Page 1 of 1

APG 3.5 Control Points Editor lose control Points

PostPosted: Fri Apr 04, 2014 11:13 pm
by Abulafia
Hi Team,
hi All,

after using the Control Points Editor & optimize the panorama all control points are lost.
This bug is since APG 3.5 RC1 and not fixed in the final version. Sorry guys.

With kind regards,
Stefan

Re: APG 3.5 Control Points Editor lose control Points

PostPosted: Fri Apr 04, 2014 11:24 pm
by mediavets
Abulafia wrote:Hi Team,
hi All,

after using the Control Points Editor & optimize the panorama all control points are lost.
This bug is since APG 3.5 RC1 and not fixed in the final version. Sorry guys.

With kind regards,
Stefan

I don't experience that with APG 3.5 final 64-bit on Windows 7 Pro 64-bit.

Which OS are you using? APG 3.5 32-bit or 64-bit?

Re: APG 3.5 Control Points Editor lose control Points

PostPosted: Sat Apr 05, 2014 11:35 am
by Abulafia
Hi All,

may I should explain it a little bit more in detail.

I'm using APG 3.5 on Windows 7 Pro 64bit.
I try it on my PC and laptop. Both with nVidia graphic chip.

I import all the images via Papywizard import, 50 images with 5 per stack.
I start control points editor and remove for example some wrong points. After new optimazion, the the RMS values are all over 50 and higher. The image look like that all images are stacked to one stack. I experence this also with RC1, but not with beta1, beta2 & beta3.

Hope it helps.

Regards,
Stefan

Re: APG 3.5 Control Points Editor lose control Points

PostPosted: Sat Apr 05, 2014 11:47 am
by mediavets
Abulafia wrote:Hi All,

may I should explain it a little bit more in detail.

I'm using APG 3.5 on Windows 7 Pro 64bit.
I try it on my PC and laptop. Both with nVidia graphic chip.

I import all the images via Papywizard import, 50 images with 5 per stack.

So what robotic pano head are you using?

Are you using AEB to bracket exposures or a Promote Control?

Are your stacks hard linked?

I start control points editor and remove for example some wrong points. After new optimazion, the the RMS values are all over 50 and higher. The image look like that all images are stacked to one stack. I experence this also with RC1, but not with beta1, beta2 & beta3.

Hope it helps.

Regards,
Stefan

The general consensus remains that it is best to process bracketed exposures in thid party software and then to stitch the processed images in APP/APG.

Re: APG 3.5 Control Points Editor lose control Points

PostPosted: Sun Apr 06, 2014 10:10 am
by Abulafia
OK. Here is my workflow:
1. Press Import...
2. Select Papywizard
3. Select XML-file
4. Select all images (150 Nikon NEF files, 3 rows, 10 stops, 5 per stack)
5. Select "Use hardlinks" and "reference level" (in german "Bezugspegel")
6. Leave all settings to default
7. Press "OK"
8. After stiching, start Control point editor. At this point the highest RMS value is 4.46, avg. RMS is 2.76
9. Remove some controls points
10. Press "Optimize Panorama"
11. After that the highest RMS value is 67.69, the lowerst is 51.55, the avg. RMS is 61.06
12. All images are stacked one above the other

Hope it helps.

Re: APG 3.5 Control Points Editor lose control Points

PostPosted: Sun Apr 06, 2014 12:53 pm
by mediavets
I've never tried editing control points in hard linked stacks - is that even a viable workflow?

If your highest RMS value is only 4.46 - and the average RMS only 2.76 - is there any need to edit CPs at all?
...............

I believe the consensus is still to process RAW files and/or bracketed exposures with third party software before stitching for optimum control and top quality results.

Re: APG 3.5 Control Points Editor lose control Points

PostPosted: Sun Apr 06, 2014 2:23 pm
by Abulafia
Thanks for helping, but sorry this is only an expample to show the problem. I don't want to disccus the workflow by it self. If you want we can do this in an extra tread, I'm open to do that. I have done meanwhile over 400 panorames in that way and I'm using APG since version 2.0. And yes there is a need to do that (in this case), because in the image there are some trees and bushes and with beta3 I was able to optimize the overlapping of them. With APG 3.5 final the panorama is destroyed. And by the way this happens also with 1 image per stack.

Thanks again for helping and with kind regards,
Stefan

Re: APG 3.5 Control Points Editor lose control Points

PostPosted: Sun Apr 06, 2014 2:35 pm
by mediavets
Abulafia wrote:And by the way this happens also with 1 image per stack.

Stefan


I spent a long time yesterday on a difficult image set - 36 images - editing Control Points and I've not seen this with single exposures; Windows 8.1 or 7 64-bit.

Re: APG 3.5 Control Points Editor lose control Points

PostPosted: Sun Apr 06, 2014 7:17 pm
by Abulafia
Many thanks for your reply. Can you reproduce it with 3 or 5 per stack?
I tryed today a new one with 3 per stack and 10 stops and after removing 1 control point & optimization. The panorama was destroid.
I'll check again a new one with 10 stops and 1 per stack to see what's happen then.
I keep you updated.

Re: APG 3.5 Control Points Editor lose control Points

PostPosted: Sun Apr 06, 2014 10:01 pm
by Abulafia
I tried a panorama with 10 images with 1 image per stack. After optimization, the panarama is good. So in fact no problems with 1 per stack, but when I use my 10 stops and 5 per stack panorama and deselect 4 images to leave 1 image per stack. The panorama is destroid after optimization.
I think we could finalize that the problem occures only with more then one image per stack.

Hope it helps.

Re: APG 3.5 Control Points Editor lose control Points

PostPosted: Mon Apr 07, 2014 9:01 am
by ThomasV
Hello Abulafia,

Thanks for reporting this bug. We sucessfully reproduce this issue. Let me explain the cause of the issue.
When importing a papywiazrd case with hard links, we assume that all images in a stack have the same position, then in the new optimizer engine, the position (yaw, pitch and roll) has its scope set to 'group', as images of a stack are in a same group.
You can learn more on that topic on this documentation page (http://www.autopano.net/wiki-en/action/view/Optimizer_engine).

The bug you are describing is that the groups are lost after the first optimization. When you open the control point editor, you can see in the layers editor that all images are in the same group. This bug will be fixed in Autopano 3.5.1.

The best solution to work around the issue is to disable grouping in the group's optimizer settings (http://www.autopano.net/wiki-en/action/view/Autopano_-_Preferences..._-_Optimization).
- Change optimization scopes to 'Manual'
- Set grouping criteria to 'No grouping'

Then, do the detection. (So you had to disable automatic detection in papywizard plugin).

Regards,
Thomas

Re: APG 3.5 Control Points Editor lose control Points

PostPosted: Mon Apr 07, 2014 7:03 pm
by Abulafia
Hi Thoms,

many, many thanks for helping. Looking forward to APG 3.5.1.

By the way I also report an issue with Nikon DF raw files.

With best regards,
Stefan

Re: APG 3.5 Control Points Editor lose control Points

PostPosted: Tue Apr 08, 2014 7:47 am
by ThomasV
Hello,

I saw the thread of Nikon Df file, and it would be good to have a sample file. (http://www.kolor.com/forum/viewtopic.php?t=22046&p=140152#p140152).

Regards,
Thomas