instant-ngp VS tiny-cuda-nn

This is really cool! Could you now use something like this to turn the new images in a 3d model? Or even use open pose (controlnet) to generate a bunch of images from different angles and use InstantNeRF to make a 3d model for free!

It is at this point that you should have Instant-NGP setup. The script for the COLMAP processing is in the repo, as well as the steps to perform it. My exact parameters were 3 fps and 16 aabb. It is pretty helpful to add the scripts directory into path for exact access system-wide.

[Project Page] https://nvlabs.github.io/instant-ngp/

instant-ngp ([1]) from NVIDIA can render NeRF in VR in real-time, assuming a very good desktop video card. Note that instant-ngp is not as photo-realistic as Zip-NeRF. But it's still very good!

1. https://github.com/NVlabs/instant-ngp

Playing around with neRF AI (https://github.com/NVlabs/instant-ngp) to create some 3d gear reveals. I think this a fun way to show off a kit, what do you think?

3 not sure if it's the one you want but the -aabb_scale is a crop. This page recommends trying a large value of 128 for some outdoor scenes: https://github.com/NVlabs/instant-ngp/blob/master/docs/nerf_dataset_tips.md

I don't know about lumalabs, but basically all NeRF projects these days are based on NVIDIAs Instant neural graphics primitives ( GitHub: instant-ngp). It utilizes COLMAP for SfM (preprocessing step for the neural network) and runs on average Geforce cards pretty good. The fox example (50 photos) on their page literally takes 5 seconds to complete.

In the opposite direction from your question is a very interesting project, TinyNN all implemented as close to the metal as possible and very fast: https://github.com/NVlabs/tiny-cuda-nn

This repo only implemented the rendering part of the NGP but is more simple and has a lesser amount of code compared to the original (Instant-NGP and tiny-cuda-nn).

...a C++ library with a CUDA backend. But these high-performance building blocks might only be saturating the GPU fully if the data is large enough.

I haven't looked at implementing these things, but I imagine uf you have smaller networks and thus less data, the large building blocks may not be optimal. You may for example want to fuse some operations to reduce memory latency from repeated memory access.

In PyTorch world, there are approaches for small networks as well, there is https://github.com/NVlabs/tiny-cuda-nn - as far as I understand from the first link in the README, it makes clever use of the CUDA shared memory, which can hold all the weights of a tiny network (but not larger ones).