nerfstudio VS CIPS-3D

You’re under the right paper for doing this. Instead of one big model, they have several smaller ones for regions in the scene. This way rendering is fast for large scenes.

This is similar to Block-NeRF [0], in their project page they show some videos of what you’re asking.

As for an easy way of doing this, nothing out-of-the-box. You can keep an eye on nerfstudio [1], and if you feel brave you could implement this paper and make a PR!

[0] https://waymo.com/intl/es/research/block-nerf/

[1] https://github.com/nerfstudio-project/nerfstudio

Try rendering NERFs with your footage, you're gonna love the result and NERFs are pretty robust to reflections. You can use your Metashape solve for Nerf Studio https://github.com/nerfstudio-project/nerfstudio

Beyond these tips, I don't have much. There's lots of research about how to improve quality of solves in the software itself. I'm hoping these get added to instant-ngp, since it's fast and free, but it is research software, not a product, so we'll see. Another thing to maybe look at is Nerfstudio. It can use instant-ngp as a solver, but there are other solvers. I briefly tried it but couldn't figure out how it worked, from the small bit of time I spent with it. I hope to get back to it.

nerf.studio | github | discord

Code for https://arxiv.org/abs/2110.09788 found: https://github.com/PeterouZh/CIPS-3D

Researchers at Huawei and Shanghai Jiao Tong University have developed CIPS-3D, an approach that synthesizes each pixel value independently, just as its 2D version did.