update

app.py

@@ -7,12 +7,13 @@ from scripts.inference import inference_process
 import argparse
 import uuid
 
-is_shared_ui = True if "fudan-generative-ai/hallo" in os.environ['SPACE_ID'] else False
+# is_shared_ui = True if "fudan-generative-ai/hallo" in os.environ['SPACE_ID'] else False
+is_shared_ui = False
 
 if(not is_shared_ui):
     hallo_dir = snapshot_download(repo_id="fudan-generative-ai/hallo", local_dir="pretrained_models")
 
-def run_inference(source_image, driving_audio, pose_weight, face_weight, lip_weight, face_expand_ratio, progress=gr.Progress(track_tqdm=True)):
+def run_inference(source_image, driving_audio, progress=gr.Progress(track_tqdm=True)):
     if is_shared_ui:
         raise gr.Error("This Space only works in duplicated instances")
         
@@ -23,10 +24,10 @@ def run_inference(source_image, driving_audio, pose_weight, face_weight, lip_wei
         source_image=source_image,
         driving_audio=driving_audio,
         output=f'output-{unique_id}.mp4',
-        pose_weight=pose_weight,
-        face_weight=face_weight,
-        lip_weight=lip_weight,
-        face_expand_ratio=face_expand_ratio,
+        pose_weight=1.0,
+        face_weight=1.0,
+        lip_weight=1.0,
+        face_expand_ratio=1.2,
         checkpoint=None
     )
     
@@ -112,17 +113,13 @@ We have provided some [samples](https://huggingface.co/datasets/fudan-generative
         with gr.Column():
             avatar_face = gr.Image(type="filepath", label="Face")
             driving_audio = gr.Audio(type="filepath", label="Driving audio")
-            pose_weight = gr.Number(label="pose weight", value=1.0),
-            face_weight = gr.Number(label="face weight", value=1.0),
-            lip_weight = gr.Number(label="lip weight", value=1.0),
-            face_expand_ratio = gr.Number(label="face expand ratio", value=1.2),
             generate = gr.Button("Generate")
         with gr.Column():
             output_video = gr.Video(label="Your talking head")
 
     generate.click(
         fn=run_inference,
-        inputs=[avatar_face, driving_audio, pose_weight, face_weight, lip_weight, face_expand_ratio],
+        inputs=[avatar_face, driving_audio],
         outputs=output_video
     )