Suggestion:About Google's MultiNeRF

I heard that Google’s AI noise reduction algorithm MultiNeRF is open source on Github, but it seems to have only the algorithm, not the model. Users need to train the model themselves. This algorithm seems to work really well, can you add this algorithm in the next update? I think we can provide the data for training the model. (This is just a suggestion, I don’t know if it’s feasible.)Suggestion:About Google’s MultiNeRF

2 Likes

Is it Free Open Source Software? I doubt it.

Do you mean GitHub - google-research/multinerf: A Code Release for Mip-NeRF 360, Ref-NeRF, and RawNeRF ?

1 Like

Looking at that, it doesn’t appear to be a noise reduction algorithm, but more of a 3D reconstruction approach?

AFAICT, it’s all-in-one: can use multi-shot to reduce noise and reconstruct 3D, all in HDR…

2 Likes

thank you very much for the info: the approach is very interesting, but the noise reduction requires several photos of the same scene.As soon as I have some time I will try to test with my photos shooted for photogrammetry 3D (motion from structure)
thank again

Yes.

Oh,thanks for your reply!