the cursor to another location, then select 
"file->Load files.."  select some files, and set insertion strategy to 
"Paste at insertion point".


You'll see the insertion point in this case is not the "flashing cursor" 
but really, here the definition of "insertion point":


if (in_point_defined)
 insertion_point = in_point;
else
 insertion_point = time-line cursor position;
end

I propose replacing the occurence of "insertion point" throughout the 
manual with "timeline cursor", or even just "cursor"


I understand there are 3 manual locations, (HV's, SVN's, and wiki).  Do 
other's agree with the change in terminology.


Pierre

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] Doubtful fade keyframe

[Pierre Marc Dumuid]

Which version of cinelerra are you using (community or HV,  and if 
community, which SVN revision number).


In the latest version of cinelerra-CV, the automation ranges have 
recently been fixed (within the last few weeks)


if you are using an older version of cinelerra-CV ensure that your 
automation range, (the textbox to the left of the "Selection Start" box) 
is 0-100 when editing video.
if it is say 0-700, the fade curve could possibly be cycling through 7 
times, (though I doubt that would be the result, depends on the color 
model, and in my experience pumping the gain of video to cause it to 
wrap-around rather than smoothly fade). 


Definition:
wrap-around
- when a signal multiplied by a gain exceeds the valid output range, and 
(in the computer) is wrapped around (i.e. a value of 257 for a range of 
0-256 would be come 1)


and example:

Consider a video signal that was recorded dimly such that no level ever 
exceeded 128, (whilst the range for color is 0-256).  If one applied a 
gain of 200 % to the track, this would cause the output video range to 
never exceeed 256 which is valid and keeps the output signal within the 
.  But if the user applied a gain of 210 % then the video pixels with an 
amplitude  of 128 would be 269, but wrap-around (within the computer) 
would result in this becoming 13 (= 269-256).  But this is only for that 
specific pixel!


P.S. I haven't seen many videos that are recorded that dimly!!

Raffaella Traniello wrote:
Thanks Nicolas. 

  

If I reposition it vertically I'll move the whole flat curve. At the
bottom end of the track I'll get a dark image, at the top end I'll get a
bright one. But in between I'll have 7 (seven!) cycles of dark to bright
image. What's for?
  

Could you show us a screenshot?


Not without hours of learning.
In the video track I set the white line shaped like this:
 
/
___/

with the beginning at the bottom border of the video track and the end
at the top border. I used all the available space, the maximum height.
If I playback the timeline I see the blinking effect in the compositor
(if I play it slow I can see the cycles).

I tried it again with another footage.
The mouse cursor never changed shape over the white line.
I had to launch Cinelerra 3 time before being obeyed. 
I had 25 light cycles this time.


Something funny is going on here...
What do you think?

Ciao 
Raffaella



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra
  



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

[CinCVS] people having problems with the --with-buildinfo flag

[Pierre Marc Dumuid]

I noticed someone having issues with the new --with-buildinfo configure 
option.  Cinelerra should definitely compile w/out specifying this flag!


If you are having difficulties, please try running:

rm cinelerra/Makefile cinelerra/.deps/main.*  cinelerra/.deps/aboutprefs.*;
./autogen.sh
/configure 
make

I've committed a patch that might make it work better for some.

If all else fails, and you don't want to wait around for it to be fixed, 
create a file cinelerra/versioninfo.h with the following contents:

more cinelerra/versioninfo.h
#define BUILDDATE "Thu Sep 28 10:37:57 CST 2006"
#define REPOMAINTXT  " (C) 2006 Heroine Virtual Ltd.\nCompiled on 
Thu Sep 28 10:37:58 CST 2006"

#define REPOABOUTPREFTXT ""

For those who make packages, I suggest a versioninfo.h file like:
#define BUILDDATE "Thu Sep 28 10:37:57 CST 2006"
#define REPOMAINTXT  "SVN 907 (C) 2006 Heroine Virtual 
Ltd.\nCompiled on Thu Sep 28 10:37:58 CST 2006\nFrom package  
cinelerra_2.1.0-1svn20060928_i386.deb for debian on an athlonxp built by 
Andraz Tori"
#define REPOABOUTPREFTXT "SVN 907: package 
cinelerra_2.1.0-1svn20060928_i386.deb for debian on an athlonxp built by 
Andraz Tori"


or come up with your own.
Pierre

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

[CinCVS] Tooltip patch for the zoombar

[Pierre Marc Dumuid]

Hi all,
I'm was writing a response to 
http://bugs.cinelerra.org/show_bug.cgi?id=326 and realised that 
cinelerra doesn't have tool tips for the zoom panel.
I've made a patch that I'm about to commit that gives these tooltips, 
but was wanting to run the tooltips by people before committing them


The patch is:
http://cvs.cinelerra.org/~pmdumuid/tooltips.diff

Let me know what you think of the proposed tooltips, they are:

Duration visible in the timeline
Audio waveform scale
Height of tracks in the timeline
Automation Type (Select which automation type to view / adjust in the 
range textbox)

Automation range minimum adjuster
Automation range
Automation range maximum adjuster
Selection start time
Selection length
Selection end time

Will commit the patch in a few days.  If anyone wants to modify the 
patch to include the language translations, send it to me, and I'll 
commit a more complete patch..


Regards,

Pierre Dumuid

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] interlaced zoom for nearest neighbour..

[Pierre Marc Dumuid]

 ...

 Would it be much harder to make it work with interpolated scaling, too?
Probably not. 



A (VERY SHORT) sample comparison is available at:

http://cvs.cinelerra.org/~pmdumuid/interzoom/myzoom.mov
http://cvs.cinelerra.org/~pmdumuid/interzoom/origzoom.mov


 A very obvious improvement, indeed.  Any effect that affects the
image spatially (that would be blur and sharpen, too) should be
interlace-aware.
Yes!!  but there's a LOT of work involved and currently most people 
consider that the best means of combating interlacing is working at 50 
fps.. Somehow this doesn't feel logical to me though..



Would it make sense to extend Cinelerra's core to present each
frame as two vertically stacked fields to the plugins?
That could do the trick.  The other thing to be aware of is that zoom, 
being a function of time should be different for each field. (which in 
my case it wasn't). 

The code that I sent was only a test-of-an idea, and was relatively 
simple to implement.  There are likely a number of ways of tacking this 
issue, (also there's the possiblity of using a de-interlacer for each 
set of fields using both fields..  LOTS of work to implement though!



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

[CinCVS] interlaced zoom for nearest neighbour..

[Pierre Marc Dumuid]

Hi All,
It's always bugged me that when zooming into interlaced material, there 
is a type of moire effect, or you end up zooming into two bars.
Some time ago, I proposed the idea of zooming for two frames, the odd 
and the even frames. 
Here is that idea somewhat implemented for the nearest neibour method.
The patch isn't complete, (since it doesn't have an on/off button, not 
take advantage of the interlace settings that should be set)


Just wondering what people think about this idea..

To test it:
1. apply the patch (patch -p 1 < nearest_...patch ), make etc..
2. get some interlaced material.  Load it up, Select "Nearest neighbor 
enlarge and reduce" from preference->playback.
3. zoom in, and pan around.  After rendering the results. reverse the 
patch, (patch -R -p1 < nearest..patch) , make etc.. and compare.


