search

VITA-Group / LightGaussian

[NeurIPS 2024 Spotlight]"LightGaussian: Unbounded 3D Gaussian Compression with 15x Reduction and 200+ FPS", Zhiwen Fan, Kevin Wang, Kairun Wen, Zehao Zhu, Dejia Xu, Zhangyang Wang
https://lightgaussian.github.io/
Other
608 stars 58 forks source link

about Octree-based Compression #21

Open cfchi opened 8 months ago

cfchi commented 8 months ago

Dear Authors,

I'm curious to know whether the Octree-based Compression G-PCC mentioned in the paper is included in this GitHub repository?

If it is, could you kindly direct me to the specific path where I can locate it?

If not, would it be possible for you to provide it in the future? Alternatively, could you offer some guidance on how to access or utilize it?

Thank you very much for your assistance.

Ericgone commented 8 months ago

hi, did you success in running the trainning process? how many ram and vram do we need for this project? thx

erenovic commented 5 months ago

Hello, I am interested in Octree-based compression of gaussian positions as well. As far as I saw, the position information is stored using np.savez_compressed which is not same as Octree-based compression. When I looked into the G-PCC / TMC3 compression, I realized that point locations are sorted (according to what, I am not sure about). Can you point out to further details about how to use the position compression with G-PCC?