I had a really hard time installing all the dependencies and getting CUDA set up but I finally got an error message that I feel like is outside of my control:
File "D:\Repos\FLAVR\interpolate.py", line 132, in
idxs = torch.Tensor(range(len(videoTensor))).type(torch.long).view(1,-1).unfold(1,size=nbr_frame,step=1).squeeze(0)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^
RuntimeError: maximum size for tensor at dimension 1 is 3 but size is 4
I guess it's possible this is some kind of complaint about my input media. I'm using an mp4.
It's also possible this is due to me using newer versions of the dependencies than the original project called for. I don't know. I originally tried matching all the versions but I was getting a number of different errors going that way. I'm currently working with the following dependency versions:
torch 2.8.0+cu126
torchvision 0.23.0
tqdm 4.67.1
av 15.1.0
opencv-python 4.12.0.88
cuda 12.6
I'm on Windows 10 with a GeForce GTX 1060 6GB