0

So, at this point, I'm not sure if this is a programming or a physics problem! I'm attempting to concatinate 2 video inputs together to cover a wider field of view. As far as I can tell, this is usually done by "stitching" using matching keypoints in the two images. This can be computationally expensive, and as I'm trying to show the output in real time, my approach has been a simple mechanical allignment and a pixel overlap, blending/merging the overlap together.

Here's the problem. Usually stitching is done with 2 cameras placed next to each other. My cameras are places 3 meters apart. Two different point of views resulting in two different perspectives.

Now, the stitching doesn't have to be absolutely perfect. My goal is to concatinate the inputs while NOT losing objects. Obviously, depending on the field of view of the cameras, there will be a "dead space" close to the cameras.

Here's my starting point:

Starting point

One idea was to do a cut with an angle and then stitch the images together. That way, I keep information close by while avoiding double objects further out. But this introduces an unwanted distortion and I still get double objects further out:

angle cut & stitched

Any ideas how to move forward? Any help appriciated!

Improve this question
2
  • You cannot just cut at an angle - every point will have a transformation required in order to make it fit. See the old autostitch programme for guidance and info - it's a fair bit more complex than you think, unfortunately
    – Dr Mayhem ♦
    Dec 8, 2022 at 11:30
  • Autostitch (by Matthew Brown?) looks pretty impressive! I'll have a look into that. At first glance it does however look like the approach is similar to most other sources I've found: cameras are placed close to each other. But yes, I'm afraid this whole project is more complicated than I thought at first. Regardless, thanks for your reply.
    – user39965
    Dec 9, 2022 at 8:33

1 Answer 1

Reset to default
2

There's a pretty good explanation of how to do this fro polyhaven (who make lots of HDRI images which require stitching of multiple shots): https://blog.polyhaven.com/how-to-create-high-quality-hdri/

They recommend using PTGui to stitch the images together https://ptgui.com/

To do this for a video you'd have to convert to an image sequence, then script something to sequentially stitch all your files. it would take a while.

I'd try mesh warping them in After Effects and carefully masking between them. You might just about get away with it.

Improve this answer

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.