A (VERY SHORT) sample comparison is available at:

http://cvs.cinelerra.org/~pmdumuid/interzoom/myzoom.mov
http://cvs.cinelerra.org/~pmdumuid/interzoom/origzoom.mov

I suggest playing them with mplayer using:

mplayer -vf pp=lb -speed .5 myzoom.mov
mplayer -vf pp=lb -speed .5 origzoom.mov



diff --git a/cinelerra/overlayframe.C b/cinelerra/overlayframe.C
index 64400ce..3ddc389 100644
--- a/cinelerra/overlayframe.C
+++ b/cinelerra/overlayframe.C
@@ -2255,7 +2255,7 @@ ScaleTranslateUnit::~ScaleTranslateUnit(
 {
 }
 
-void ScaleTranslateUnit::scale_array_f(int* &table, 
+void ScaleTranslateUnit::scale_array_x_f(int* &table, 
 	int out_x1, 
 	int out_x2,
 	float in_x1,
@@ -2269,6 +2269,31 @@ void ScaleTranslateUnit::scale_array_f(i
 		table[i] = (int)((float)i / scale + in_x1);
 }
 
+void ScaleTranslateUnit::scale_array_y_f(int* &table, 
+	int out_y1, 
+	int out_y2,
+	float in_y1,
+	float in_y2,
+	int interpolate)
+{
+	float scale = (float)(out_y2 - out_y1) / (in_y2 - in_y1);
+
+	float scale2 = (float)(in_y2 - in_y1)  / (out_y2 - out_y1);
+	table = new int[(int)out_y2 - out_y1];
+	int tmp = (int)in_y1 + out_y1;
+
+	if (1)
+		for(int i = 0; i < out_y2 - out_y1; i++)
+			table[i] = (int)( (float)i * scale2/2 ) * 2 + (int) in_y1 + (tmp + i & 0x01);
+	else
+		for(int i = 0; i < out_y2 - out_y1; i++)
+			table[i] = (int)((float)i / scale + in_y1);
+
+//	for(int i = 0; i < out_y2 - out_y1; i++)
+//		printf("table: %d %d %d\n", i, out_y1 + i, table[i]);
+
+}
+
 void ScaleTranslateUnit::process_package(LoadPackage *package)
 {
 	ScaleTranslatePackage *pkg = (ScaleTranslatePackage*)package;
@@ -2298,17 +2323,18 @@ void ScaleTranslateUnit::process_package
 //printf("ScaleTranslateUnit::process_package 1 %d\n", mode);
 	if(out_w != in_x2 - in_x1)
 	{
-		scale_array_f(x_table, 
+		scale_array_x_f(x_table, 
 			out_x1, 
 			out_x2,
 			scale_translate->in_x1,
 			scale_translate->in_x2);
 	}
-	scale_array_f(y_table, 
+	scale_array_y_f(y_table, 
 		out_y1, 
 		out_y2,
 		scale_translate->in_y1,
-		scale_translate->in_y2);
+		scale_translate->in_y2,
+		0);
 
 
  	if (mode == TRANSFER_REPLACE && (out_w == in_x2 - in_x1)) 
diff --git a/cinelerra/overlayframe.h b/cinelerra/overlayframe.h
index 7e2c14d..32dac4a 100644
--- a/cinelerra/overlayframe.h
+++ b/cinelerra/overlayframe.h
@@ -269,12 +269,19 @@ public:
 	~ScaleTranslateUnit();
 
 	void process_package(LoadPackage *package);
-	void scale_array_f(int* &table, 
+	void scale_array_x_f(int* &table, 
 		int out_x1, 
 		int out_x2,
 		float in_x1,
 		float in_x2);
 	
+	void scale_array_y_f(int* &table, 
+		int out_x1, 
+		int out_x2,
+		float in_x1,
+		float in_x2,
+		int interpolate);
+
 	OverlayFrame *overlay;
 	ScaleTranslateEngine *scale_translate;
 };

Re: [CinCVS] [Bug 322] New: fit_autos and buttons do something again.

[Pierre Marc Dumuid]

Hi  Nicolas,

- The little up and down arrows located just on the right of the textbox
  
I'm not so sure what to do here, as previously it did some weird 
expansion, in the assumption that the range would be [-10, 10] and 
increment them somewhat.  Considering that you can use the scroll button 
in the textbox to change the min and max values by scrolling on the left 
or right side of the textbox respectively.

- The Alt f shortcut (Fit curve amplitude to highlighted section of curves)
  
I'm a bit wierded by this idea.  What should I do for those autos where 
there is now change?  I'll look into it.


I am on IRC if ya wanna chat

Pierre

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] [Bug 322] New: fit_autos and buttons do something again.

[Pierre Marc Dumuid]

I noticed that problem. I'll use an older revision till the problem
is solved.
  
Is this problem really so bad as to use the old version?  I find that 
rather curious as the range is the much more appropriate range now. I'm 
wanting to get back into study, and not sure when I'll get to fix it.



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

[CinCVS] My hack of the bluedot theme to 2.1 is at..

[Pierre Marc Dumuid]

My hack of the bluedot theme to 2.1 is at..
http://cvs.cinelerra.org/~pmdumuid/cinelerra_bluedot.tar.bz2

It compiles, but is still a bit of cleaning up.
I'm sorry to say that I didn't think it looked that pretty..

Pierre

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] [Bug 200] Fade keyframes lack maximum range

[Pierre Marc Dumuid]

Patch r898 finally fixes the automation curve's range annoyance that has 
probably been bugging people since the 1.2.2 to 2.0 merge.


Notes:
As a result of patch 896 you can scroll through the automation group 
type, making it quicker to select your range you wish to adjust.
It is possible to use your scroll button on either side of the text box 
to adjust the range of the minimum / maximum respectively.

The audio fade steps in 2 dB
The Video Fade steps in 1 (won't go below 0)
The Zoom halves / doubles, (won't go below 0)
The X and Y step in 5 pixels.

Also for each type, I the number of decimal points are displayed as 
appropriate for each type.


Pierre

[EMAIL PROTECTED] wrote:

http://bugs.cinelerra.org/show_bug.cgi?id=200

[EMAIL PROTECTED] changed:

   What|Removed |Added

 Status|NEW |RESOLVED
 Resolution||FIXED



--- Additional Comments From [EMAIL PROTECTED]  2006-09-25 08:31 ---
This bug should be fixed by the r898.  FINALLY!!!



--- You are receiving this mail because: ---
You are the assignee for the bug, or are watching the assignee.

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra
  



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

[CinCVS] bluedot theme for cinelerra 2.1

[Pierre Marc Dumuid]

Hi Koen Muylkens / cinelerra people,

I noticed that you were the one who implemented the blue dot theme for 
cinelerra some time ago.  I have recently attempted to morph your code 
into the 2.1 branch, but have since given up (got better bugs to chase)


If you (or anyone else) is interested in my git repository that contains 
the various commits I made in try to as nicely as possible morph the 
bluedot theme into 2.1 let me know.


Regards,
Pierre Dumuid

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] dv grabbing (off topic)

[Pierre Marc Dumuid]

BTW, that's the same for burning DVD. If you want to be SURE the DVD are
perfectly burn, I have to do nothing except burning on my computer. I
burn at 4x speed only.
  
Huh??  I've burnt 2 DVD disks at once on my computer (using 2 dvd 
burners) and didn't really have any problems..
Given that you can buy tower DVD duplicators that duplicate 3 disks at 
once, I was wondering why linux isn't able to do it either..


but, I guess it shouldn't be recommended..
:)


___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

[CinCVS] Versioning text for cinelerra

[Pierre Marc Dumuid]

Hi All,

I have recently developed a patch that allows the display of the version 
info on startup, and in the preferences to hold info about the 
repository that the source was built from.  I haven't quite finished it 
yet but want to get feedback from package developers.


