Tone

For the problem that all output frames are the same, I suggest that you can read the paper and run provided code to understand the working principle of IFRNet, then...

I have not realized this phenomenon. What is your hardware device? I use the standard PyTorch implementation of IFRNet and test it on the NVIDIA Tesla V100 and NVIDIA RTX...

>Does this model benefit from the use of VRAM or Core Frequency? Because I usually see that the GPU Core is at 0%, but the VRAM does heat up with...

For Middlebury dataset, we zero pad each input frame pair to spatial size of 640 x 480, and here is the key implementation code. ```python I0_pad = torch.zeros([1, 3, 480,...

The input size of IFRNet should be divided by 16. You can first pad the input frames, and then unpad the output frame. You can refer to [benchmarks/SNU_FILM.py](https://github.com/ltkong218/IFRNet/blob/dfe18113ee2e345aab550aec2ab832259964dfe1/benchmarks/SNU_FILM.py). Thanks.

Thanks for your interest. Our pretrained models can already get relative good frame interpolation visual quality on common videos. To get best visual quality on your specific videos, you can...

The input size of IFRNet should be divided by 16. You can first pad the input frames, and then unpad the output frame. ```python class InputPadder: """ Pads images such...

For 8x frame interpolation, we follow the [FLAVR](https://github.com/tarun005/FLAVR) project. It seems that [FLAVR](https://github.com/tarun005/FLAVR) does not provide the Adobe240 evaluation dataset, so we select a subset of all Adobe240 video sequences...

Please refer to the code.

Is PSNR of IFRNet_S equals to 34.45 or 35.46 according to your experiment?