Update app.py

app.py

@@ -99,10 +99,10 @@ audio2exp_yaml_path = os.path.join(current_root_path, 'src', 'config', 'auido2ex
 free_view_checkpoint = os.path.join(current_root_path, 'checkpoints', 'facevid2vid_00189-model.pth.tar')
 
 # Function for running the actual task (using preprocessed data)
-def process_chunk(audio_chunk, args):
+def process_chunk(audio_chunk, args, preprocessed_data):
     print("Entered Process Chunk Function")
     global audio2pose_checkpoint, audio2pose_yaml_path, audio2exp_checkpoint, audio2exp_yaml_path, wav2lip_checkpoint
-    global path_of_lm_croper, path_of_net_recon_model, dir_of_BFM_fitting
+    # global path_of_lm_croper, path_of_net_recon_model, dir_of_BFM_fitting
     global free_view_checkpoint
     if args.preprocess == 'full':
         mapping_checkpoint = os.path.join(current_root_path, 'checkpoints', 'mapping_00109-model.pth.tar')
@@ -111,16 +111,16 @@ def process_chunk(audio_chunk, args):
         mapping_checkpoint = os.path.join(current_root_path, 'checkpoints', 'mapping_00229-model.pth.tar')
         facerender_yaml_path = os.path.join(current_root_path, 'src', 'config', 'facerender.yaml')
 
-    # first_coeff_path = preprocessed_data["first_coeff_path"]
-    # crop_pic_path = preprocessed_data["crop_pic_path"]
-    # crop_info_path = "/home/user/app/preprocess_data/crop_info.json"
-    # with open(crop_info_path , "rb") as f:
-            # crop_info = json.load(f)
+    first_coeff_path = preprocessed_data["first_coeff_path"]
+    crop_pic_path = preprocessed_data["crop_pic_path"]
+    crop_info_path = "/home/user/app/preprocess_data/crop_info.json"
+    with open(crop_info_path , "rb") as f:
+            crop_info = json.load(f)
 
-    first_frame_dir = os.path.join(args.result_dir, 'first_frame_dir')
-    os.makedirs(first_frame_dir, exist_ok=True)
-    preprocess_model = CropAndExtract(path_of_lm_croper, path_of_net_recon_model, dir_of_BFM_fitting, args.device)
-    first_coeff_path, crop_pic_path, crop_info =  preprocess_model.generate(args.source_image, first_frame_dir, args.preprocess, source_image_flag=True)
+    # first_frame_dir = os.path.join(args.result_dir, 'first_frame_dir')
+    # os.makedirs(first_frame_dir, exist_ok=True)
+    # preprocess_model = CropAndExtract(path_of_lm_croper, path_of_net_recon_model, dir_of_BFM_fitting, args.device)
+    # first_coeff_path, crop_pic_path, crop_info =  preprocess_model.generate(args.source_image, first_frame_dir, args.preprocess, source_image_flag=True)
 
     print(f"Loaded existing preprocessed data")
     print("first_coeff_path",first_coeff_path)
@@ -228,10 +228,10 @@ def generate_audio(voice_cloning, voice_gender, text_prompt):
     elif voice_cloning == 'yes':
         set_api_key('92e149985ea2732b4359c74346c3daee')
         user_voice_path = '/home/user/app/images/AUDIO-2024-10-04-09-51-34.m4a'
-        voice = clone(name = "User Cloned Voice",
-                    files = [user_voice_path] )
-        # voice = Voice(voice_id="DeZH4ash9IU9gUcNjVXh",name="Marc",settings=VoiceSettings(
-        #                 stability=0.71, similarity_boost=0.5, style=0.0, use_speaker_boost=True),)
+        # voice = clone(name = "User Cloned Voice",
+        #             files = [user_voice_path] )
+        voice = Voice(voice_id="DeZH4ash9IU9gUcNjVXh",name="Marc",settings=VoiceSettings(
+                        stability=0.71, similarity_boost=0.5, style=0.0, use_speaker_boost=True),)
 
         audio = generate(text = text_prompt, voice = voice, model = "eleven_multilingual_v2",stream=True, latency=4)
         with tempfile.NamedTemporaryFile(suffix=".mp3", prefix="cloned_audio_",dir=TEMP_DIR.name, delete=False) as temp_file:
@@ -271,7 +271,7 @@ def generate_video():
     try:
         if request.method == 'POST':
             # source_image = request.files['source_image']
-            image_path = '/home/user/app/images/Image.jpg'
+            image_path = '/home/user/app/images/marc_smile_enhanced.jpg'
             source_image = Image.open(image_path)
             text_prompt = request.form['text_prompt']
             
@@ -336,8 +336,8 @@ def generate_video():
         
 
     try:
-        # preprocessed_data = run_preprocessing(args)
-        base64_video, temp_file_path = process_chunk(driven_audio_path, args)
+        preprocessed_data = run_preprocessing(args)
+        base64_video, temp_file_path = process_chunk(driven_audio_path, args, preprocessed_data)
         final_video_path = app.config['final_video_path']
         print('final_video_path',final_video_path)