The main idea is that:
./configure --with-buildinfo-repotype=git 


would cause the terminal output to display:

Cinelerra 2.1CV  GIT::cd8d0e02dd39d612889c8cffe0fd7cc5b95a1418 (C) 2006 
Heroine Virtual Ltd.

Compiled on Sat Sep 23 14:09:21 CST 2006

Cinelerra is free software, covered by the GNU General Public License,



Whilst
./configure --with-buildinfo-repotype=svn

would cause the terminal output to display:

Cinelerra 2.1CV  SVN 896M (C) 2006 Heroine Virtual Ltd.
Compiled on Sat Sep 23 14:43:35 CST 2006

Cinelerra is free software, covered by the GNU General Public License,
...

Using this flag alone, it will only create / update the versioninfo.h 
file if it doesn't exist.  If you want it to be created on each compile, 
then you can pass the flag --with-buildinfo-recompile, which will force 
a re-creation of versioninfo.h on each execution of make, (and result in 
recompiling main.o and aboutprefs.o and re-linking)


Part of the idea with this patch is that package developers could create 
their own versioninfo.h files to specify the distribution, and compile 
specific info etc.


Before I commit it, I wouldn't mind getting feedback on the patch..

Index: configure.in
===
--- configure.in	(revision 892)
+++ configure.in	(working copy)
@@ -228,6 +228,26 @@
 AM_CONDITIONAL(HAVE_FIREWIRE,test "x$enable_firewire" = "xok")
 ## END OF FIREWIRE
 
+# BUILDINFO display, (for displaying version / date)
+AC_ARG_WITH(buildinfo-repotype,
+  AC_HELP_STRING([--with-buildinfo-repotype], [sets the repository type (git or SVN at the moment) for build info]),
+		[ binfo_repotype=$withval ],
+		[ binfo_repotype="none" ])
+AC_SUBST(buildinfo_repotype)
+
+AM_CONDITIONAL(BUILDINFO_GIT, test "x$binfo_repotype" = "xgit")
+AM_CONDITIONAL(BUILDINFO_SVN, test "x$binfo_repotype" = "xsvn")
+
+AC_ARG_WITH(buildinfo-recompile,
+  AC_HELP_STRING([--with-buildinfo-recompile], [recompile the affected objects every instance of make.]),
+		[ binfo_recompile=$withval ],
+		[ binfo_recompile="no" ])
+AC_SUBST(buildinfo_repotype)
+
+AM_CONDITIONAL(BUILDINFO_RECOMPILE, test "x$binfo_recompile" = "xyes")
+
+# END BUILDINFO display, (for displaying version / date)
+
 # CSS SUPPORT IN LIBMPEG3
 if test "x$enable_css" = "xyes"; then
 	CSS_CFLAGS="-DHAVE_CSS"
Index: cinelerra/aboutprefs.C
===
--- cinelerra/aboutprefs.C	(revision 892)
+++ cinelerra/aboutprefs.C	(working copy)
@@ -5,6 +5,7 @@
 #include "quicktime.h"
 #include "theme.h"
 #include "vframe.h"
+#include "versioninfo.h"
 
 
 
@@ -42,10 +43,9 @@
 
 	y += get_text_height(LARGEFONT);
 	char license2[BCTEXTLEN];
-	sprintf(license2, "%s%s%s%s%s", 
+	sprintf(license2, "%s%s%s%s", 
 		_("(C) 2006 Heroine Virtual Ltd.\n\n"),
-		_("SVN Version: "),
-		SVNVERSION,
+		REPOABOUTPREFTXT,
 		_("\nBuild date: "), 
 		BUILDDATE);
 	set_font(MEDIUMFONT);
Index: cinelerra/Makefile.am
===
--- cinelerra/Makefile.am	(revision 893)
+++ cinelerra/Makefile.am	(working copy)
@@ -1,5 +1,32 @@
 BUILDDATE=$(shell date)
+
+if BUILDINFO_RECOMPILE
+main.o: versioninfo
+aboutprefs.o: versioninfo
+
+versioninfo:
+else
+main.o: versioninfo.h
+aboutprefs.o: versioninfo.h
+
+versioninfo.h:
+endif
+	echo '#define BUILDDATE "$(BUILDDATE)"' > versioninfo.h
+
+if BUILDINFO_GIT
+GITVERSION=$(shell if [ -d $(top_srcdir)/.git ]; then cd $(top_srcdir); cg-object-id; else echo "unk"; fi)
+	echo '#define REPOMAINTXT " GIT::$(GITVERSION) (C) 2006 Heroine Virtual Ltd.\nCompiled on $(BUILDDATE)"' >> versioninfo.h
+	echo '#define REPOABOUTPREFTXT "Git SHA-1: $(GITVERSION)"' >> versioninfo.h
+else
+if BUILDINFO_SVN
 SVNVERSION=$(shell svnversion $(top_srcdir))
+	echo '#define REPOMAINTXT  " SVN $(SVNVERSION) (C) 2006 Heroine Virtual Ltd.\nCompiled on $(BUILDDATE)"' >> versioninfo.h
+	echo '#define REPOABOUTPREFTXT "SVN Version: $(GITVERSION)"' >> versioninfo.h
+else #default
+	echo '#define REPOMAINTXT  "(C) 2006 Heroine Virtual Ltd.\nCompiled on $(BUILDDATE)"' >> versioninfo.h
+	echo '#define REPOABOUTPREFTXT ""' >> versioninfo.h
+endif
+endif
 
 SUBDIRS = data
 
@@ -309,9 +336,7 @@
 	$(MJPEG_CFLAGS) \
 	$(OPENEXR_CFLAGS) \
 	$(LIBDV_CFLAGS) \
-	-DPLUGIN_DIR=\"$(plugindir)\" \
-	-DSVNVERSION="\"$(SVNVERSION)\"" \
-	-DBUILDDATE="\"$(BUILDDATE)\""
+	-DPLUGIN_DIR=\"$(plugindir)\"
 
 AM_LDFLAGS = -export-dynamic
 
Index: cinelerra/main.C
===
--- cinelerra/main.C	(revision 892)
+++ cinelerra/main.C	(working copy)
@@ -14,6 +1

Re: [CinCVS] Pipes question ...

[Pierre Marc Dumuid]

Heh,
There are actually "pipe presets" that cinelerra demonstrates  how to 
use it..


Cinelerra actually starts whatever program is specified in the textbox, 
and pipes the YUV data into that program.


Basically, what you place in the textbox next to the "Use Pipe" check 
box is the actual program that receives the pipe)

In regards to the statement, make a bash script that contains:

#!/bin/bash
yuvcorrect -v 0 -T INTERLACED_BOTTOM_FIRST | mpeg2enc -v 0 -r 32 -4 1 
-2 1 -D 10 -g 15 -G 15 -q 5 -b 9400 -f 8 -o $1



calling yencode.sh
(possibly needing a chmod +x on it)

Then place yencode.sh in the textbox.





"Export as an YUV4MPEG stream and use a pipe to encode the YUV stream :
 yuvcorrect -v 0 -T INTERLACED_BOTTOM_FIRST | mpeg2enc -v 0 -r 32 -4 1 
 -2 1 -D 10 -g 15 -G 15 -q 5 -b 9400 -f 8 -o $1 


 Export an AC3 sound stream at 224 kbps

 Mplex everything with :
 mplex -f 8 -V audio.ac3 video.m2v -o movie.mpeg"

And there are many more references to piping - quite justified, as it
prevents *huge* temporary files from being created (the YUV4MPEG stream
is about 1GB per minute of video). Otrherwise, one would need at least
60GB of free HD space at all times for every hour of video rendering.


