An error is reported in the video input

[txy00001]

Hello, I would like to ask, can't a new video be directly input and then generate a new clear video?Does it have to be framed?1.I use restormer and get an error "assert inp_img.shape[1] == 1, "Input images should contain only 1 frame." AssertionError: Input images should contain only 1 frame.” 2. "AttributeError: 'EasyDict' object has no attribute 'model'" is also reported with dbn

[ljzycmd]

Hi @txy00001, the video should be first split into frames. Meanwhile, Restormer is an image model, thus can only restore 1 frame when inference once.