search

nerlfield / wild-gaussian-splatting

DUSt3R + Gaussian Splatting
146 stars 12 forks source link

Doubt regarding camera behaviour #5

Closed mbortolon97 closed 5 days ago

mbortolon97 commented 2 weeks ago

Hi, Thanks for the fantastic work. I was playing with it a bit, and I saw that on scene/cameras.py:59, there is a multiplication per 10,000. But it seems strange to me. Would you happen to have an idea why?

Thanks Matteo

nerlfield commented 1 week ago

I've done it to overcome the issue of numerical instability. Firstly, there's no need for concern, as the values are normalized before being used in the render function.

The challenge arises because Dust3r outputs camera parameters in a very small range and there are values super close to 0, and these sometimes saturate to 0 after normalization before rendering. By multiplying them by 10000, the values increase and, post-normalization, they nearly reach but do not hit zero. Without this adjustment, artifacts appear in the rendered images. You may comment out this line; theoretically, it shouldn't affect the classical Gaussian splatting pipeline.

mbortolon97 commented 5 days ago

Thanks a lot for your explanation. I never noticed the issue; this effectively makes quite a few differences in my tests. Thanks again Matteo