Heya.. Was going through this repo and was interested in running DepthSplat.. I was able to run it for the provided sample but I saw that the code reads the values of camera intrinsics ad pose from the inferece_dir but what if I don't have that information to begin with? What if all I have is a sequence of images or frames extracted by let's say maybe a video. Isn't there a way to point at just the images rather than pointing to them with the intrinsic and pose information as well? Thanks : )
Heya.. Was going through this repo and was interested in running DepthSplat.. I was able to run it for the provided sample but I saw that the code reads the values of camera intrinsics ad pose from the
inferece_dirbut what if I don't have that information to begin with? What if all I have is a sequence of images or frames extracted by let's say maybe a video. Isn't there a way to point at just the images rather than pointing to them with the intrinsic and pose information as well? Thanks : )