Alas, there's one thing I somehow don't understand: how exactly do I
pipe the output from Cinelerra to the yuvcorrect line?

I thought I understood pipes (they're equally simple as useful), but I
can't seem to make it work here. I tried the following procedure:

[CLI]:   mkfifo cinepipe
[Cinelerra]: Render to YUV4MPEG Stream - file name: cinepipe
[CLI]:   yuvcorrect ... -o video.m2v < cinepipe

However, Cinelerra says it's busy rendering, but it just sits there and
does nothing; nor does top show any activity in the yuvcorrect/mpeg2enc
department.
I suspect that Cinelerra somehow has trouble with named pipes - if I
replace the last CLI command with cat cinepipe > /dev/null, there's
still no progress in Cinelerra.

I'm pretty certain that the solution is *really* simple - but I'm afraid
that I'm even simpler - too simple in fact to see it. Google turned up
lots of tips about plumbing, and the usual explanations and examples
about the | symbol, but nothing much useful on using pipes to process
(stream) output from GUI applications.


Thanks in advance for any explanation,

Regards,

Richard Rasker


___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

[CinCVS] How this last merge occured

[Pierre Marc Dumuid]

Hi all,

For those interested, I have drawn up the process of how the 2.1 merge 
occurred:


http://cvs.cinelerra.org/~pmdumuid/2.0_to_2.1_notes/20_21merge.png
http://cvs.cinelerra.org/~pmdumuid/2.0_to_2.1_notes/20_21merge.fig

The Magenta line shows connections where commits came from the SVN, the 
red lines show commits as a result of the merge, and the black lines are 
simple mergers.


Here are the steps taken:

  1. Determine which patch from 2.0CV had made it into 2.1, ( see
 http://cvs.cinelerra.org/~pmdumuid/2.0_to_2.1_notes/patch_yesno )
  2. Apply these patches onto the 2.0 base code, (SECTION "2.0CV into 2.1")
  3. Another branch "2.0 to 2.1" then was developed containing
 incremental patches that would arrive at 2.1.
  4. Patches made in the "2.0 to 2.1" branch where then pulled into
 "2.1CV plan" branch using the git merge tool.
  5. Due to some git-issues, the "2.1CV plan" branch required an
 isolation from it's mergers, so a new branch, "2.1CV plan w/out
 merge" was created.
  6. Patches were then collated, and ordered in a better manner and a
 new branch, "2.1CV reordered" is formed.
  7. At multiple points along "2.1CV reordered", the changes are
 committed to the SVN.

At this stage, the relationships between the figure branch names and the 
git branches are:


"2.1 to 2.1 patch"http://.../~j6t/.../cinelerra2.git
"2.1CV plan"  http://.../~j6t/.../cinelerra2.git#svn-plan
"2.1CV plan w/out merge"  http://.../~j6t/.../cinelerra2.git#svn-plan-linear
"2.1CV reordered" http://.../~j6t/.../cinelerra2.git#svn-planj6t

Regards,

Pierre

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

[CinCVS] Re: further to the lock_compressor

[Pierre Marc Dumuid]

hi Johannes (and other's who have pulled the git!),

whilst at work, I thought I'd get the computer to test compile 
individual revisions.  Thought this script may be useful:


--
#!/bin/bash

cd testcompile
# This is a really boring process!!
for i in $(git-rev-list origin ^42ca31c9f); do
# into Dir
cg-reset
cg-clean
if [ -a ../results/$i.done ]; then
   echo Skipping $i, it is already done!
else
   cg-seek $i
   ./autogen.sh
   echo Configuring for $i
   ./configure --prefix=/usr > ../results/$i.config1 2> 
./results/$i.config2

   echo Making for $i
   make > ../results/$i.make1a 2> ../results/$i.make2a
   make > ../results/$i.make1b 2> ../results/$i.make2b
   cat ../results/$i.make2b
   touch ../results/$i.done
   make clean
fi
# out of dir
done
cd ..

--

After modifying the git-rev-list start and end points, let it rip, then 
search in the ../results/*.make2b for bigger than normal file sizes.


The 2 makes are to get rid of the warning messages...(i.e. the first 
make compiles, and warnings end up in make2a


Please note, that this is an EXTREMELY slow method, and the autogen.sh 
&& configure && cleaning of the dir may not REALLY be needed every time!!


Cheers,

Pierre

Johannes Sixt wrote:

Pierre,

I've almost done it - all patches up to 
2e02de0baf1909cd3b7b4ccadaab97e90fe0a536 are moved to svn-plan2. There are 
now only 9 patches remaining, one of which is the recording window patch that 
is already on svn-plan2.


I would have preferred to complete everything, but it's already too late night 
now. I've only commited the cache and garbage collection patches to svn.


BTW, I haven't continued from your svn-linearpmd this time because it looked a 
bit messy, although I did reuse the splitted-off patch related to cache.* - 
thanks for doing this!


-- Hannes



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] A detail about the latest SVN

[Pierre Marc Dumuid]

Thank's for the compile error report... I assume j6t will fix this soon.

If people like test-compiling code, here are some instructions on how 
you could help j6t and myself by doing so.


1. Install git, cogito
2. make an empty directory
3. within this directory, follow these commands:
cg-init
# Type ":wq" in vi to save an initial commit, (which is empty)
# Now what the next lines do is declares remote branches to your git 
repository according to

# cg-branch-add  
cg-branch-add svn-plan-linear 
http://cvs.cinelerra.org/~j6t/cinelerra2.git#svn-plan-linear
cg-branch-add svn-planj6t 
http://cvs.cinelerra.org/~j6t/cinelerra2.git#svn-planj6t
cg-branch-add svn-linearpmd 
http://cvs.cinelerra.org/~pmdumuid/2.0_to_2.1_notes/git.pmd_cin#svn-linearpmd


# After declaring the remote branches, you then need to fetch the latest 
updates for each of these branches:

# (note, these could be bandwidth intensive!!)
cg-fetch svn-plan-linear
cg-fetch svn-planj6t
cg-fetch svn-linearpmd

# Now your master is currently on your empty repository, (so you'll 
likely see no changes in your dir)

# Perform the following:
git-update-ref refs/heads/master 7e637073eb
cg-reset
cg-clean
# Your master is now on commit 7e637073eb (which is r852 for the real SVN)

# Now fire up gitk:
gitk
# and select
# View-> New View...
# Name: svn-plans
# [x] remember this view
# Commits to include: svn-plan-linear svn-planj6t svn-linearpmd

# Now after you've switched to that view, here is what you are seeing in 
git:
# (type the SHA1 ID's into the textbox in gitk to jump quickly to these 
commits)


# fc8df179735c7f1ee58737cdf6078b51184adcdd
#  - this is the last commit in SVN before j6t and I started the merge, 
(it is that patch that people had

# errors compiling!)

# b85916e7118ac4f2163dc1822179d0283ef9b2f5 -> svn-plan-linear
# - this is a branch is the svn-plan-linear branch.  This branch 
contains a heap of commits that were formed

# that should cover all the new features in 2.1
# ( we actually jointly developed these patches on another commit branch at
#  http://cvs.cinelerra.org/~j6t/cinelerra2.git#svn-plan  )

