stitching issues with different cameras/focal lengths  

Share your tips and tricks here or get help with any Autopano Pro / Giga problem!
No bug reports (of any kind) in this forum!
no avatar
waterproofmonk
New member
 
Topic author
Posts: 7
Likes: 1 post
Liked in: 2 posts
Joined: Fri Jul 14, 2017 4:47 pm
Info

stitching issues with different cameras/focal lengths

by waterproofmonk » Sun Jan 21, 2018 2:02 am

I've been using autopano pro for a while now but still struggling with certain cases.
I often take a bunch of close-up shots with one camera/lens (long focal length), and then take a shorter-focal-length "overview" shot to give myself more to work with when stitching/composing the image afterwards. The idea is that this will help create a bigger crop if I want it, as well as fill in areas where my long-focal-length shots end up blurry or something else goes wrong with them.

However, Autopano often chokes on that and does a terrible job of stitching. There are 2 issues:
1. "auto control points" does not work. I've attached an example of images which clearly match but where I can't get Autopano to create any automatic control points no matter what areas I select with the "auto control points" tool:
manual control points.PNG
Example of images where automatic control point selection does not work


2. OK, so I have to create control points manually. Despite my manual control point selection, when I "optimize" the panorama or just the image, Autopano moves the image to a completely wacky spot that makes no sense. This so-called "optimization" always increases the RMS error value compared with the values I get by moving the image manually.
For example, in this case the RMS of my 2 manual control point links are around 5 and 10 beforehand:
before optimization.PNG
Example of images where automatic control point

After I "local optimize this image", they are 61 and 143!
after optimization.PNG
Example of images where automatic control point


Why is this happening? Is there anything I can do to get Autopano to put this image in the right place?
Thanks in advance for your help!

User avatar
olivierbo
Kolor Team
 
Posts: 2225
Likes: 1 post
Liked in: 133 posts
Joined: Thu Nov 06, 2014 11:44 am
Info

Re: stitching issues with different cameras/focal lengths

by olivierbo » Mon Jan 22, 2018 11:39 am

Hello,

Sometimes Autopano fails to find some control points depending on the picture content: too much overlap, a lack of details, homogeneous areas, etc.
In that case, you can try to increase the quality of detection and filtering model maybe.
Also, in case you have a variation of the nodal point, try viewpoint option in optimization tab (Giga only).

Let me know if it helps.

Best regards,
Olivier

no avatar
waterproofmonk
New member
 
Topic author
Posts: 7
Likes: 1 post
Liked in: 2 posts
Joined: Fri Jul 14, 2017 4:47 pm
Info

Re: stitching issues with different cameras/focal lengths

by waterproofmonk » Tue Jan 23, 2018 1:58 am

Thanks for your response Olivier. I don't think it solves my problem, though.

Sometimes Autopano fails to find some control points depending on the picture content: too much overlap, a lack of details, homogeneous areas, etc.

Maybe you could explain why having too much overlap would prevent creating control points? The images in this case have plenty of detail so the last should not be an issue. You should be able to clearly see in the screenshot that I provided that there is plenty of detail - lots of edges, texture, etc.

you can try to increase the quality of detection and filtering model maybe.

I'm not sure what you mean by this.

Anyway, my main problem, numbered 2 above, is that even when I manually enter control points, the "optimization" does a terrible job of placing the image relative to the others in the panorama. Do you have any explanation for why that would happen? Isn't the basic idea of the optimization that the RMS values should be minimized?

no avatar
waterproofmonk
New member
 
Topic author
Posts: 7
Likes: 1 post
Liked in: 2 posts
Joined: Fri Jul 14, 2017 4:47 pm
Info

Re: stitching issues with different cameras/focal lengths

by waterproofmonk » Tue Jan 23, 2018 2:35 am

Interestingly, I just ran a test where I created a new panorama with only 2 images - the same ones that fail to find control points above. When I create a panorama that has only those two images, plenty of control points are found and optimization works just fine. Is there an explanation for why two images find perfect alignment when alone, but in a larger pano with many more images, they fail?


Who is online

Users browsing this forum: No registered users and 3 guests

cron