Zhihang Zhong

Unfortunately, we only have normal-sized EMA-VFI checkpoints. https://drive.google.com/drive/folders/17p-hsSnkp40ck6tyt8x4V8erIVChGOeO?usp=sharing

In the training and inference stage, the model only supports 3 nearby images as inputs. As for out-of-order and more than 3 blurred images, the model needs to be retrained...

Same issue of "demo_colmap.py". Out of memory (80G) with 250 images. Any help, thanks?

> Same issue of "demo_colmap.py". Out of memory (80G) with 250 images. Any help, thanks? I noticed that the increase in GPU memory consumption in demo_colmap.py is due to the...

This method only supports 3 nearby frames as inputs. As for a single image, you can try: https://github.com/zzh-tech/Animation-from-Blur https://github.com/MeiguangJin/Learning-to-Extract-a-Video-Sequence-from-a-Single-Motion-Blurred-Image