Update app.py

app.py

@@ -24,7 +24,7 @@ import PIL
 import torch.nn.functional as F
 from torchvision import transforms
 
-#import spaces
+import spaces
 from huggingface_hub import snapshot_download
 
 DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
@@ -122,7 +122,7 @@ def init_pipe():
 
     return pipe
     
-#@spaces.GPU()#duration=120
+@spaces.GPU()#duration=120
 def inference(source_images, 
         target_images, 
         text_prompt, negative_prompt, 
@@ -133,6 +133,7 @@ def inference(source_images,
     pipe.vae.to(DEVICE)
     pipe.transformer.to(DEVICE)
     pipe.controlnet_transformer.to(DEVICE)
+    print("pipe.vae/transformer/controlnet移动到GPU里")
 
     source_pixel_values = source_images/127.5 - 1.0
     source_pixel_values = source_pixel_values.to(torch.float16).to(DEVICE)
@@ -161,6 +162,7 @@ def inference(source_images,
             image_latents = None
             latents = source_latents
 
+    print("准备调用pipe()函数")
     video = pipe(
         prompt = text_prompt,
         negative_prompt = negative_prompt,
@@ -207,7 +209,7 @@ def process_video(video_file, image_file, positive_prompt, negative_prompt, guid
     
     video:List[PIL.Image.Image]=[]
 
-    
+    print("所有事都干完了，准备inference。")
     for i in progress.tqdm(range(video_shard)):
         if i>0: #first frame guidence
             first_frame=transforms.ToTensor()(video[-1])