sniklaus / 3d-ken-burns

an implementation of 3D Ken Burns Effect from a Single Image using PyTorch
Other
1.52k stars 225 forks source link

Strategy of generating the camera intrinsic #72

Open Binyr opened 1 month ago

Binyr commented 1 month ago

Hi, thanks for your great work!

I found that each sample in your dataset has various field-of-view. Are these field-of-view generated completely randomly?

Is there any strategy that I can get a set of temperally continuous image with constant intrinsic?

Thank you, Yanrui

sniklaus commented 1 month ago

Thank you for your interest in our research! Yes, the field of view varies, and it is indeed completely randomized. Since we target single image inputs, we did not design the dataset to provide useful feedback for video depth estimation for example, sorry!