Shengyin Jiang
Shengyin Jiang
> Thanks for reply. My nvs dataset actually assumes a strictly rigid scene. So in your opinion, it's better to finetinue VGGT based on camera_head+depth_head's output, instead of point_head's output?...