janusch

Results 97 comments of janusch

Should not be hard at all. Of course, depending on how wild you go with the UI.

Hey, no current plans at the moment as I am too busy. However, maybe the opencv model. That one should be easy

We should support it. For reference, https://github.com/nerfstudio-project/nerfstudio/blob/main/nerfstudio/process_data/colmap_utils.py#L187-L387

You could pass in just the sphere with the other gaussians. The sphere just needs the gaussians to be at opacity 1. However, you can see from the depth video...

I missed that. Anyway, thank you! I guess, that implementation has been surpassed meanwhile by faster ones :)

I have everything upgrade to the latest torch version 2.7

This might be relevant: https://github.com/InternLandMark/FlashGS https://arxiv.org/abs/2408.07967 https://github.com/humansensinglab/taming-3dgs

You should use cuda 12.8 I don't know what your issue is. I am on Ubuntu 22.04 and it works perfectly fine after g++14 is installed and made default. Anyway,...