nerf-pytorch VS OpenMVG (open Multiple View Geometry)

Found relevant code at https://github.com/yenchenlin/nerf-pytorch + all code implementations here

I have been curious about NeRF and have tried out the software. I really like the idea, but I have found generating the NeRF representation to be too slow and does not do seem to do too well if the number of cameras in the rig are a bit sparse. I have seen that there have been some evolution and variations of the NeRF concept and I am looking into some of them currently.

While there are other PyTorch implementations out there (e.g., this one and this one), I personally found them somewhat difficult to follow, so I decided to do a complete rewrite of NeRF myself. I tried to stay as close to the authors' text as possible, and I added comments in the code referring back to the relevant sections/equations in the paper. The final result is a tight 374 lines of heavily commented code (320 sloc—"source lines of code"—on GitHub) all contained in a single file. For comparison, this PyTorch implementation has approximately 970 sloc spread across several files, while this PyTorch implementation has approximately 905 sloc. A Colab notebook for the full model can be found here.

Try to find a community that will know more about openMVG, like the discussion tab on their github.

OpenMVG (JSON, NVM, PLY)

I find openMVG very decent, FTXUI might be a good one and nlohmann's json library is also pretty nice. I don't really know of any project that strictly adheres to the core guidelines, except maybe for some of Jason Turner's (sample) projects.

I am looking for some pointers to papers/open-source software that can do fast and accurate depth estimation from multi-view videos. I am aware of colmap and OpenMVG software but was curious to know if there are others out there that outperform these. I am also interested in any state-of-the-art for depth estimation using unsupervised or self-supervised neural networks.

OpenMVG is super cool, and the project lead Pierre is very generous with his time.

Meshroom is just a Python wrapper for VSfM. OpenMVS seems like the package most customizers are starting to use. https://github.com/openMVG/openMVG