AlexanderGroshev

Hello, @zhuya1996! Have you downloaded the arcface model? Use `sh download_models.sh` for this. This command creates the `arcface_model` folder with weights inside.

Hi, @epicstar7! The output face may look blurry because our model only generates a 256x256 resolution. To solve this problem, you can look at the face enhancement function. If you...

@epicstar7 check [inference_GFPGAN.py](https://github.com/AlexanderGroshev/sber-swap/blob/superres/inference_GFPGAN.py): ```bash python inference_GFPGAN.py --source_paths examples/images/beckham.jpg --target_video examples/videos/nggyup.mp4 --use_sr True ``` You need to clone GFPGAN repository, download weights `GFPGANCleanv1-NoCE-C2.pth` and place to `GFPGAN/experiments/pretrained_models/` folder

@epicstar7, try to reduce batch size, it might help.

@bmc84 Thanks for your feedback! Jittering of the face is a challenge we are working on. Longer training may partially solve this task. Two copies of the eyebrows may appear...

Hi, @alcanunsal, sorry for the late response! We trained model in 2 steps. For first step we did not use eye loss and set other weights as follows: `weight_adv 1...

Hi, @bmc84 ! You can check our colab, it can help you set up environment, since it based on CUDA 11.

You need to match the versions of pytorch, mxnet and onnxruntime-gpu to your CUDA. If you are using CUDA 10, you can follow requirements.txt. For CUDA 11.2 you can try...