YuelangX / Gaussian-Head-Avatar

[CVPR 2024] Official repository for "Gaussian Head Avatar: Ultra High-fidelity Head Avatar via Dynamic Gaussians"
Other
703 stars 40 forks source link

Confused about the preprocess_nersemble.py #18

Closed starcxl closed 2 months ago

starcxl commented 2 months ago

Hi, yuelang, many thanks for your awesome job!

May I ask the format of your "raw_NeRSemble"? Cause when I run "python preprocess_nersemble.py", the code here https://github.com/YuelangX/Gaussian-Head-Avatar/blob/2b798a8e60af9997c827fdb1d950725053980cee/preprocess/preprocess_nersemble.py#L58 is absence in my NeRSemble dataset. I am really confused about this step.

Looking forward to your reply. Thanks!

YuelangX commented 2 months ago

NeRSemble What does your raw_NeRSemble look like?

starcxl commented 2 months ago

Many thanks for your quick reply! My raw_NeRSemble looks like this: image

I am using the "participant_036.zip" and "camera_params.zip", and unzip them to get my raw_NeRSemble, but it seems all the names are not aligned. Could you please tell me which zip you using in your raw_NeRSemble?

Looking forward to your reply! Thanks!

starcxl commented 2 months ago

Ah, it seems that the datasets you are using is the "# View 3 (vertical): All sequences from all participants, grouped by sequence type" in the download_links_nersemble.txt. Am I right?

NathanHuanrongLIU commented 2 months ago

I have the same questions. Did you use View 1: 10 sequences from NeRSemble paper (NVS benchmark) in download_links_nersemble.txt or others? Could you give a dataset folder? Thx~~~

YuelangX commented 2 months ago

I just used "# View 3 (vertical): All sequences from all participants".

starcxl commented 2 months ago

Many thanks for your reply! I am clear now.