Question about the pretrained models

zju3dv / 4K4D

[CVPR 2024] 4K4D: Real-Time 4D View Synthesis at 4K Resolution

https://zju3dv.github.io/4k4d/

Other

1.59k stars 69 forks source link

Question about the pretrained models #45

Closed brianneoberson closed 3 months ago

brianneoberson commented 3 months ago

Hello,

I was wondering if the pretrained models were trained on the training views only or did you train on the whole dataset? (so including the 4 test views)

Best regards

dendenxu commented 3 months ago

The released models are trained on full-resolution images of all views for the best visual quality for the release version of the code. For evaluation (matching the view selection and resolution selection used in the paper), please consider downloading the reference images. Note that images for the paper are produced using the training views only.