Open FantasticOven2 opened 4 days ago
Thanks for the question! It is applied to the entire image - this also helps the background "move" appropriately (i.e. in manner that reflects the inverse of the camera's motion).
Hey author,
I have a new question regarding camera poses estimation; did you try to visualize the learned camera trajectory and compare with the colmap estimated camera trajectory?
Thanks a lot
Hey author,
Thanks for this amazing work!
For iphone experiments without camera poses, is the tracking loss applied to the entire image or only the dynamic foreground part?