Mingdeng

In fact, I will release the pretrained checkpoints gradually. Please stay tuned.

If you want to output all frames, you should add a FOR-LOOP in Sec.4 of the Colab Notebook. Just like the following: ```python for i, batch_data in enumerate(frames_data): out =...

Yep, I will update the notebook with saving the output frames soon!

Thanks for your attention and interest, we will release the trained checkpoints as quickly as possible.

Hi, in fact, I provide a simple demo notebook on [Colab](https://colab.research.google.com/drive/13dNBB38U_scDI46EHQ-V1GEgDMVtj5mx?usp=sharing), and you can run it to make a try quickly.

Hi, thanks for your interest in this project. In fact, most of these methods are trained with various datasets under the SimDeblur framework, and we release some of these CKPTs...

Hi, note that EDVR and STFAN models are required to be compiled on the GPU platform, and the script "inference_image.py" is just responsible for single image deblurring models, *e.g.*, SRN....

The blurs caused by moving objects occur locally, resulting in some local blur patterns. Thus you can add some priors corresponding to the object shape or some local operations to...