The model is getting trained but I am facing some challenges towards the end at the video rending part.I think somehow the end output is having a zoom-in effect which I am unable to understand why? Can anyone please help me to understand where exactly the problem lies and the possible solution around it ?
I am using Neural Radiance filed's origional Git repository to run 'tiny_nerf.ipynb' colab which gives 360 degree rotating output of a 3D render. Input for this is generated from colmap which is in .npz format containing details about images, poses amd focal length of the images.