Firstly, the script on the project homepage will throw an error when run, as the parameters are not quite right.
Secondly, could you please support the reconstruction of any frame? For example, if I input 150 frames at once, it will report an Out Of Memory (OOM) error, stating that 190GB of VRAM is needed.
Looking at the code, during video reconstruction, there is no chunking, and all frames are encoded and then decoded at once, which our VRAM cannot handle.
Could the author please explain how the 24-second, 7-frame high-definition large video of 20*1280 resolution mentioned on the homepage was reconstructed?
Firstly, the script on the project homepage will throw an error when run, as the parameters are not quite right.
Secondly, could you please support the reconstruction of any frame? For example, if I input 150 frames at once, it will report an Out Of Memory (OOM) error, stating that 190GB of VRAM is needed.
Looking at the code, during video reconstruction, there is no chunking, and all frames are encoded and then decoded at once, which our VRAM cannot handle.
Could the author please explain how the 24-second, 7-frame high-definition large video of 20*1280 resolution mentioned on the homepage was reconstructed?