search

nianticlabs / acezero

[ECCV 2024 - Oral] ACE0 is a learning-based structure-from-motion approach that estimates camera parameters of sets of images by learning a multi-view consistent, implicit scene representation.
https://nianticlabs.github.io/acezero/
Other
651 stars 41 forks source link

Multiple cameras , intrinsics? #3

Closed antithing closed 3 months ago

antithing commented 3 months ago

Hi, and thanks for making this code available!

Could it run on datasets created with multiple cameras, in the same scene?

Does it also calculate intrinsic data per camera? What format is used?

Thanks!

ebrach commented 3 months ago

Hi! You can use ACE0 if all cameras share intrinsics, as ACE0 only estimates a single focal length value shared by all images. I think, estimating intrinsics per image should work in principle, but needs some changes to the code. See also our FAQ.