Max Zhaoshuo Li 李赵硕

Hi @Adolfhill If everything is set up correctly and you have increased the batch size, it is likely that this capture is quite hard to reconstruct for the current implementation....

Hi @Adolfhill Neuralangelo and most NeRF methods will require dense camera coverage similar to taking photos of an object in a hemisphere. In your case, walking circularly around the scene...

Hi @ZiyangYan Thank you for your interest in the project! The current dataloader only supports single camera. However, it is possible to extend this to multiple cameras. If you look...

Hi @FrankSpalteholz I would extend the current implementation to support multi-cam instead of assuming that they all have the same intrinsics :) I haven't tested something like yours, but I...

The suggestion of @HanakoAA is correct.

Have you tried to use [this script](https://github.com/NVlabs/neuralangelo/blob/main/projects/neuralangelo/scripts/visualize_mesh.ipynb) for visualization?

Interesting. I have requested access to your files. Once I have access, I can take a look.

Hi @Futurer90 It does look like your checkpoint failed to load the color network. The last layer weights of the rgb network are all zero. What checkpoint did you use?

Hi @Futurer90 Oh I mean which commit of the branch you are on. Not the command nor checkpoint.

Hi @kf-zhang The [TNT datasets provide an official code for evaluation](https://github.com/isl-org/TanksAndTemples/tree/master/python_toolbox/evaluation). We used the [python code for DTU evalution](https://github.com/fdarmon/NeuralWarp/blob/main/eval.py) borrowed from NeuralWarp.