# 93a8e83215efb356539eab47042b1c468ead51ae -> svn-planj6t
# - this branch is a re-ordering (and re-fining) of the commits on 
svn-plan-linear branch

# It is from this branch that we are pushing into the SVN

# As both j6t and myself are on the job, sometimes (in Aussie time) I'll 
look at collating some patches
# together for j6t to review and place in his svn-planj6t branch.  This 
is what the svn-linearpmd branch

# is about.

# of course, these branches are updated on a {bi,tri,dec...}-daily 
basis, so you'll need to

# perform cg-fetch to bring your local git up-to-date with ours...

# o.k. after that tour, here's how to test compile for us:
# you are currently at 7e637073eb,

# perform your standard:  ./autogen; ./configure; make; make install

# hopefully this should work!
# if it does, then you can test another commit for us, say 2cc60c1, to 
try this commit out:


./cms_movelocalmaster 2cc60c10aba  # this updates the master reference only!
cg-reset   # this actually sync's the files in 
the repo

make; make install

# and so forth. Note, that if you go onto the svn-plan-linear branch, 
instead of typing:


# ./cms_movelocalmaster ...
# you should type
# git-update-ref refs/heads/master ...

# if you manage to run-across a commit that is bad, you could email us 
and let us know which commit doesn't compile for you...


# that way we don't commit non-compiling commits to SVN :) !

Note:
The above instructions were written without testing, (I didn't want to 
chew my bandwidth just for a demo!)

the above steps could possibly need refining!

Oh, HV, if you're listening in, there's a branch especially for you:

cg-branch-add pmd-ee 
git+ssh://cvs.cinelerra.org/home/pmdumuid/cinelerra.git#embraceextend


This branch contains a series of commits that you might want to get.
(I'll also assume you didn't read this whole email, and will send you a 
private email later)


Cheers,
Pierre

Bas Alphenaar wrote:

How did you manage to compile r845? I get this error:


if g++ -DHAVE_CONFIG_H -I. -I. -I.. -I../quicktime -I../guicast 
-I../libmpeg3 -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE 
-D_FILE_OFFSET_BITS=64  -DHAVE_ALSA -DHAVE_OSS 
-I/usr/include/mjpegtools -I/usr/include/mjpegtools/mpeg2enc 
-I/usr/include/mjpegtools/mplex   -I/usr/include/OpenEXR 
-DPLUGIN_DIR=\"/usr/local/lib/cinelerra\" -g -O2 -MT videodevice.o -MD 
-MP -MF ".deps/videodevice.Tpo" \
  -c -o videodevice.o `test -f 'videodevice.C' || echo 
'./'`videodevice.C; \

then mv -f ".deps/videodevice.Tpo" ".deps/videodevice.Po"; \
else rm -f ".deps/videodevice.Tpo"; exit 1; \
fi
videodevice.C: In constructor ‘VideoDevice::VideoDevice(MWindow*)’:
videodevice.C:110: error: invalid use of undefined type ‘struct MWindow’
mwindow.inc:16: error: forward declaration of ‘struct MWindow’
make[3]: *** [videodevice.o] Error 1
make[

Re: [CinCVS] Bezier automation

[Pierre Marc Dumuid]

The merge is now a joint effort, (j6t and myself) and we are VERY 
close.  We've made the patch, and are just re-ordering, and cleaning 
them up.


[EMAIL PROTECTED] wrote:

Hannes,
I've been out of the loop for awhile..is the 2.1 merge complete?
scott
 -- Original message --
From: Johannes Sixt <[EMAIL PROTECTED]>
  
2.1 has a menu entry "Straighten keyframes". I haven't played with it yet, but 
from the naming I'd conclude that it is a workaround that allows to have 
straight line segments instead of bezier interpolation between keyframes.


-- Hannes

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra





___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra
  



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] Will not compile - Open SUSE 10.1

[Pierre Marc Dumuid]

Looking at the href you gave, I agree that your patch seems alright.
Can't really apply it to svn until merge is complete...
Pierre

Alexis Ballier wrote:

Yes I think the best solution is to ask x264 devs.
Before writing this patch I had a look at how it has been fixed in 
other softs, and for the ffmpeg part included in cinelerra it's how it 
has been fixed by ffmpeg devs. The part I had to think more about is 
the quicktime part, but it should be similar to the fixes I've seen.
You might also want to have a look at [1] to see how the api changed, 
and I agree that's strange to replace cbr by abr...





[1] https://trac.videolan.org/x264/changeset/537 





--
Alexis Ballier. 



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] Will not compile - Open SUSE 10.1

[Pierre Marc Dumuid]

I don't know much about the x264 stuff, but was thinking of forwarding 
this patch on to the x264 mailing list to ask if the changes look about 
right? (I am sure they'd know the best...)


Looking at the patch, I can guess that b_cbr means "?? constant 
bit-rate" and X264_RC_ABR means "?? average bit rate",


Just a quick (possibly dumb) question:  Aren't constant and average 
might be different??  (i.e. constant would be constant frame size, 
whilst average, would be varying frame size, but average over the whole 
file)


Alexis Ballier wrote:

Hi,
I've had exactly the same problem and wrote a patch to fix this. It 
seems to work, but I think more tests are needed.


I'm attaching the patch (against svn of trunk/hvirtual rev 840).

Regards,

--
Alexis Ballier.


--- cinelerra-cvs-20060827/quicktime/qth264.c   2006-08-27 00:32:07.130566427 
+0200
+++ working/quicktime/qth264.c  2006-08-27 03:00:58.486699204 +0200
@@ -136,7 +136,11 @@
x264_param_t default_params;
x264_param_default(&default_params);
 // Reset quantizer if fixed bitrate
+#if X264_BUILD < 48
if(codec->param.rc.b_cbr)
+#else
+   if(codec->param.rc.i_rc_method == X264_RC_ABR )
+#endif
{
codec->param.rc.i_qp_constant = 
default_params.rc.i_qp_constant;
codec->param.rc.i_qp_min = default_params.rc.i_qp_min;
@@ -469,7 +473,11 @@
}
else
if(!strcasecmp(key, "h264_fix_bitrate"))
+#if X264_BUILD < 48
codec->param.rc.b_cbr = (*(int*)value) / 1000;
+#else
+   codec->param.rc.i_bitrate = (*(int*)value) / 1000;
+#endif
}
 }
 
--- cinelerra-cvs-20060827/quicktime/ffmpeg/libavcodec/x264.c	2006-08-27 00:32:00.564013252 +0200

+++ working/quicktime/ffmpeg/libavcodec/x264.c  2006-08-27 02:47:49.224599225 
+0200
@@ -135,8 +135,13 @@
 x4->params.i_keyint_max = avctx->gop_size;
 x4->params.rc.i_bitrate = avctx->bit_rate / 1000;
 x4->params.rc.i_vbv_buffer_size = avctx->rc_buffer_size / 1000;
-if(avctx->rc_buffer_size)
+if(avctx->rc_buffer_size){
+#if X264_BUILD < 48
 x4->params.rc.b_cbr = 1;
+#else
+   x4->params.rc.i_rc_method = X264_RC_ABR;
+#endif
+}
 x4->params.i_bframe = avctx->max_b_frames;
 x4->params.b_cabac = avctx->coder_type == FF_CODER_TYPE_AC;
 
  



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] Cinelerra on OSX - new patch

[Pierre Marc Dumuid]

Adrian Prantl wrote:

Hi everybody!

