search

VITA-Group / LightGaussian

"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
506 stars 41 forks source link

example ckpt for room scene #14

Closed cfchi closed 5 months ago

cfchi commented 6 months ago

Hello, I would like to inquire whether the example checkpoint you provided for the room scene has undergone pruning, SH distillation, and Vectree Quantization processes. I noticed that the point cloud size in the file is 77MB, but Table 2 in the paper indicates that the model size after Vectree Quantization is 20MB. Could you kindly clarify if there might be a misunderstanding on my part? Thank you for your assistance.

Kevin-2017 commented 5 months ago

Hello! Thank you for interest in our work! To clarify the extreme_saving folder is the result of the Vectree Quantization process, and its size is approximately 20MB. This quantized version can be rendered using the command python render_video.py --load_vq -m <model_path>

However, it's important to note that current interactive viewer does not support this quantized format directly. To accommodate this, we have implemented a function that converts the quantized data back into the .ply format. This conversion results in the 77MB file size that you observed.

yangqing-yq commented 3 weeks ago

python render_video.py --load_vq -m but it show error "no cfg_args found in folder"