Thanks for your released code!
Could you please provide me which frames are you utilized in your experiments on the Scannet dataset? I'm also developing a similar approach for novel view synthesis from unknown cameras and want to compare with the Nope-NeRF on ScanNet.
Thanks!