Just in case anybody is interested in running Cinelerra on MacOS X:
I just updated the old patch to work with the current SVN version.

This is the first raw version and it still needs some cleanup, but I 
surely would like to see it commited to svn one day!

Thank you, and I will do once the merge is over..
I'm rather interested to see it running on my Friends mac.   Does it 
work on both x86 and PPC?? (which are you using?)


Pierre

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] feature request

[Pierre Marc Dumuid]

Modesto Alexandre wrote:

Hello,

Feature request :

* Automated backup of project (every xx minutes)

=> documentation : how to remote backup .xml files :)

Alex.
  
bugzilla is the general place for feature requests. 


___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] Contributing to development

[Pierre Marc Dumuid]

Oops, that was meant for you personally!

Pierre Dumuid wrote:

Dean,

If you have a box on the net that you can give me an account to use 
ssh on, I could probably get you to help me with helping j6t to do the 
merge.
Basically, It'd involve a small amount of learning about git, (a 
wonderful program that linus created, that I have only learnt in the 
last few weeks)


Pierre

Dean Edmonds wrote:

Thanx for the info, Pierre.  I guess I'll just have to sit on my hands
until the merge is finished.

=== 

  - deane  Gooroos Software: Plugging you 
into Maya


  Visit http://www.gooroos.com for more 
information


___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra
  






___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] Free media positioning in track?

[Pierre Marc Dumuid]

[EMAIL PROTECTED] wrote:

On Fri, 21 Jul 2006, Stefan de Konink wrote:
  

Is it possible to move a track around in the 'arrow' mode instead with the
mouse moving the track with the keyboard frame by frame.

I want to lipsync individual recordings. But it is kinda hard doing it
with a mouse, play to check, and move it a bit around again.



What I'd really like to see would be keystrokes for "move this edit a
frame earlier" and "...a frame later" that would work during playback,
from the compositor.  Then I could modify timing for lip sync, and play
and replay to see that it works, all from the keyboard without switching
windows.
  

Try pressing the keys on the Num Pad and see what they do (1 & 4 especially)
Also, doc/cinelerra.html actually contains some shortcut keys, as well 
as when you hover your mouse over some of the buttons around the place, 
(see that bar just below the menu bar) you'll see what keys correspond 
to which button.




___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] updated of wiki of cinelerra to add info about the community version of cinelerra..

[Pierre Marc Dumuid]

mark stavar wrote:
Well, a good deal better than the "about.php dribble" IMHO.   I 
finally have a better understanding of why there are two distinct 
forks of the code. 


Well done and thanks,

marks

thanks,  though the correctness of grammar I'd have to attribute to 
other authors.  (look at the history) I was quite surprised at how 
quickly grammar and wiki-correct people got onto what I wrote, (check 
the history, I am from 129.127.??, all the rest are the editors).  Makes 
me think I should a few of chapters of my PhD thesis and let them fix it 
for me :).


Pierre

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

[CinCVS] svn_log.php vs. Changelog

[Pierre Marc Dumuid]

Hi All,

Given someone on IRC indicated that the website links were quite broken, 
I thought I'd take a look.  In looking, I noticed the Changelog menu 
item.  Given that we now have the svn_log.php page, I was wondering if 
people find the Changelog useful anymore?  I personally find it quite 
redundant.  So I am curious if anyone if for / against the idea of 
removing the ChangeLog?


Pierre


___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] Cinelerra dialogs with translated texts

[Pierre Marc Dumuid]

Jean-Luc,

Just call me Pierre, (not Pierre-Marc)

I have been using the widget-grid system at home for quite a while now, 
and haven't had it crash on me much.


The widget grid "engine" is actually already in SVN.  To use the engine 
with the dialog box, code needs to be added to every declaration of a 
widget that uses the engine. (basically the widget are added to a grid, 
and then the grid engine cycles through each column, and row, finding 
the max width / height and then calls the repositioning function to move 
each widget accordingly.  rowspan and colspan are also implemented 
similar to html)


Obviously, "adding every widget" to the grid makes the code quite 
different, and can make it really confusing when doing a merge with 
upstream.


Having done (and doing) some merging with j6t, I have a better 
understanding of branching tools and will look into the idea of making a 
branch in SVN that contains widget-grid enabled dialog boxes, and 
ensuring that it is kept somewhat up with the main SVN branch. 

I will send a patch of the latest version of BC_widgetgrid ( against 2.1 
) upstream in hope that HV consider it worth of including in his next 
version, but until we see the next version we won't know if he's taken 
it upstream.


Regards,
Pierre Dumuid

Jean-Luc Coulon (f5ibh) wrote:

Hi,

Some time ago, I had a discussion with Pierre-Marc and submitted a
question to the list about the way to have some kind of dynamic
resizing of the panels to that the translated text could fit in the
room left by the theme.

Is there any nexs on this front?

Best regards

Jean-Luc


2005/6/9, Pierre Marc Dumuid <[EMAIL PROTECTED]>:

Hi All,

Jean-Luc Coulon asked me about possibly developing a patch to resolve
the problem shown between languages in:

http://jean.luc.coulon.free.fr/images/format-genuine.png
http://jean.luc.coulon.free.fr/images/format-modified.png

My proposition is as follows, (and I wouldn't mind a bit of feedback on
the idea!):

Please note that I've got a few things on, so couldn't really clean the
idea up for a few weeks... (Also need to get around to fixing a little
bug that cinelerra has when going to Settings->Format, the "interlace
mode" shown isn't always the interlace mode set for that project, ..
then I need to get a patch of the interlace fixups applied to Heroine's
source)

Here's A ROUGH SKETCH what I  would propose:

void BlondTheme::new_bg_get_positions(int *width, int *height, int *x1,
int *x2, int *x3, int *x4) {
  char *s;
   s = getenv("LANG");
   if (strstr(s,"en_")) {
   width = 500;
   height = 700;
   x1 = 20;
   x2 = 60;
   x3 = 120;
   x4 = 180;
   } else if (strstr(s,"fr_")) {
   width = 600;
   height = 700;
   x1 = 20;
   x2 = 70;
   x3 = 140;
   x4 = 190;
  } else {
   printf("Lang Test -> Don't know that language, will use english
\n");
   width = 600;
   height = 700;
   x1 = 20;
   x2 = 70;
   x3 = 140;
   x4 = 190;
  }
}

would sit in defaulttheme.C

Then in new.C, in void NewThread::run() at the top, place:

   int width, int height, int x1, int x2, int x3, int x4 ;
mwindow->theme->new_bg_get_positions(width, height, x1, x2, x3, x4);

and change:
   int x = mwindow->gui->get_root_w(0, 1) / 2 - WIDTH / 2;
   int y = mwindow->gui->get_root_h(1) / 2 - HEIGHT / 2;

to

   int x = mwindow->gui->get_root_w(0, 1) / 2 - width / 2;
   int y = mwindow->gui->get_root_h(1) / 2 - height / 2;

and also,

   nwindow = new NewWindow(mwindow, this, x, y);
to
   nwindow = new NewWindow(mwindow, this, x, y,width,height);

and
NewWindow::NewWindow(MWindow *mwindow, NewThread *new_thread, int x, 
int y)

: BC_Window(_(PROGRAM_NAME ": New Project"),
x,
   y,
   WIDTH,
   HEIGHT,
   -1,
   -1,
   0,
   0,
   1)

to

NewWindow::NewWindow(MWindow *mwindow, NewThread *new_thread, int x, int
y, int width, int height)
: BC_Window(_(PROGRAM_NAME ": New Project"),
x,
   y,
   width,
   height,
   -1,
   -1,
   0,
   0,
   1)

and in int NewWindow::create_objects(), call

   int width, int height, int x1, int x2, int x3, int x4 ;
mwindow->theme->new_bg_get_positions(width, height, x1, x2, x3, x4);

again, and ensure that the x1,x2,x3,x4 refer correctly..

