Re: [Bf-committers] ffmpeg library update
Hi, I've done some testing with configurations. Here are some results. So, first of all i've disabled version3 and nonfree otions. First of them leads to gpl3 or higher library version, second maked libraries un-redistributable. Now libraries are gpl2 or higher. Also, i've disabled debugging symbols (--disable-debug flag). This shouldn't affect on size doe to all symbols (strip --strip-all) are stripping from blender binary. Then i've made simple config which gave exactly the same result in dependencies as ffmpeg which was used for 2.57a release. Conficuration is published in [1]. Blender binary after stripping was 43Mb. Don't think it's matetr due to linking happened against libraries which are ~1year newer, i think, and codebase of ffmpeg became larger (additional checkings, maybe architecture changed so some symbols which were unused and stripped now used and so on). Next test was with initial config from this thread with license cleaning and cleaning up flags which Peter and Joerg marked as unused and also disabled most of marked as UNSURE options. This options are related on speech (don't think this kind of audio is useful for Blender, but this libraries aren't big -- ~300Kb so could be added if they're necessery) and streaming (but dirac and schro keeped enabled -- it's algos of encoding/decoding raw videos). Configuration is published in [2]. Blender binary growed u[ to 49mb. This is also understandable, because all new dependencies (vpx, ogg, vorbis, theora, dirac, schro) are ~8mb in total. More codecs -- higher size of binary. Next step was disabled schro and dirac. Configuration posted to [3]. This saved 2mb and Blender binary became 47mb. I haven't tested with rtmp enabled because don't find stream sources b useful in Blender and this gives too much dependencies which weren't easy to solve -- some libraries came from Debian Sid environment, some from Debian Lenny, some from Lenny Backports.. Also, this adds ~2mb of Blender binary size. So, question is: do we want to support more codecs (ogg, vorbis, teora, vpx, dirac, schro) and have 8mb bigger Blender binary as we've got now? [1] http://www.pasteall.org/21098 [2] http://www.pasteall.org/21099 [3] http://www.pasteall.org/21100 -- With best regards, Sergey I. Sharybin ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
Re: [Bf-committers] ffmpeg library update
Hi, maybe this is not the moment or not the good thread but i have a problem to import correctly pal avi video in the blender 2.57 sequencer with the 25 fps framerate. In this case, the sound is too short and of course the video in blender is unsynchronized. But if i select a custom framerate at 24 fps divided by the proportion of sound-time on the image-time, 0.96, it gives me a quite correct synchronization (one image lag in the sound time). Curiously, it displays a framerate of 25 fps. All this seems a bit absurd and silly but true. Of course, I can post all this on the bug tracker with screen captures. Regards, jms. ps : the video is a sony digital handicam one imported through win 7. ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
Re: [Bf-committers] ffmpeg library update
Also, forgot to mention. Pre-rc1 binaries were 52mb. Not sure why with the same configuration Ken had got such bigger binary -- maybe it's because of collada (not sure i've used exactly the same configuration for it). Really strange thing that even with binary 10mb bigger tham mine Ken's archive was 2-3mb smaller. Matbe it's bytes sequence issue. So, maybe enabling all that codecs aren't so big issue from binary side point of view? Sergey I. Sharybin wrote: Hi, I've done some testing with configurations. Here are some results. So, first of all i've disabled version3 and nonfree otions. First of them leads to gpl3 or higher library version, second maked libraries un-redistributable. Now libraries are gpl2 or higher. Also, i've disabled debugging symbols (--disable-debug flag). This shouldn't affect on size doe to all symbols (strip --strip-all) are stripping from blender binary. Then i've made simple config which gave exactly the same result in dependencies as ffmpeg which was used for 2.57a release. Conficuration is published in [1]. Blender binary after stripping was 43Mb. Don't think it's matetr due to linking happened against libraries which are ~1year newer, i think, and codebase of ffmpeg became larger (additional checkings, maybe architecture changed so some symbols which were unused and stripped now used and so on). Next test was with initial config from this thread with license cleaning and cleaning up flags which Peter and Joerg marked as unused and also disabled most of marked as UNSURE options. This options are related on speech (don't think this kind of audio is useful for Blender, but this libraries aren't big -- ~300Kb so could be added if