Hello,
I captured a 360 video using my Insta x4.
I extracted 360 (equirectangular) images every 1 second - total of 24 frames.
Each frame was then converted to a cubemaps - so now I have 144 2D images from this short video.
I tried to run alignment on this dataset on CapturingReality - but it doesn’t really seem to create any 3D model.
Any suggestion for the correct workflow from a 360 video to 3D model?
Thank you,
Amir
Hello Amir,
as RealityCapture doesn’t natively support 360 images, there is no a correct workflow.
Such images could be quite deformed (distorted) and therefore it could be problematic to align them.
Is there a good overlap between the images? What is the resolution of the images?
What you can try it to separate your dataset into 6 parts (one for each cube side) and try to align only that one part. If that is successful, you can export the components and them import them into new project. Then you can try to align. Also, it is possible you will need to use control points to merge such data.