That type of patch would allow the text authors to have good access to
tuning the column positions of their language.  In terms of fixing the
background, I would propose modifying:

void BlondTheme::draw_new_bg(NewWindow *gui)
{
   int width, int height, int x1, int x2, int x3, int x4 , int x5, x6;
   new_bg_get_positions(width, height, x1, x2, x3, x4,x5,x6);
   gui->draw_vframe(new_bg_topredbox, 0, 0);
   gui->draw_vframe(new_bg_redcurve, x5, 0);
   gui->draw_vframe(new_bg_

Re: [CinCVS] faad issue - need help with make

[Pierre Marc Dumuid]

Hey Tom,

that was quite a few posts!  Anyhow, I assume you installed the faad2 
rpm rather trying to compile it yourself.. and so the error below isn't 
an issue..
lengthy discussions with people may be easier in the IRC channel, 
#cinelerra (use xchat)


Regards,
Pierre

tom poe wrote:

[EMAIL PROTECTED] faad2]# make
Makefile:623: *** missing separator.  Stop.

Pierre:  Took a look, but couldn't see what is missing
Tom




___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] How many audio tracks can i add?

[Pierre Marc Dumuid]

I think this is changed in 2.1

Bernhard Frühmesser wrote:

Hello,

I have one Video track and 7 audio tracks in my project, which all 
work fine, but i need to add more audio tracks to the project, but if 
i add for example another one or more audio tracks etc... they don´t 
play anymore. Is cinelerra limitted to 7 audio tracks.


Not all play at the same time, most of them are used for short audio 
(wav) clips.


Thanks,
BF.


___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] libavc1394 questions; now configure issues

[Pierre Marc Dumuid]

Hi Tom,

You wrote:

Pierre:  the ./configure command you suggest results with same missing 
items.  After adding livna, and installing, then running the suggested 
./configure command, the results still show libtheora missing, as well 
as all the others.  I wonder if I need to have some way of pointing to 
the right directory for each of the missing items?  How best to proceed? 
Tom



I find it wierd that it is the "Same" missing items, (I assume there 
would be LESS missing items!! - please double check!)
I did say it was a "starter" of a list of packages you would likely 
need, and that there would be more.  Looking at the list, obviously I 
missed out the libtheora or libtheora-devel or whatever it is called.  
For each package that is still listed as missing, I advise doing this:


$ yum list '*theora*'

if you then install the package and it still shows up as being missing 
then let me know of that particular package name and I will investigate.


For theora as an example. The above command will list the name of the 
packages available from all the yum repositories set up that would 
satisfy this dependency.  Once you have the name of the package, then 
simply install it:

$ yum install libtheora  libtheora-devel

If you are really lazy, like me, you can do stuff like:
$ yum install '*theora*'

which will automatically list all packages with the words theora in them 
and ask you if you want to download and install them all.


Once you have satisfied these dependencies, please send me a list of all 
the required packages, (including the list I already sent you.)


Cheers,
Pierre



and then yum install for each of the following:

nasm
glibc-kernheaders# for /usr/include/linux/videodev2.h
xorg-x11-proto-devel # I am guessing this is needed
xvidcore-devel   # I am guessing this is needed
libXxf86vm-devel
alsa-lib-devel
mjpegtools-libs  # Naughty!! they don't follow the -devel 
notation!!

libjpeg-devel
fftw-devel
djbfft
e2fsprogs-devel # for libuuid
libdv-devel
libpng-devel
freetype-devel
a52dec-devel
libsndfile-devel
OpenEXR
faac-devel
faad2-devel
libraw1394-devel
libiec61883-devel
libavc1394-devel
librom1394-devel
libogg-devel
libvorbis-devel
lame-devel

(you could place them all on one line!) And then manually install:

x264 - I manually installed this from their website..

This should satisfy many of your dependancy problems... Let me know 
what extra needed, and I'll cut and paste it into the compile 
instructions on our website.


j6t: "xorg-x11-proto-devel" could be what needs to be added to:
AC_MSG_WARN([Please install xlib-dev or xorg-x11-devel.])


since I couldn't find either xlib-dev or xorg-xll-devel on my system..


By the way,
should I be doing something between running configure?  I saw some
install instructions that recommend doing a make clean sometimes.
Tom
Umm, I wouldn't do it too often.  If the build fails because a file 
missing in "faad.h" (i.e. you don't have the devel package installed 
for faad) or something like that, I usually install the devel 
package and try to make again... It should compile.  If it gets to 
the end and complains when linking, then I do a make clean.


In terms of the configure, I personally use:
./configure --prefix=/usr --enable-x86 --enable-mmx 
--enable-freetype2 \
   CFLAGS='-O3 -pipe -fomit-frame-pointer -funroll-all-loops 
-falign-loops=2 -falign-jumps=2 -falign-functions=2 -ffast-math 
-march=pentium-m -mfpmath=sse,387 -mmmx -msse'


when compiling, (I actually placed this in a file 
~/bin/cinelerra_conf.sh cause I got sick of typeing it so much!!) to 
ensure I get a version optimised for my cpu (pentium-m).  You may 
want to investigate this at a later date.


Also, don't forget, the latest svn is crashy, (since it's undergoing 
a merge with the new upstream source)  Use the 836 for best luck 
compiling without any error messages!





___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] libavc1394 questions

[Pierre Marc Dumuid]

Pierre:  Found it, and installed, reran configure and got back a dozen
dependencies missing.  Looks like FC5 didn't want to tackle the
multimedia, and leave that to a group to develop a fedora multimedia
niche distribution, or something.
I personally use FC5.. it does suck a bit, but I am fairly used to it 
now... FC5 is quite scared of patent issues involved with many 
multi-media apps, (lame, mpeg, etc..)


I personally use the livna yum repository, and most of the dependencies 
I was able to be fixed from there..

Well, fooey on that.  I shall use this as a learning experience, and see
if I can get a clean ./configure in the next few days.  
If you decided to do the yum method, can you keep track of what packages 
you installed, and send me a list.  I'll add them to a list somewhere 
later...


Just quickly going through the configure.in script myself, and 
rpm-querying my system to see the package names, I obtained the 
following steps.. You might start with


rpm -ivh http://rpm.livna.org/livna-release-5.rpm

and then yum install for each of the following:

nasm
glibc-kernheaders# for /usr/include/linux/videodev2.h
xorg-x11-proto-devel # I am guessing this is needed
xvidcore-devel   # I am guessing this is needed
libXxf86vm-devel
alsa-lib-devel
mjpegtools-libs  # Naughty!! they don't follow the -devel notation!!
libjpeg-devel
fftw-devel
djbfft
e2fsprogs-devel # for libuuid
libdv-devel
libpng-devel
freetype-devel
a52dec-devel
libsndfile-devel
OpenEXR
faac-devel
faad2-devel
libraw1394-devel
libiec61883-devel
libavc1394-devel
librom1394-devel
libogg-devel
libvorbis-devel
lame-devel

(you could place them all on one line!) And then manually install:

x264 - I manually installed this from their website..