they're necessery) and streaming (but dirac and schro keeped enabled -- it's algos of encoding/decoding raw videos). Configuration is published in [2]. Blender binary growed u[ to 49mb. This is also understandable, because all new dependencies (vpx, ogg, vorbis, theora, dirac, schro) are ~8mb in total. More codecs -- higher size of binary. Next step was disabled schro and dirac. Configuration posted to [3]. This saved 2mb and Blender binary became 47mb. I haven't tested with rtmp enabled because don't find stream sources b useful in Blender and this gives too much dependencies which weren't easy to solve -- some libraries came from Debian Sid environment, some from Debian Lenny, some from Lenny Backports.. Also, this adds ~2mb of Blender binary size. So, question is: do we want to support more codecs (ogg, vorbis, teora, vpx, dirac, schro) and have 8mb bigger Blender binary as we've got now? [1] http://www.pasteall.org/21098 [2] http://www.pasteall.org/21099 [3] http://www.pasteall.org/21100 -- With best regards, Sergey I. Sharybin -- With best regards, Sergey I. Sharybin ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
Re: [Bf-committers] Camera view navigation rant
I'm talking about this for long time now but I always get this beautiful answer: We don't need that because we have fly mode. And if we really want to align camera to view, we have Ctrl+0. If you really need that (although we don't know why would you need such a thing), Blender is opensource so make a patch by your self. Buzz off. Thumbs up Daniel! On 25 April 2011 05:50, Campbell Barton ideasma...@gmail.com wrote: On Sun, Apr 24, 2011 at 9:55 PM, Daniel Salazar - 3Developer.com zan...@gmail.com wrote: So can we take a look at camera view navigation? It's like playing a space flight simulator every time I press Shift F, I just want to position and orient my camera, not to deal with inertia and atmosphere entering angle! (feels like that at least) http://chrome.atari.com/lunarlander/ The flight mode panning (WASD keys) is of no use.. again calculating directions and inercia and keeping in mind how many presses I've done for each side so I can try and counter act when I want a stop! (which never works btw, the counter act seems to never be able to reach zero again) this could be a test for new NASA astronauts. Trackball is a decent tool for orienting the camera but it is about 10 times too sensitive, and somehow seems like it's going the wrong direction (inverted axis?) I have asked for simple game like controls in the past, you know those first person shooter controls? those are intuitive and precise but the answer was something like naah fly simulator is cooler, we can fly like birds!, really that was an actual answer, supposedly for live recording the camera to keyframes. Also a lock camera to view option would be great so we can just move the camera like we move any other viewport am I the only one that doesn't get this? peace, Daniel Salazar 3Developer.com Alternative view navigation modes would definitely be worth trying out and can be scripted without doing anything really complicated. See: Script Templates - Operator Modal View3D, this is a very simple operator that uses the mouse to move the view offset. in execute() you can set rv3d.view_location and rv3d.view_rotation API Ref: http://www.blender.org/documentation/blender_python_api_2_57_release/bpy.types.RegionView3D.html Reading you're mail I remembered rv3d.view_matrix was readonly, just committed support for assignment r36317. This makes manipulating the view easier, so you can assign the transform without worrying about the 'view_distance'. Of course once the basics are done this can become more involved, casting rays, smoother motion, mouse-look strafe :-) ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
Re: [Bf-committers] ffmpeg library update
Am 2011-04-25 08:26, schrieb Sergey I. Sharybin: So, question is: do we want to support more codecs (ogg, vorbis, teora, vpx, dirac, schro) and have 8mb bigger Blender binary as we've got now? Ogg, vorbis and theora are definitely needed as they are in the output file formats and codecs and commonly used codecs in the FOSS world. For the others I'm fine if we don't support them, but I'm not the video guy here, just audio. Regards ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
Re: [Bf-committers] ffmpeg library update
Made full release build cycle (with blenderplayer, strippng, packing and so on). Archive size with all that codecs enabled growed up from 30 to 33mb for linux64 platforms Sergey I. Sharybin wrote: Also, forgot to mention. Pre-rc1 binaries were 52mb. Not sure why with the same configuration Ken had got such bigger binary -- maybe it's because of collada (not sure i've used exactly the same configuration for it). Really strange thing that even with binary 10mb bigger tham mine Ken's archive was 2-3mb smaller. Matbe it's bytes sequence issue. So, maybe enabling all that codecs aren't so big issue from binary side point of view? Sergey I. Sharybin wrote: Hi, I've done some testing with configurations. Here are some results. So, first of all i've disabled version3 and nonfree otions. First of them leads to gpl3 or higher library version, second maked libraries un-redistributable. Now libraries are gpl2 or higher. Also, i've disabled debugging symbols (--disable-debug flag). This shouldn't affect on size doe to all symbols (strip --strip-all) are stripping from blender binary. Then i've made simple config which gave exactly the same result in dependencies as ffmpeg which was used for 2.57a release. Conficuration is published in [1]. Blender binary after stripping was 43Mb. Don't think it's matetr due to linking happened against libraries which are ~1year newer, i think, and codebase of ffmpeg became larger (additional checkings, maybe architecture changed so some symbols which were unused and stripped now used and so on). Next test was with initial config from this thread with license cleaning and cleaning up flags which Peter and Joerg marked as unused and also disabled most of marked as UNSURE options. This options are related on speech (don't think this kind of audio is useful for Blender, but this libraries aren't big -- ~300Kb so could be added if they're necessery) and streaming (but dirac and schro keeped enabled -- it's algos of encoding/decoding raw videos). Configuration is published in [2]. Blender binary growed u[ to 49mb. This is also understandable, because all new dependencies (vpx, ogg, vorbis, theora, dirac, schro) are ~8mb in total. More codecs -- higher size of binary. Next step was disabled schro and dirac. Configuration posted to [3]. This saved 2mb and Blender binary became 47mb. I haven't tested with rtmp enabled because don't find stream sources b useful in Blender and this gives too much dependencies which weren't easy to solve -- some libraries came from Debian Sid environment, some from Debian Lenny, some from Lenny Backports.. Also, this adds ~2mb of Blender binary size. So, question is: do we want to support more codecs (ogg, vorbis, teora, vpx, dirac, schro) and have 8mb bigger Blender binary as we've got now? [1] http://www.pasteall.org/21098 [2] http://www.pasteall.org/21099 [3] http://www.pasteall.org/21100 -- With best regards, Sergey I. Sharybin -- With best regards, Sergey I. Sharybin -- With best regards, Sergey I. Sharybin ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
Re: [Bf-committers] Camera view navigation rant
Quick note that there have been a few new-devs posting on the mailing list wondering where to start - IMHO a fly mode alternative is a good project for beginner developers so get their feet wet and write something useful. On Mon, Apr 25, 2011 at 7:22 AM, Damir Prebeg blend.fact...@gmail.com wrote: I'm talking about this for long time now but I always get this beautiful answer: We don't need that because we have fly mode. And if we really want to align camera to view, we have Ctrl+0. If you really need that (although we don't know why would you need such a thing), Blender is opensource so make a patch by your self. Buzz off. Thumbs up Daniel! On 25 April 2011 05:50, Campbell Barton ideasma...@gmail.com wrote: On Sun, Apr 24, 2011 at 9:55 PM, Daniel Salazar - 3Developer.com zan...@gmail.com wrote: So can we take a look at camera view navigation? It's like playing a space flight simulator every time I press Shift F, I just want to position and orient my camera, not to deal with inertia and atmosphere entering angle! (feels like that at least) http://chrome.atari.com/lunarlander/ The flight mode panning (WASD keys) is of no use.. again calculating directions and inercia and keeping in mind how many presses I've done for each side so I can try and counter act when I want a stop! (which never works btw, the counter act seems to never be able to reach zero again) this could be a test for new NASA astronauts. Trackball is a decent tool for orienting the camera but it is about 10 times too sensitive, and somehow seems like it's going the wrong direction (inverted axis?) I have asked for simple game like controls in the past, you know those first person shooter controls? those are intuitive and precise but the answer was something like naah fly simulator is cooler, we can fly like birds!, really that was an actual answer, supposedly for live recording the camera to keyframes. Also a lock camera to view option would be great so we can just move the camera like we move any other viewport am I the only one that doesn't get this? peace, Daniel Salazar 3Developer.com Alternative view navigation modes would definitely be worth trying out and can be scripted without doing anything really complicated. See: Script Templates - Operator Modal View3D, this is a very simple operator that uses the mouse to move the view offset. in execute() you can set rv3d.view_location and rv3d.view_rotation API Ref: http://www.blender.org/documentation/blender_python_api_2_57_release/bpy.types.RegionView3D.html Reading you're mail I remembered rv3d.view_matrix was readonly, just committed support for assignment r36317. This makes manipulating the view easier, so you can assign the transform without worrying about the 'view_distance'. Of course once the basics are done this can become more involved, casting rays, smoother motion, mouse-look strafe :-) ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers -- - Campbell ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
Re: [Bf-committers] Camera view navigation rant
+1 this side, While on the topic of camera view navigation, if i may just add something slightly related: not being able to zoom with ctrl+mmb while in camera view catches me out a lot. I'll often have two viewports alongside one another - one in camera view and one in 3d - while animating to the camera etc. As i use a tablet and don't have a mouse wheel, it becomes 'trickier' to focus on a certain area in both views because i can use ctrl+mmb for the 3D view but have to drop the pen and grab a second mouse/laptop-trackpad-scroll to focus in the camera view. Rgds, J ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
Re: [Bf-committers] Camera view navigation rant
Ok so we got some attention :) now let's get this thing working great. Any py coders around? (better than me) Daniel Salazar 3Developer.com On Sun, Apr 24, 2011 at 3:55 PM, Daniel Salazar - 3Developer.com zan...@gmail.com wrote: So can we take a look at camera view navigation? It's like playing a space flight simulator every time I press Shift F, I just want to position and orient my camera, not to deal with inertia and atmosphere entering angle! (feels like that at least) http://chrome.atari.com/lunarlander/ The flight mode panning (WASD keys) is of no use.. again calculating directions and inercia and keeping in mind how many presses I've done for each side so I can try and counter act when I want a stop! (which never works btw, the counter act seems to never be able to reach zero again) this could be a test for new NASA astronauts. Trackball is a decent tool for orienting the camera but it is about 10 times too sensitive, and somehow seems like it's going the wrong direction (inverted axis?) I have asked for simple game like controls in the past, you know those first person shooter controls? those are intuitive and precise but the answer was something like naah fly simulator is cooler, we can fly like birds!, really that was an actual answer, supposedly for live recording the camera to keyframes. Also a lock camera to view option would be great so we can just move the camera like we move any other viewport am I the only one that doesn't get this? peace, Daniel Salazar 3Developer.com ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
Re: [Bf-committers] Camera view navigation rant
Yep, I agree :) so lets do this! Daniel Salazar 3Developer.com On Mon, Apr 25, 2011 at 7:07 PM, Matt Ebb m...@mke3.net wrote: On Mon, Apr 25, 2011 at 7:55 AM, Daniel Salazar - 3Developer.com zan...@gmail.com wrote: Also a lock camera to view option would be great so we can just move the camera like we move any other viewport This is really all that's needed imo. Funky fly modes and game controls may seem cool at first, but they're red herrings - not nearly as useful as just being able to rotate the view consistently as you would normally. I use it all the time in Houdini. ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
[Bf-committers] page for notes by those new to the codebase
Hi all, please add notes here http://wiki.blender.org/index.php/Dev:Ref/GoogleSummerOfCode/2011/NotesOnCode that explain variables that took you awhile to understand, as well as particularly useful functions, files or directorys. Hopefully we can reduce the time and pain of learning blenders code base. ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
[Bf-committers] recent commits after 36315 broke scons building with player
hi i get this error after the commit 36315 to build blender: - Creating library C:\Development\blender\build\win64-docs\bin\blender.lib and object C:\Development\blender\build\win64-docs\bin\blender.exp Generating code Finished generating code Trying to remove existing py bundle. Unpacking '../lib/win64/release/python32.tar.gz' to '..\install\win64-docs-36331 \2.57\python\lib' scons: building terminated because of errors. - i verified that i still able to compile 36315, but after updating i get the error, any one got the same thing. i can Build with 336331 without blenderPlayer i'm using windows x64 with scons and msvcsp1 any thought on that ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers