Update app.py

app.py

@@ -1,6 +1,6 @@
 import gradio
 
-import basicsr, realesrgan, gfpgan, av, pathlib, diffusers, torch, transformers, builtins, numpy, re
+import av, pathlib, diffusers, torch, transformers, builtins, numpy, re
 from animatediff.generate import controlnet_preprocess, img2img_preprocess, wild_card_conversion, region_preprocess, unload_controlnet_models
 from animatediff.settings import get_model_config, get_infer_config
 from animatediff.utils.pipeline import send_to_device
@@ -128,16 +128,13 @@ music = pipeline(prompt='Light rhythm techno', negative_prompt='low quality, ave
 del pipeline
 torch.cuda.empty_cache()
 
-model = basicsr.archs.rrdbnet_arch.RRDBNet(num_in_ch=3, num_out_ch=3, num_feat=64, num_block=23, num_grow_ch=32, scale=4)
-upsampler = realesrgan.RealESRGANer(scale=4, model_path='https://huggingface.co/chaowenguoback/pal/resolve/main/RealESRGAN_x4plus.pth', model=model, half=True, device='cuda')
-face_enhancer = gfpgan.GFPGANer(model_path='https://huggingface.co/chaowenguoback/pal/resolve/main/GFPGANv1.4.pth',upscale=4, bg_upsampler=upsampler)
 with av.open('video.mp4', mode='w') as writer:
     video = writer.add_stream('h264', rate=8)
     video.width = width * 4
     video.height = height * 4
     video.pix_fmt = 'yuv420p'
     audio = writer.add_stream('aac', rate=16000)
-    for frame in frames: writer.mux(video.encode(av.VideoFrame.from_ndarray(face_enhancer.enhance(frame)[-1])))
+    for frame in frames: writer.mux(video.encode(av.VideoFrame.from_ndarray(frame)))
     writer.mux(video.encode())
     for _ in builtins.range(0, music.shape[0], audio.frame_size):
         frame = av.AudioFrame.from_ndarray(music[_:_ + audio.frame_size][None], format='fltp', layout='mono')