This should satisfy many of your dependancy problems... Let me know what extra 
needed, and I'll cut and paste it into the compile instructions on our website.

j6t: "xorg-x11-proto-devel" could be what needs to be added to:
AC_MSG_WARN([Please install xlib-dev or xorg-x11-devel.])

since I couldn't find either xlib-dev or xorg-xll-devel on my system..


By the way,
should I be doing something between running configure?  I saw some
install instructions that recommend doing a make clean sometimes.
Tom
Umm, I wouldn't do it too often.  If the build fails because a file 
missing in "faad.h" (i.e. you don't have the devel package installed for 
faad) or something like that, I usually install the devel package and 
try to make again... It should compile.  If it gets to the end and 
complains when linking, then I do a make clean.


In terms of the configure, I personally use:
./configure --prefix=/usr --enable-x86 --enable-mmx --enable-freetype2 \
   CFLAGS='-O3 -pipe -fomit-frame-pointer -funroll-all-loops 
-falign-loops=2 -falign-jumps=2 -falign-functions=2 -ffast-math 
-march=pentium-m -mfpmath=sse,387 -mmmx -msse'


when compiling, (I actually placed this in a file 
~/bin/cinelerra_conf.sh cause I got sick of typeing it so much!!) to 
ensure I get a version optimised for my cpu (pentium-m).  You may want 
to investigate this at a later date.


Also, don't forget, the latest svn is crashy, (since it's undergoing a 
merge with the new upstream source)  Use the 836 for best luck compiling 
without any error messages!


Pierre


___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] Definite MPEG Sync Problem in CinelerraCVS (Is there a bug list I can peruse?)

[Pierre Marc Dumuid]

I have uploaded the file I had issues with to:

http://cvs.cinelerra.org/~pmdumuid/pmdumuid_mpeg_extraframe_audiolength.mpg

Pierre

Pierre Marc Dumuid wrote:
It is a known bug, (i.e. some other people know about it) but it 
hasn't been reported in the bug report, (would you be willing to write 
one up!).  I personally would contribute to a bounty for the fix...


I've had issues with mpeg from a standalone DVD-recorder (Panasonic).  
My issues were:

* Audio was always 3 seconds longer than actually in the file
* Video always had 1 frame extra

Other's who have noted the issue include:
Minmax had an issue with a certain video: 
http://www.kiberpipa.org/~minmax/MOV00977.MPG 
<http://www.kiberpipa.org/%7Eminmax/MOV00977.MPG>
goldfish2001 had an issue with: 
www.nicholaslidster.plus.com/video/testvid.mpg


I personally fixed the first issue that I had with a patch ( 
http://cvs.cinelerra.org/~pmdumuid/hvirtual_mpeg_number_of_samples.diff 
) but I'm not completely sure that's its safe for everyone to use. The 
second issue has always evaded me.  Since the MPEG specifications need 
to be purchased, I personally don't understand the structure of mpeg 
files, and hence how to debug libmpeg.


Here's a mail I sent out:
https://init.linpro.no/pipermail/skolelinux.no/cinelerra/2005-December/004880.html 



For a discussion that occured on IRC, see:

http://cvs.cinelerra.org/irclog/index.php?file=2005-12-23.html  
(specifically 13:44 onwards)
http://cvs.cinelerra.org/irclog/index.php?file=2005-12-21.html  
(goldfish2001's posting)


Cheers,
Pierre

Computer Lists wrote:
Over the past week I've worked on getting all of my systems up to the 
latest version of Cinelerra from cvs.cinelerra.org.  I am currently 
running it on the following systems:


1. A Celeron (P4 family) 2.4 GHz box with 512 Megs or RAM running 
Fedora Core 3

2. A Pentium 4 (HT) 2.6 GHz box with 1 gig of RAM running Gentoo 2005.1
3. A Pentium M 1.7 GHz latop with 512 Megs or RAM running Fedora Core 3

With every MPEG (1 or 2) file that I open the video is always shorter 
than the audio track.  Whether it's a capture done with my PVR250 
board, a video created on my Sony Mavica digital camera or a 
transcoded file created with ffmpeg, the problem can be easily 
reproduced.  All I have to do is open the file and drag it onto the 
timeline view, then magnify at the end of the tracks and the audio 
always last many minutes longer than the video.  Since I'm new here 
I'm not sure if this is the right place to post things like this.  If 
not, let me know where I should post.  Also, if there is a resource 
that I can use to see if there is already a bug report for this issue 
and it's status, please let me know.  I don't want to annoy anyone, 
but I want to get some answers and hopefully participate in solving 
some problems. :)


Thanks,
Deck

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra

Re: [CinCVS] Definite MPEG Sync Problem in CinelerraCVS (Is there a bug list I can peruse?)

[Pierre Marc Dumuid]

It is a known bug, (i.e. some other people know about it) but it hasn't 
been reported in the bug report, (would you be willing to write one 
up!).  I personally would contribute to a bounty for the fix...


I've had issues with mpeg from a standalone DVD-recorder (Panasonic).  
My issues were:

* Audio was always 3 seconds longer than actually in the file
* Video always had 1 frame extra

Other's who have noted the issue include:
Minmax had an issue with a certain video: 
http://www.kiberpipa.org/~minmax/MOV00977.MPG 

goldfish2001 had an issue with: 
www.nicholaslidster.plus.com/video/testvid.mpg


I personally fixed the first issue that I had with a patch ( 
http://cvs.cinelerra.org/~pmdumuid/hvirtual_mpeg_number_of_samples.diff 
) but I'm not completely sure that's its safe for everyone to use. The 
second issue has always evaded me.  Since the MPEG specifications need 
to be purchased, I personally don't understand the structure of mpeg 
files, and hence how to debug libmpeg.


Here's a mail I sent out:
https://init.linpro.no/pipermail/skolelinux.no/cinelerra/2005-December/004880.html

For a discussion that occured on IRC, see:

http://cvs.cinelerra.org/irclog/index.php?file=2005-12-23.html  
(specifically 13:44 onwards)
http://cvs.cinelerra.org/irclog/index.php?file=2005-12-21.html  
(goldfish2001's posting)


Cheers,
Pierre

Computer Lists wrote:
Over the past week I've worked on getting all of my systems up to the 
latest version of Cinelerra from cvs.cinelerra.org.  I am currently 
running it on the following systems:


1. A Celeron (P4 family) 2.4 GHz box with 512 Megs or RAM running 
Fedora Core 3

2. A Pentium 4 (HT) 2.6 GHz box with 1 gig of RAM running Gentoo 2005.1
3. A Pentium M 1.7 GHz latop with 512 Megs or RAM running Fedora Core 3

With every MPEG (1 or 2) file that I open the video is always shorter 
than the audio track.  Whether it's a capture done with my PVR250 
board, a video created on my Sony Mavica digital camera or a 
transcoded file created with ffmpeg, the problem can be easily 
reproduced.  All I have to do is open the file and drag it onto the 
timeline view, then magnify at the end of the tracks and the audio 
always last many minutes longer than the video.  Since I'm new here 
I'm not sure if this is the right place to post things like this.  If 
not, let me know where I should post.  Also, if there is a resource 
that I can use to see if there is already a bug report for this issue 
and it's status, please let me know.  I don't want to annoy anyone, 
but I want to get some answers and hopefully participate in solving 
some problems. :)


Thanks,
Deck

___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra



___
Cinelerra mailing list
Cinelerra@skolelinux.no
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra