[hugin-ptx] Re: Error While Using Stacks

[Calvin McDonald]

Follow up:

Removing all the masks prevented the error and the pano completed, with 
stacks.
The problem appears to be related to the masking.

-- 
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
--- 
You received this message because you are subscribed to the Google Groups 
"hugin and other free panoramic software" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to hugin-ptx+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/hugin-ptx/ad38-265f-4713-ba2e-864fa5bd447f%40googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

[hugin-ptx] Error While Using Stacks

[Calvin McDonald]

I'm trying to stitch a 360x180 from 38 images (3x12+1N+1Z) exposure-stacked 
2-deep (76 total images) with Hugin 2011.2.0.3d9649aa241a.  I have a 
handful of hand-placed control points and 20 or so masks to remove unwanted 
lens-flare artifacts.  The only thing I'm doing different than the 100's of 
other times I've successfully run Hugin is the stacking.  I ran this one 
unstacked and it completed fine.  Here is the error message:

enblend: excessive overlap detected; remove one of the images
enblend: info: remove invalid output image "crackcanyon-pairs.tif"
make: *** [crackcanyon-pairs.tif] Error 1

Can someone please give me an idea how to work past this?

Thanks
Calvin

-- 
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
--- 
You received this message because you are subscribed to the Google Groups 
"hugin and other free panoramic software" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to hugin-ptx+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/hugin-ptx/61107e4d-3697-4ea3-a46f-0bf26d2f3ea5%40googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

Re: [hugin-ptx] Help My Skies

[Calvin McDonald]

Carlos:

Thanks so much for taking the time to respond.  Your tips will help me out.

I do play with masking a bit but I wouldn't call it methodical at all.  I 
kind of do a hit-and-miss approach as I have little skill at it.  
Occasionally I generate improvement with masks but usually not.  I haven't 
tried using other images.  I'll give that a try.  My PS skills improve with 
every pano I make.  However, I have a long way to go to fully use the 
capability of the tool.  Thanks for the image editing pointers - they will 
help.

Calvin



-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

Re: [hugin-ptx] Re: Exposure Stack Troubles

[Calvin McDonald]

My apologies John and Carl.  With both of you telling me it should work I 
decided to try again - even though I was convinced I did as you directed 
and it didn't work.  I tried again, and it worked beautifully.  
Unfortunately, I don't know what I (or Hugin?) did wrong before, but it's 
working nicely now.  Thanks for prodding me along.

Now that the blending anomaly is gone my interest has move to the shade of 
blue it selected for the sky.  It picked a paler shade of blue that I would 
like.  Is there a way to direct it to pick a darker exposure?

Calvin



On Wednesday, April 4, 2012 4:57:13 AM UTC-6, zarl wrote:
>
> JohnG schrieb am 04.04.12 02:40:
> >
> > I'm very disappointed to hear that the zenith stack include mask
> > didn't work ... in theory this should be an extremely elegant
> > workaround for Enblend's inability to read across the poles :-(
>
> It should work. If not, I'd check if the stack is properly defined in 
> the images tab, i.e. they have the same number in the "Stack Number" row.
>
> It works so nicely for me that I almost never fall back to masking my 
> input images in Photoshop.
>
> >> On Apr 3, 10:58 pm, Calvin McDonald  wrote:
> >>> On Saturday, March 31, 2012 6:15:46 AM UTC-6, zarl wrote:
> >>>
> >>>> Another option could be to force hugin to use the "sky" parts of your
> >>>> zenith shot with an "include" mask. And since you're using stacks you
> >>>> should define the mask to include every other image in the same stack
> >>>> (there's a drop down list for that in the mask tab).
> >>>
> >>> It seems non-intuitive to me to set an include mask on all images in 
> the
> >>> stack to solve the problem.  It would seem that setting an include 
> mask on
> >>> the one stack image with the properly exposed sky would be right. 
>  However,
> >>> I tried both ways (mask on 1 image, mask on all 7 in the stack) and 
> neither
> >>> worked.  Both produced the same bad result as pictured in my first 
> post.
>
> Calvin, make sure you explicitely set the mask in the drop-down list 
> "Mask type:" to -> "Include region from stack"
> (I'm not sure if this is indeed an intuitive phrase)
>
> Carl
>
>
On Wednesday, April 4, 2012 4:57:13 AM UTC-6, zarl wrote:
>
> JohnG schrieb am 04.04.12 02:40:
> >
> > I'm very disappointed to hear that the zenith stack include mask
> > didn't work ... in theory this should be an extremely elegant
> > workaround for Enblend's inability to read across the poles :-(
>
> It should work. If not, I'd check if the stack is properly defined in 
> the images tab, i.e. they have the same number in the "Stack Number" row.
>
> It works so nicely for me that I almost never fall back to masking my 
> input images in Photoshop.
>
> >> On Apr 3, 10:58 pm, Calvin McDonald  wrote:
> >>> On Saturday, March 31, 2012 6:15:46 AM UTC-6, zarl wrote:
> >>>
> >>>> Another option could be to force hugin to use the "sky" parts of your
> >>>> zenith shot with an "include" mask. And since you're using stacks you
> >>>> should define the mask to include every other image in the same stack
> >>>> (there's a drop down list for that in the mask tab).
> >>>
> >>> It seems non-intuitive to me to set an include mask on all images in 
> the
> >>> stack to solve the problem.  It would seem that setting an include 
> mask on
> >>> the one stack image with the properly exposed sky would be right. 
>  However,
> >>> I tried both ways (mask on 1 image, mask on all 7 in the stack) and 
> neither
> >>> worked.  Both produced the same bad result as pictured in my first 
> post.
>
> Calvin, make sure you explicitely set the mask in the drop-down list 
> "Mask type:" to -> "Include region from stack"
> (I'm not sure if this is indeed an intuitive phrase)
>
> Carl
>
>

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

Re: [hugin-ptx] Exposure Stack Troubles

[Calvin McDonald]

On Saturday, March 31, 2012 6:15:46 AM UTC-6, zarl wrote:
>
> Hi Calvin,
>
> Calvin McDonald schrieb am 30.03.12 21:32:
> > Just wanted to ask the group if the Hugin behavior I'm seeing is
> > expected and if not, get some feedback on workarounds for the problems
> > I'm having.
> >
> > Here's the configuration: 3x12+1N+1Z - all but the nadir are stacks of 7
> > exposure bracketed shots (total 260 images). Each stack of 7 is position
> > Linked. There is just one nadir shot - which is properly exposed. The
> > CP's are inserted only between the middle image of the 7 in each stack.
> > The exposure fusing is working nicely in general and I'm getting a
> > really good stitch alignment-wise, but, I'm having unusual problems with
> > the nadir and zenith.
> >
> > ZENITH:
> > Below is a pic of the zenith. I've never seen this before. Why is Hugin
> > doing this and what can I do to fix it?
>
> This is an enblend problem, the panotools wiki has a tutorial that deals 
> with it:
>
> http://wiki.panotools.org/How_to_remove_blending_error_caused_by_enblend_and_enfuse_at_zenith_and_nadir_%28automatic%29
>
> Another option could be to force hugin to use the "sky" parts of your 
> zenith shot with an "include" mask. And since you're using stacks you 
> should define the mask to include every other image in the same stack 
> (there's a drop down list for that in the mask tab).
>
It seems non-intuitive to me to set an include mask on all images in the 
stack to solve the problem.  It would seem that setting an include mask on 
the one stack image with the properly exposed sky would be right.  However, 
I tried both ways (mask on 1 image, mask on all 7 in the stack) and neither 
worked.  Both produced the same bad result as pictured in my first post.

I'll go read up on and try the panotools method
 

> > NADIR:
> > Several of the 12 groups of 7 images of the bottom row include the feet
> > of my tripod. This is very typical for me and I've never had Hugin not
> > remove them for me with no intervention when including and blending in
> > the nadir. This is the first time I've done exposure stacks on the
> > bottom row. For some reason Hugin is leaving in the tripod feet. I've
> > included an image of this below.
> >
> > So I went in and masked them out. But this makes Hugin crash.
>
> Can you reproduce that crash and is there an error message? Consider 
> filing a bug report together with a detailed description of what leads 
> to that crash.
>
I didn't try to reproduce it.  With a 4 hour build time I'm not very 
motivated to run a lot of experiments.  There was no gui error message, 
Hugin just terminate.  The tools might have left a log file around but I 
don't know them well enough to know what to look for.

I do know that changing from 3-4 masks in each stack (on individual images) 
to a single mask on all 7 images in the stack made the crash go away.
 

> > I note that when cropping, Hugin will allow multiple images to be
> > cropped at once, but when masking, it will only allow one image
> > selection at a time. When I masked the tripod feet out I had to do it
> > one image at a time. To save time I only masked the images with
> > exposures close to correct. So of the group of 7, I only masked 3 from
> > each of the 12 sets. Hugin crashes when I do this. It crashes when doing
> > the build.
>
> First of all there are two different types of cropping: you can crop 
> input images (individually or with the same coordinates if you use the 
> "load lens parameters" button in the "camera and lens" tab), but you can 
> also crop the output image (See the "fast preview" window's "crop" tab).
> All images that use the same stack (manually: in "images" tab select all 
> images of one stack and click "new stack", repeat this for all other 
> stacks) you can now set a mask to include (or exclude) taht area in 
> every image of the same stack.
>
> Hugin (2011.2 and later) also lets you copy/paste masks. Before you had 
> to save/load a mask for a similar workflow.
>
> There's another option: tiffs (instead of e.g. jpgs as input image file 
> format) can include masks directly in the files (as alpha masks).
>
> > If I have the stacked images Linked, does the mask affect all 7 images,
> > or just the one's masked? Could this be what's causing Hugin to crash?
> > Do I need to mask all 7? Just one? Is there a way to select all 7 and
> > apply a mask to all 7 at once?
>
> Mask behaviour is defined in the "mask" tab, se

[hugin-ptx] Exposure Stack Troubles

[Calvin McDonald]

Just wanted to ask the group if the Hugin behavior I'm seeing is expected 
and if not, get some feedback on workarounds for the problems I'm having.

Here's the configuration:  3x12+1N+1Z - all but the nadir are stacks of 7 
exposure bracketed shots (total 260 images).  Each stack of 7 is position 
Linked.  There is just one nadir shot - which is properly exposed.  The 
CP's are inserted only between the middle image of the 7 in each stack.  
The exposure fusing is working nicely in general and I'm getting a really 
good stitch alignment-wise, but, I'm having unusual problems with the nadir 
and zenith.

ZENITH:
Below is a pic of the zenith.  I've never seen this before.  Why is Hugin 
doing this and what can I do to fix it?


NADIR:
Several of the 12 groups of 7 images of the bottom row include the feet of 
my tripod.  This is very typical for me and I've never had Hugin not remove 
them for me with no intervention when including and blending in the nadir.  
This is the first time I've done exposure stacks on the bottom row.  For 
some reason Hugin is leaving in the tripod feet.  I've included an image of 
this below.

So I went in and masked them out.  But this makes Hugin crash.  

I note that when cropping, Hugin will allow multiple images to be cropped 
at once, but when masking, it will only allow one image selection at a 
time.  When I masked the tripod feet out I had to do it one image at a 
time.  To save time I only masked the images with exposures close to 
correct.  So of the group of 7, I only masked 3 from each of the 12 sets.  
Hugin crashes when I do this.  It crashes when doing the build.  

If I have the stacked images Linked, does the mask affect all 7 images, or 
just the one's masked?  Could this be what's causing Hugin to crash?  Do I 
need to mask all 7?  Just one?  Is there a way to select all 7 and apply a 
mask to all 7 at once?

Thanks in advance for any suggestions anyone might have.

Calvin



-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

Re: [hugin-ptx] Looking for recomendations on a new camera

[Calvin McDonald]

I'm not familiar with Canon but if Nikon is appealing to you, the D90, D300 
or D700 would be excellent choices, depending on budget.  I'd also look at 
the D5100 or D7000 if higher sensor resolution is important to you.

The new Nikon D4 just showed up on Nikon's website recently.  It's a little 
disappointing that it's only 16mp.  Not sure if it's really available yet.

http://imaging.nikon.com/lineup/dslr/index.htm

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

[hugin-ptx] Re: Help with Enfuse within Hugin

[Calvin McDonald]

Kay

On Saturday, December 31, 2011 1:26:32 AM UTC-7, kfj wrote:

Reading your more detailed description, I actually had the same idea. 
> The pattern of CP misalignment looks like it might have been the 
> result of a slight change in focal length between the first and second 
> series. Luckily this is easy to correct. 


You got it - making a second lens did it!  I loaded my latest .pto file, 
gave the second set of 12 images their own lens, enabled optimization for 
FOV and Optimized and the error instantly dropped down to 1.7 - a "good 
fit".  The resulting image is beautiful.  I'm building the full resolution 
version now.

Thanks so much for taking the time to help me!  I really learned a lot 
about Hugin in the process.
Calvin

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

[hugin-ptx] Re: Help with Enfuse within Hugin

[Calvin McDonald]

Kay:

On Thursday, December 29, 2011 4:21:52 AM UTC-7, kfj wrote:
>
> On 29 Dez., 01:48, Calvin McDonald  wrote: 
>
> > Result was good other than it didn't fix the misalignment of the stacked 
> > images.  I'm getting 20 or so pixels of edge ghosting on the sky 
> boundary 
> > still.  I experimented with many CPs on the sky boundary - didn't help. 
>  I 
> > experimented with "Exposure fused from any arrangement" - changed things 
> a 
> > little but didn't improve the result. 
> > 
> > Ideas? 
>
> If your CPs don't have an effect on the alignment, are you sure to 
> have run the optimizer after placing the CPs? CPs per se don't have 
> any effect whatsoever. 
>
> Why do you bother with handmade control points? Just run cpfind on 
> your image set, this should give you much better alignment than just 
> using four handmade points. 
>

Sorry, my description was misleading.  I do use cpfind.  As an experiment I 
manually placed a bunch of CPs on the sky boundary to see if that would 
help (it didn't).  I also Optimize, usually a bunch of times while tuning 
up/out the occasional misplaced CPs.
 

> Now this has been going on for quite a while and I feel I might help 
> you more effectively if I see 
>
> a) your images 
> b) the pto you use for them 
>
> do you have a place where you can put them online so I can have a go 
> at them (doesn't have to be the full set, and can be JPEG to save 
> space)? 
>

Given I'm really just trying to fix a mistake I made upstream (not taking 
the bracketed exposure shots back-to-back) I'm starting to feel like I'm 
wasting my time and yours.  I'll forgo sending you a test-case, unless you 
want one.

I started from scratch in Hugin and followed all your guidance.  I got a 
similar bad result.  Something is very odd about this image-set.

One last question.  What would happen if the focal length of my zoom lens 
changed just a small amount (<1mm) between the time I took the two sets of 
12 upper-row exposure-bracketed shots?  Could that cause what I'm 
experiencing?  My zoom (Nikkor 18-200 VR-II) has a zoom lock at 18mm and it 
was locked when I took the images in question.  But, I've noticed that even 
when it's locked the zoom barrel can still move a little.  I haven't had a 
problem in the past like this and I've done over 100 panos with this lens.  
Just a thought - kind of a long shot.  In general though, if the focal 
length did change, can Hugin/tools deal with that?

Calvin

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

[hugin-ptx] Re: Help with Enfuse within Hugin

[Calvin McDonald]

Kay:

On Thursday, December 29, 2011 4:21:52 AM UTC-7, kfj wrote:

On 29 Dez., 01:48, Calvin McDonald  wrote: 
>
> > Result was good other than it didn't fix the misalignment of the stacked 
> > images.  I'm getting 20 or so pixels of edge ghosting on the sky 
> boundary 
> > still.  I experimented with many CPs on the sky boundary - didn't help. 
>  I 
> > experimented with "Exposure fused from any arrangement" - changed things 
> a 
> > little but didn't improve the result. 
> > 
> > Ideas? 
>
> If your CPs don't have an effect on the alignment, are you sure to 
> have run the optimizer after placing the CPs? CPs per se don't have 
> any effect whatsoever. 
>
> Why do you bother with handmade control points? Just run cpfind on 
> your image set, this should give you much better alignment than just 
> using four handmade points. 
>

Sorry, my description was poor.  I did use cpfind, I meant that afterwards 
I placed a bunch of CPs on the sky boundary by hand to see if that would 
help.  I use optimization also - actually quite a few times, as I tune the 
cpfind CPs.

There is something very odd about this particular image set.  In Hugin I 
completely started over from scratch and followed all your guidance, 
checked all the cpfind generated CPs (no manual ones this time) very 
closely for misplaced ones.  After several optimization runs I could only 
get the error down to 5.04 with the max error being 22.4 - I've never had a 
pano converge to such a poor fit.  Most of the CPs at the top of the error 
list are between the pairs of stacked images.  Final result - no change - 
still bad.

>From my inexperienced observation - it simply appears that the tools are 
unable to align the top portions of the top row of 12 stacked images.  
Everything looks fine from about the middle (going up/down) of this row of 
images and down.  But from the middle going up they progressively get more 
what looks like misaligned until the sky boundary where it looks the 
worst.  It's likely the worst at the top in the sky but the errors all get 
lost in the sky/clouds.

Now this has been going on for quite a while and I feel I might help 
> you more effectively if I see 
>
> a) your images 
> b) the pto you use for them 
>
> do you have a place where you can put them online so I can have a go 
> at them (doesn't have to be the full set, and can be JPEG to save 
> space)? 
>

Given I'm really just trying to find a method to fix a mistake I made 
upstream (didn't take the stacked images back-to-back) I'm starting to feel 
like I'm wasting both our times.  I've fought this sufficiently to force me 
never to forget to shoot my exposure bracketing shots back-to-back.  :)

I'll forgo troubling you with handing off a test-case to you, unless you 
want it.  I do have a place I could drop it online.

One last question. Back on the thought that something very unusual is 
wrong What would happen if the focal length of a zoom lens were to 
change just a bit between a pair of otherwise identical exposure-stacked 
images?  I ask because I did shoot these images with a zoom lens, a Nikkor 
18-200mm VR-II.  I shoot all my panos with this lens, with it zoom-locked 
at 18mm.  It was definitely locked at 18mm when I took the images in 
question but I've noticed that even when it's locked the zoom barrel can 
still move a little.  This is probably a long-shot, but what if the focal 
length got bumped slightly between the two sets of stacked images.  Could 
that account for the trouble I'm having?  I'm having a hard time believing 
this could because I've never had this problem before in the 100+ other 
panos I've done with this same lens.

Thanks
Calvin

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

[hugin-ptx] Re: Help with Enfuse within Hugin

[Calvin McDonald]

On Tuesday, December 27, 2011 8:56:03 AM UTC-7, Calvin McDonald wrote:

With your tips I will go back and try again with Hugin.  I'll let you know 
> how it goes. 
>

Kay:

Here's what I have done.

 1.  Load 38 basic images (3x12+1N+1Z)
 2.  Read in placement template  -- (my standard workflow to 
this point)
 3.  Read in 12 (1x12 top row) images with properly exposed sky
 4.  Read in properly exposed Z shot and put it in place of the washed out 
one (it only has sky)
 5.  Stacked the 12 images on top of their counterpart images
 6.  Removed the "Link" checkbox for all 12 stacked images by "Image 
Position" on the Images page
 7.  Used no masking, other than on the tripod feet in the nadir shot.
 8.  Selected "Exposure corrected, low dynamic range" AND "Exposure fused 
from stacks"
 9.  Placed at least 4 CPs on all image boundaries, including the 12 
stacked pairs.
10.  Did the regular tuning of CPs and got the std-dev error down to about 
20.  (best I could get it)
11.  Ran Align.  (std-dev came down to about 4)
12. Created the pano

Result was good other than it didn't fix the misalignment of the stacked 
images.  I'm getting 20 or so pixels of edge ghosting on the sky boundary 
still.  I experimented with many CPs on the sky boundary - didn't help.  I 
experimented with "Exposure fused from any arrangement" - changed things a 
little but didn't improve the result.

Ideas?

Calvin

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

[hugin-ptx] Re: Help with Enfuse within Hugin

[Calvin McDonald]

On Tuesday, December 27, 2011 5:22:15 PM UTC-7, Calvin McDonald wrote: 
 

> > > If you're on windows, there are also the enfuse-align droplets, where 
>> > > you can just drag and drop sequences of images onto them and have an 
>> > > exposure fusion made from an aligned unlinked stack. This might be an 
>> > > option if you want to manually enfuse but not use hugin for it. 
>> > 
>> > I am on Windows but I'm not sure what you are referring to.  Can you 
>> > elaborate a little? 
>>
>> When I was still using Windows, the hugin packets had a script called 
>> enfuse_align_droplet.bat. You could link it onto your desktop and drag 
>> and drop sets of images onto it to have exposure fusions made from 
>> them. Maybe it's been taken out of the bundles? Look in your hugin 
>> installation's folder. 
>>
>
> Ah, I have it.  I'll play with it and see what I get.
> So many fun toys, so little time
>

Kay:

I used enfuse_align_droplet out of the box (as-is, no tweaks).  I dropped 
two of my images on it, one with washed out skies, and one with good 
skies.  The two images have the slight alignment problem we have been 
discussing.  The script ran fine, the skies look great and the fuse 
boundary is at the sky where I wanted it.  But, it didn't fix the 
alignment.  There is about 20 pixels (width) of edge ghosting along most of 
the sky boundary. 

Calvin

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

[hugin-ptx] Re: Help with Enfuse within Hugin

[Calvin McDonald]

Kay:

On Tuesday, December 27, 2011 2:38:02 PM UTC-7, kfj wrote:


>
> On 27 Dez., 16:56, Calvin McDonald  wrote: 
>
> > > Sadly I couldn't find a way to make hugin keep the fused stacks before 
> > > feeding them to enblend, which would be nice for you to play with (am 
> > > I blind? or is there really no such option?). What you can do, though, 
> > > is keep all the remapped images (no photometric correction, LDR) and 
> > > enfuse them manually. They should be well-aligned now they have been 
> > > made from unlinked stacks. 
> > 
> > You're not asking me these questions are you?  I'm such a novice I'm not 
> > sure I understand the questions, let alone the answers!  :) 
>
> Not really - I just thought maybe someone else is reading this, too. I 
> find the stitcher tab hard to figure out, and I think a novice will 
> have a hard time figuring out what different types of output he can 
> get.


Right you are... I'm a novice for sure and I find the Stitcher page highly 
bewildering.  :)
 

> > > If you're on windows, there are also the enfuse-align droplets, where 
> > > you can just drag and drop sequences of images onto them and have an 
> > > exposure fusion made from an aligned unlinked stack. This might be an 
> > > option if you want to manually enfuse but not use hugin for it. 
> > 
> > I am on Windows but I'm not sure what you are referring to.  Can you 
> > elaborate a little? 
>
> When I was still using Windows, the hugin packets had a script called 
> enfuse_align_droplet.bat. You could link it onto your desktop and drag 
> and drop sets of images onto it to have exposure fusions made from 
> them. Maybe it's been taken out of the bundles? Look in your hugin 
> installation's folder. 
>

Ah, I have it.  I'll play with it and see what I get.
So many fun toys, so little time

Thank
Calvin
 

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

[hugin-ptx] Re: Help with Enfuse within Hugin

[Calvin McDonald]

Kay:

Thanks for the response.  Comments below

You should be able to do it all in hugin. When you say your attempts 
> failed, what went wrong? 
>

First attempt I got hit by the misalignment.  Reading the Hugin docs I 
decided to select "Exposure fused from any arrangement" rather than 
"Exposure fused from stacks" and that fixed the misalignment problem.  My 
next problem was where Hugin (or sub-tool) decided to join the images.  It 
didn't in general join them at the sky boundary.  In many places it did it 
lower in the image and it appeared the blender (I think) was really messing 
up the canyon walls.  My final few attempts where messing with masks trying 
to get the joint on the sky boundary.  This failed because it made the sky 
boundary look bad.

There is one common mistake in your situation. I suppose you still 
> have your initial attempt with all the images in hugin. Since your 
> stacks are not perfectly aligned the way you took them, you have to 
> uncheck the 'link' checkbox next to the image position fields in the 
> images tab for each of your misaligned stacks. If you don't do that, 
> the stacked images will be placed precisely on top of each other and 
> their positions will be linked, and it sounds as if this was happening 
> with your panorama.
>

Yes, I made this mistake.  I did not unselect "link" for the 12 stacked 
images on any attempt I tried.

You need control points between the images in the stacks in order to 
> align them, so in case you don't have them (linked stacked images can 
> do without control points), you'll need another run of a control point 
> generator - align_image_stack is popular for the specific case, but 
> cpfind with settings to create a few more points than usual (try -- 
> fullscale --sieve2size 5 -o %o %s) works well, too - and another 
> optimizer run.


I also made this mistake.  I typically get most of my panos to stitch 
without any CPs.  This one had non for every attempt I made - not even 
between the non-stacked images. 
 

> Once everything is optimized, the minor shifts between the stacked 
> images should be dealt with and you can proceed to let hugin create 
> the fused panorama. You can check the quality of the alignment in the 
> preview by setting the viewing mode to difference in the preview tab 
> of the preview. If the overlap is generally very good, instead of 
> creating a panorama from fused stacks you can also try and go the 
> other way by creating an exposure fusion from an arbitrary 
> arrangement, which has the added benefit that it lets you keep the 
> intermediate layers to blend yourself if you want. 
>

About the time I started this thread and before you replied, I got 
frustrated and used Photoshop to merge in the properly exposed sky in my 12 
upper row images.  If it wasn't for the misalignment, this method wouldn't 
have been that painful.  With a little work however, I got this done in PS 
and then used the resulting 12 images for the top row in Hugin.  This 
worked fairly well - but not great - the land-sky boundary looks a little 
unnatural.  See the result here:  

http://www.360cities.net/image/squaw-creek-canyon-falls-idaho-usa#261.30,-79.19,110.0

With your tips I will go back and try again with Hugin.  I'll let you know 
how it goes.
 

> Do trials with small output sizes to see what looks about right, then 
> when you think you're getting there, crop out a typical section and do 
> it higher-res, then when all seems fine have a go at the full pano. 
>
> Sadly I couldn't find a way to make hugin keep the fused stacks before 
> feeding them to enblend, which would be nice for you to play with (am 
> I blind? or is there really no such option?). What you can do, though, 
> is keep all the remapped images (no photometric correction, LDR) and 
> enfuse them manually. They should be well-aligned now they have been 
> made from unlinked stacks. 
>

You're not asking me these questions are you?  I'm such a novice I'm not 
sure I understand the questions, let alone the answers!  :) 
 

> If you're on windows, there are also the enfuse-align droplets, where 
> you can just drag and drop sequences of images onto them and have an 
> exposure fusion made from an aligned unlinked stack. This might be an 
> option if you want to manually enfuse but not use hugin for it.


I am on Windows but I'm not sure what you are referring to.  Can you 
elaborate a little?

And I'm sure there are many more possibilities... it's confusing at 
> first, but you'll get there ;-) 
>
> Kay


Thanks again Kay - I'm off to give Hugin another try.

Calvin 

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit

[hugin-ptx] Re: Help with Enfuse within Hugin

[Calvin McDonald]

Thanks for the help Kay.

After 5-6 failed attempts trying to exposure-fuse within Hugin, I'm now 
trying image-set by image-set with command-line enfuse.

I'm getting pretty good (acceptable) exposure-fusing results via 
command-line, so I'm encouraged.  However, I'm on to the next problem 
alignment.  I know better now but when I took the image-set I made the 
mistake of not taking the two exposure shots back-to-back.  I took the 12 
upper row shots at exposure setting #1, then went back and took the 12 with 
the alternate exposure.  Although tripod mounted with a nice detent pano 
head, it of course didn't create perfect alignment between the image 
pairs.  The alignment error is about 20 or so pixels - just enough to make 
the transition from land to sky look horrid.

How do I align them?

Calvin

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

[hugin-ptx] Help with Enfuse within Hugin

[Calvin McDonald]

I'm needing a little driving instructions when using Hugin to combine 
stacked images of bracketed exposures.

I normally take image sets 3x12+1N+1Z.  I took a set in a dark canyon, 
exposed for the canyon which naturally washed out the sky.  The sky is only 
visible in the zenith shot and the top row of 12 images.  For these 13 
images I took two shots at each position, one exposed for the dark canyon 
and one for the blue sky.

Now I'm in Hugin and fumbling around (never done this before) trying to get 
it to enfuse the two exposures of the top row for me.  I loaded the 
standard 38 images and positioned them properly with a template.  Then I 
loaded the extra 13 images of the properly exposed sky and stacked them on 
top of their counterpart images.

My question is, what options do I want on the "Stitcher" tab to get it to 
combine the two exposures automatically?

Calvin

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

[hugin-ptx] Hints to Hugin about Image Positions Relationships

[Calvin McDonald]

It's very likely pilot error but I'm having a hard time getting cpfind 
(2010.4.0) to figure out my image positions.
It seems to find the horizontal overlaps OK,... other than it seems to 
frequently miss the overlap between image 11 back to 0 (of 12 per row).
I'm having the most problem with the vertical direction.  I'm doing 
3x12+1Z+1N and cpfind almost never figures out that image 12 is above 0 and 
so on.  It's fairly common for it to find none of the vertical overlaps 
(1/13, 2/14, 3/15, ... 12/24, 13/25, .. etc).

I'm a novice user so am doing nothing presently to assist the tool.  Is 
there a way to give the tool hints about the image positions so it more 
reliably finds all the overlaps?
If I select just images 0 and 12 (for example) and run cpfind it does a 
great job of finding CPs between them - but if I run cpfind on all the 
images it commonly misses the fact the 0/12 overlap.

I'm using this cpfind command-line from within Hugin:
--multirow -n 4 --fullscale --kdtreeseconddist 0.25 --minmatches 1 
--sieve2width 2 --sieve2height 2 --sieve2size %p -o %o %s

Thanks
Calvin

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

Re: [hugin-ptx] How Does Hugin Arrive at the Resulting Canvas Size?

[Calvin McDonald]

Bruno:

Thanks - you pointed out my problem correctly.  My downscaling preference 
was set to 70.  I didn't even know Hugin had this feature and have been 
suffering the results of it on every pano I've done apparently since I last 
updated Hugin.

Does Hugin really come built with this set to 70?

In reading the Hugin help it makes it sound like the downscaling is a post 
processing activity.  It even claims the algorithm is inferior and suggests 
if downscaling is needed to use an image editor post-Hugin.  If it's really 
a post processing activity I wouldn't think it would speed up the stitching 
process.

Can someone tell me why in general someone would want their pano downscaled 
by Hugin?
In other words, is there a good reason to set this to something besides 
100%?

I did change my "Downscale final pano" to 100% and the resulting pano was 
19,450 x 9,725 - the same as Hugin estimated when I click "Calculate optimal 
size..." under the Stitcher tab.

Thanks Bruno and Carlos for the help.
Calvin

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

[hugin-ptx] How Does Hugin Arrive at the Resulting Canvas Size?

[Calvin McDonald]

Why does Hugin consistently build results much smaller than the dimensions 
it gives me if I click "Calculate optimal size"?

My lens has an angle-of-view of 46 degrees (on the sensor) in the horizontal 
direction.
I take 12 images per row.
So of the 46 degrees captured per image, 30 degrees are needed to complete 
the pano.
I believe this is approximately a 35% overlap.
This tells me that of each image of the 12 that 65% of the pixels will end 
up being used (on average).
My sensor in the horizontal direction is 2,592 pixels.

My simple mind wants to believe that the resulting stitched image will be 
approximately 2,592 * 0.65 * 12 = 20,218 pixels.
But Hugin generates a 13,250 x 6,625 result (or thereabouts).

When I load an image-set and click on "Calculate optimal size" Hugin sets 
the Canvas Size to 19,450 x 9,725.  It appears my math is consistent with 
what Hugin tells me the "optimal" size is.  But without exception when a 
pano is complete the X dimension is in the low 13,000 range.

Would someone kindly explain to me the reason for this difference?

Thanks

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

[hugin-ptx] cpfind halts

[Calvin McDonald]

I'm using Hugin 2011.0.0.0fd3e119979c on a quad-core 64bit Win7 machine.  I 
noticed recently that if I create a new line across two adjacent images and 
then select these same two images only in the "Images" panel and tell Hugin 
to "Create control points" that I get a message that cpfind halted.  This is 
repeatable.

Is this a bug or am I using the tool incorrectly.

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

Re: [hugin-ptx] Fast Panorama Preview is freeze

[Calvin McDonald]

Harry:

I have seen this same freezing behavior on my 64-bit Win7 box and Hugin 
2011.0.0.  However, it is not consistent, meaning it only hangs sometimes.  
It also hangs sometimes when I invoke the Viewer manually.  I need a little 
help understanding what you are suggesting in the way of using the correct 
display driver or driver update.

My system has an ATI Radeon HD 4350 and a Properties interrogation of the 
device says it using an ATI Technologies driver dated 6/14/2009.

Q1:  I assume this means I'm *not* using the "Windows screen drivers", 
correct?
Q2:  Are you suggesting that my hanging problem might be solved if I update 
to a new ATI driver?

Thanks
Calvin

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

[hugin-ptx] Re: help! none of my lenses seem to point straight ahead!

[Calvin McDonald]

Hi Kay.

Interesting experiment I just had to try it.  I did as close to what you 
described as I could from your description with my Nikon D200 / 18mm.  Over 
1m I got a downward deflection of 0.15cm.  However, the bottom of my D200 
isn't all that large and so there is a bit of play in the movement of the 
camera when I attempt to hold it flat on the tabletop.   Also, I'm not sure 
how planar my tabletop is either.  Considering these two variables, I think 
this could account for the 0.15cm deflection I'm seeing so I'm not prepared 
to believe my camera/lens are misaligned.

If I view the ruler through the viewfinder while holding the camera bottom 
tight to the tabletop, but then press a little harder on the front or rear 
of the camera, I can see deflections of 1/10 of a cm, all the while the 
camera feels like it's flat on the tabletop.

I don't have a second lens to try at the moment.

I would think that if your camera body is in some way out of alignment that 
you'd get a consistent direction of deviation from lens to lens.  The fact 
that your two lenses deviate in different directions could suggest a sloppy 
lens mount or simply error in the measurement - such as the camera not 
really being flat to the tabletop or the tabletop not being planar.


Calvin

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

Re: [hugin-ptx] HDR for Beginners

[Calvin McDonald]

Thanks Carlos.  Command-line is working for me, other than one thing...
I gave it two images, one with a properly exposed foreground and a washed 
out sky and the other the opposite, under exposed foreground and properly 
exposed blue sky.  The resulting image has a properly exposed foreground and 
an improved sky but not as well exposed as the second photo.  It's like it's 
blending bad sky with good sky and coming up with fair - rather than just 
taking the properly exposed sky and using it.

Is this life or cockpit error?

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

[hugin-ptx] Re: Just tried out 2011.0.0, but the fast panorama preview screen is "Not Responding" (ie. hangs)

[Calvin McDonald]

I'm experiencing the same thing on 64-bit win7.
I filed a bug report - ticket #792896.


-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

Re: [hugin-ptx] HDR for Beginners

[Calvin McDonald]

Milan:

Thank you for the pointers - very helpful.  You are right, I'm more 
interested in exposure fusion than HDR - I hadn't separated the two in my 
mind.

I gave Hugin a run as you suggested... but it crashed and asked me to file a 
bug report.

I dumped the exposure stacks and run just a regular stitch and it crashed 
too.  So rather than learning how to do exposure fusion I'm off filing bug 
reports on Hugin.

Thanks again.  I'll start up on this again when Hugin is more stable.
Calvin

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

[hugin-ptx] HDR for Beginners

[Calvin McDonald]

I have a set of 190 images defining a spherical pano taken with an 18mm lens 
in this arrangement:  (12x3 + 1Z + 1N) x 5, where the 5 is 5 levels of 
exposure bracketing (-2, -1, 0, +1, +2).  The scene is outdoors in a shaded 
canyon.  The normal exposure was for the shaded area so I'm getting washed 
out mountain peaks and skies.

I'm reasonably comfortable running Hugin but have never messed with the HDR 
stuff before.  I'm running Hugin version 2011.0.0.  I have read in the 190 
files, positioned them and grouped them 5 to a Stack.  I'm expecting Hugin 
to automatically select and stitch/blend in the proper exposure segments for 
the mountain peaks and skies.

I'm reading but not completely understanding the help notes about the 
options available on the bottom of the "Stitcher" screen (the 9 check-box 
items).

Can someone help walk a beginner through these?

Thanks in advance.

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

[hugin-ptx] Re: Hugin Session Recipes

[Calvin McDonald]

Thanks Kay, insightful.

Your comment "perfection is an hypothetical endpoint" did remind me that we 
are playing with an art here, not an exact science.  My personal perspective 
is that a spherical is "perfect" when I can zoom in fully and study the 
image for 10 minutes and not see anything I know is a stitch/blend 
artifact.  Indeed there may be artifacts hidden nicely by the 
stitcher/blender but if they are not noticeable or distracting in any way 
from the visual experience then it's irrelevant to me if they are there.

Prior to my last equipment upgrade and refined processing workflow I use to 
get 20 to 30 percent failures - meaning with my limited experience in the 
Hugin cockpit I wasn't able to remove enough of the visual artifacts to 
publish the image.  When I was successful, I found I was spending 2-3 hours 
/ pano (30-40 images/set) pulling it together, which I found highly 
undesirable.

Going from 2-3 hours to 5-10 minutes was a major breakthrough for me.

I was intrigued by your comments of using differing resolution depths in the 
same pano.  This is something I haven't messed with yet.  Would you happen 
to have such a pano published on the web somewhere that I could look at?  
This might be something I want to explore.

Thanks
Calvin




-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

Re: [hugin-ptx] Re: Hugin Session Recipes

[Calvin McDonald]

Christopher:

> > In the meantime I have been developing a new recipe of my own.  I now 
own a
> > precision head (detents vert & hor)...

> What kind?

NN4
> > None the less, I found that if I dial in the FOV that I
> > am able in some cases to get pretty much perfect stitches with no CPs!

> Impressive!

Here is an example with no CPs:
http://www.360cities.net/image/hells-canyon-snake-river-oregon-idaho-usa

I provide an example because one person's definition of an artifact-free 
pano may be different than another person's.  Maybe mine isn't all that 
great - I would love some negative feedback on mine if anyone finds it less 
than desirable - so I can improve.

> > With this new recipe I'm getting great stitches and only investing about 
5
> > to 10 minutes per pano (excluding computer processing time).  I'm pretty
> > happy with this.

> Yes, that's excellent.

> Do you not bother with CPs for the top/bottom row?

It might be the nature of the type of pictures I'm taking but no, I haven't 
placed any CPs in the top or bottom row of 12, nor in the N or Z.  I only do 
the CPs in the horizontal row when a no CP stitch produces a little 
parallax.  I've only done 4 or 5 this way so far but all have stitched near 
perfectly.  Here are two examples of CPs just in the horizontal row:

http://www.360cities.net/image/canyon-hill-cemetery-caldwell-idaho-usa
http://www.360cities.net/image/the-boise-river-at-caldwell-canyon-idaho-usa

Again, critique is welcome - I like raising my bar.

Calvin


-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

[hugin-ptx] Re: Hugin Session Recipes

[Calvin McDonald]

Thanks all for the comments... I've been digesting them with interest.

In the meantime I have been developing a new recipe of my own.  I now own a 
precision head (detents vert & hor) and have derived distortion correction 
parameters for my lens.  I'm shooting a 18-200mm zoom locked at 18mm.  I 
shoot 3x12 + 1N + 1Z.  

I reported earlier that I was still having more difficulty than I expected 
to have (realistic exceptions is questionable) using an image position 
template and a lens distortion profile only.  I was expecting (hoping) to 
get quality stitches with no CPs, but I wasn't.  I have since discovered 
that it was my FOV that was getting me.  My FOV was coming from the camera 
and I was naively believing it was therefore exact.

It's not exact - and I found that is drifts from pano to pano.  This might 
be because it's a zoom and even though it has a lock at 18mm it probably 
still moves a little.  Or, for all I know, the camera/lens are just 
inaccurate in reporting the actual FOV.  I find that if I use a FOV in the 
range of 18.4mm - 18.6mm I can easily dial the Optimization error down below 
1 or 2 pretty easy.  None the less, I found that if I dial in the FOV that I 
am able in some cases to get pretty much perfect stitches with no CPs!

Like Kay, I'm doing most of these outdoors where mother nature helps hide 
little parallax sins.  Slightly more common is the case where Hugin just 
needs a little help.  I have had 100% success just auto generating CPs in 
the middle (horizon) row of 12 images.  The generator places about 200-400 
CPs in these 12 images and I do the regular routine of removing the outliers 
in a series of Optimizations, Fine Tunings and removing of any CPs in the 
sky or near-field.  This takes 3-5 minutes total and I'm done!

With this new recipe I'm getting great stitches and only investing about 5 
to 10 minutes per pano (excluding computer processing time).  I'm pretty 
happy with this.

Calvin

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

Re: [hugin-ptx] Hugin Session Recipes

[Calvin McDonald]

Thanks Carlos for sharing your experiences - exactly what I wanted to hear.

My best recipe so far, at least for most images, is to manually place 3 CPs 
on each image overlap.  
My process(recipe) is:
 - Load the images (usually around 38 of them).
 - Place 3 CPs manually at each overlap.
 - Optimize CPs and watch the CP list for outliers and mistakes in 
placement.
 - When Optimization is the best I can do, I Align.
 - I next do a Fast Preview and correct perspective and level the horizon 
(if needed).
 - Stitch
 - View in detail as a spherical and zoom in to look for artifacts.  I 
usually find some the 
   first and second pass and go back in and add more manual CPs in trouble 
areas.
 - Re-Align, Re-Stitch.
 - Done.

The above recipe has two significant problems.  1.  Very time consuming.  
2.  Becomes very difficult to place CPs manually when there are large 
featureless areas like blue sky.

I have found that if I use this recipe I usually get a good result needing 
little or no post image editing (GIMP/PS) and it appears to be very much 
independent of moderate inaccuracies in the input parameters like head 
adjustment, lens distortion data, minor misplacement of nodal point, 
levelness of the camera, etc.

I recently bought a NN4, a very nice sturdy tripod, extracted my lens 
distortion data, loaded starting templates for image position and very 
carefully tuned in the nodal point and carefully leveled the NN4.  With all 
this, Hugin still fails to give an acceptable result unless I spend the 2-3 
hours and manually place 3 CPs/overlap.

Anyone else out there want to add their experience and recipes?

Calvin

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

[hugin-ptx] Hugin Session Recipes

[Calvin McDonald]

I typically stitch 360x180 panos starting with between 30-60 images/pano.  
Usually:
  - Outdoor open sky scenes
  - Most commonly: 3 rows of 12 images + 1Z + 1N
  - Images shot RAW then converted (WB and LCA correction) to TIFF with 
NikonCapture

Can I get forum members to comment on their most successful Hugin session 
recipes for this type of work?
Things like use of templates, lens distortion parameters, number and 
location of CPs, use of auto CP generators, etc.
I very seldom get usable results using auto CP generation - almost always 
end up doing manual CPs.

Via trail-and-error, I've come up with a couple of recipes that eventually 
generate quality results (no or very little image artifacts) but are very 
time consuming - something like 2-3 hours per pano.

Is it reasonable to expect quality results in less time?
If so, how?

Thanks
Calvin

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

[hugin-ptx] Hugin Session Recipes

[Calvin McDonald]

I typically stitch 360x180 panos starting with between 30-60 images/pano.  
Usually:
  - Outdoor open sky scenes
  - Most commonly: 3 rows of 12 images + 1Z + 1N
  - Images shot RAW then converted (WB and LCA correction) to TIFF with 
NikonCapture

Can I get forum members to comment on their most successful Hugin session 
recipes for this type of work?
Things like use of templates, lens distortion parameters, number and 
location of CPs, use of auto CP generators, etc.
I very seldom get usable results using auto CP generation - almost always 
end up doing manual CPs.

Via trail-and-error, I've come up with a couple of recipes that eventually 
generate quality results (no or very little image artifacts) but are very 
time consuming - something like 2-3 hours per pano.

Is it reasonable to expect quality results in less time?
If so, how?

Thanks
Calvin

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

Re: [hugin-ptx] No Control Points?

[Calvin McDonald]

Thanks for the responses everyone - very helpful.  Admittedly my primary 
problem is inexperience, but that's why I'm here asking questions.

To date I have achieved my best (meaning acceptable) results manually 
placing 3 CPs between each image boundary.  Given I shoot with an 18mm lens 
this makes for a rather time consuming process.  I prefer image quality and 
deeper resolution so if I were to move to a different lens I might even go 
to something with a narrower FOV.  In other words, using a wider lens isn't 
a solution for me for much of my work.

Again, I'm inexperienced so don't claim I've come close to exhausting all 
the ticks/techniques that might be "in the book", but I very seldom end up 
with anything close to an acceptable pano using automatically generated 
CPs.  I've only tried CPfind and autopano-sift-c.  It is very likely this is 
due to not having yet learned the process and ticks of using a CP generator.

The bottom line is I'm growing tired of the time it takes to place CPs 
manually.  I blamed my difficulty getting auto generated CPs to work on my 
rather cheap tripod/head and not bothering to characterize and load my lens 
distortion parameters.  Over the past week or two I upgraded my tripod to a 
very nice heavy sturdy model, purchased a NN4 pano head and spent the time 
to accurately get my lens distortion parameters.  I expected at this point 
to get reasonable results without any CPs.  I just made my first attempt 
with the new equipment/method and was disappointed with a not horrible but 
unacceptable result.

I then tried auto generating CPs (on the same image set) - no success - in 
fact, got almost an identical result to no CPs.  My technique was to auto 
generate over the entire 38 image set, then remove the CP in the sky - then 
stitch (with no Align). My images tend to have a lot of open sky area and 
distinct lines below the horizon.  Moving clouds are also common.  (the test 
image I refer to is a open cloudy/overcast sky)

My quality expectation is being able to zoom in on the completed pano and 
not see any obvious stitching artifacts.

Maybe I should be asking the question of the forum
What is a typical Hugin session process and time duration for a 38 image 
set?
and
What does a reasonable auto CP method look like?

Thanks again for sharing your experience
Calvin

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

[hugin-ptx] No Control Points?

[Calvin McDonald]

I need someone experienced to check my expectations.

If I ...
  )  Carefully and robustly (as per these 
instructions) 
obtain distortion parameters a, b and c for my lens.
  )  Provide these distortion parameters to Hugin (Load Lens).
  )  Set my camera/lens/head to the exact nodal point calibration.
  )  Use a high precision pano head with very accurate detents both 
vertically and horizontally (w/ proper overlap).
  )  Apply image yaw/pitch/roll consistent with above and load into Hugin as 
a template.

Can I expect to get successful stitches from Hugin without any control 
points?

Calvin

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

[hugin-ptx] Re: strahov gigapixel

[Calvin McDonald]

If so, it's wrong both numerically and alphabetically - unacceptable!
I say we commission Jeffrey to go in and fix it with Photoshop.  ;)

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

[hugin-ptx] Re: Lens Profiles

[Calvin McDonald]

Thanks Kay.

> go to file-preferences-control point detectors 
Done.

> pick autopano-sift-c
There is a "autopano-sift-c" and a "autopano-sift-c (multirow/stacked)" to 
pick from.
I tried both - results below

> to the parameters there, add --maxdim  where  is your largest 
image dimension
Done

> On a set of well-overlapping (50% is good) images
Done

> Detect control points 
If I pick "autopano-sift-c (multirow/stacked)" I get an error message about 
not finding generatekeys.exe (which I don't have and can't find on the net).
I'm using a 56 count image set that is multirow and stacked so guessed this 
was the right selection.
If I pick "autopano-sift-c" it marches through finding CPs for each image 
then gives me a "Can not execute" error message when it calls 
autopano-sift-c.exe.  I can run autopano-sift-c.exe manually and it give me 
the help info so I know I have it (in hugin/bin) and it will execute.

Which autopano should I be using and any ideas how to get past the issue 
with each?

I'm using Hugin 2010.4.0.854952d82c8f on Win7

Calvin

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

Re: [hugin-ptx] Lens Profiles

[Calvin McDonald]

Thanks Jim - that helps!

If you would, check my understanding (remember, I'm a bit new to this).

Presently I don't load a lens profile.  Hugin gets a bit of a headstart 
getting a close FOV from the EXIF info but it (and me) have to struggle 
though a stitch partially because the distortion data for my lens isn't 
known.  We (Hugin and I) eventually (but not always) get a satisfactory (but 
never perfect) result.  This takes longer than I would like it to.

My thinking is that I can improve my stitch times and results if I can load 
an appropriate/accurate lens profile.  I just recently discovered that my 
lens (Nikkor 18-200mm VR-II used locked at 18mm) has particularly poor 
distortion characteristics.  I realize there are better lenses for the job 
but until my ship comes in, that's what I'm using.  ;)

Am I on base so far?

I'm trying to first determine if my thinking (above) is correct and if so, 
where/how to get/create a lens profile for my Nikkor 18-200 and Nikon D200.  
Given the lens and camera are fairly common, shouldn't I be able to find the 
a,b,c parameters published?  I'd still need d,e.  Are d,e different for 
every D200, or, the same for every D200 (holding the lens constant)?  If d,e 
are constant for all D200/18-200 then it appears a,b,c,d,e are just 5 
possibly known numbers that if I can get my hands on, I just type them into 
Hugin and save a lens profile, no?

I didn't quite follow what the process is to get a lens profile if I 
generate it with Hugin.  This also makes me a bit nervous because I assume 
the Hugin-generated lens profile numbers are not perfect and are likely 
coming from a stitch that's not perfect.  Doesn't this spread the 
imperfection to the next pano that uses the lens profile?

Can you comment in just a little more in detail what my best option is - or 
provide a reference? 
If I need to generate one with Hugin, how is that done?

Thanks
Calvin

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx

[hugin-ptx] Lens Profiles

[Calvin McDonald]

Please excuse the newcomer questions...  about distortion and lens models 
for Hugin.

1.  Are the parameters a,b,c,d,e,g,t different for EVERY lens, or, only 
different for each lens brand/model?

2.  If the latter, where does one get the lens parameter file to read into 
Hugin?

3.  If the former, how does one go about getting the parameters for their 
particular lens?

4.  Are these parameters stored in the lens itself (Nikkor DX) and presented 
in the EXIF or other meta data?

5.  If these parameters are not entered but enough CPs are use to produce a 
good stitch, will then Hugin know what these parameters are and be able to 
write out a lens profile?

Thanks
Calvin

-- 
You received this message because you are subscribed to the Google Groups 
"Hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to hugin-ptx@googlegroups.com
To unsubscribe from this group, send email to 
hugin-ptx+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/hugin-ptx