OpenStitching / stitching

A Python package for fast and robust Image Stitching
Apache License 2.0
1.99k stars 156 forks source link

missing images when using SeamFinder #60

Open algopert opened 1 year ago

algopert commented 1 year ago

Hi Lukas. Hope you are well!

Another issue!

result with Seamfinder image

result without Seamfinder image

image

What's wrong?

lukasalexanderweber commented 1 year ago

I've never worked with so many images. Can you add two more images and see if there are two more holes? Maybe opencv only supports a limited number of imgs...

lukasalexanderweber commented 1 year ago

maybe it's also a problem with the transparency. Can you plot the result without seam finder with transparency as well?

algopert commented 1 year ago

I had added 14 more images. But the same result and there was only 1 or 2 rifts.

lukasalexanderweber commented 1 year ago

I hope you understand that it's really hard for me to say whats going wrong

algopert commented 1 year ago

Ok, I will find out what's wrong.

lukasalexanderweber commented 1 year ago

great thank you so much for your effort!

phamkhactu commented 8 months ago

Hi @algopert

I've encountered the same. You can try set:

seam_finder = SeamFinder(finder='no')
lukasalexanderweber commented 8 months ago

@phamkhactu could you share the images that lead to the black seams?

phamkhactu commented 8 months ago

@phamkhactu could you share the images that lead to the black seams?

@lukasalexanderweber here is my link data I've tried to stitch all image to create full 360 image

lukasalexanderweber commented 7 months ago

maybe you could add your images to https://github.com/OpenStitching/images so the error can be reproduced :)

claytonrothschild commented 3 months ago

@phamkhactu did you have luck stitching a full 360 image?

phamkhactu commented 3 months ago

@phamkhactu did you have luck stitching a full 360 image?

Hi @claytonrothschild

I have made full 360 image in good condition. However, I used stitching as a base and used cube360 to stitch top and bottom image.