Update app.py

app.py

@@ -13,7 +13,6 @@ from src.config.crop_config import CropConfig
 from src.config.argument_config import ArgumentConfig
 from src.config.inference_config import InferenceConfig
 import spaces
-import cv2
 
 # import gdown
 # folder_url = f"https://drive.google.com/drive/folders/1UtKgzKjFAOmZkhNK-OYT0caJ_w2XAnib"
@@ -27,8 +26,8 @@ tyro.extras.set_accent_color("bright_cyan")
 args = tyro.cli(ArgumentConfig)
 
 # specify configs for inference
-inference_cfg = partial_fields(InferenceConfig, args.__dict__)  # use attribute of args to initial InferenceConfig
-crop_cfg = partial_fields(CropConfig, args.__dict__)  # use attribute of args to initial CropConfig
+inference_cfg = partial_fields(InferenceConfig, args.__dict__)
+crop_cfg = partial_fields(CropConfig, args.__dict__)
 
 gradio_pipeline = GradioPipeline(
     inference_cfg=inference_cfg,
@@ -44,17 +43,10 @@ def gpu_wrapped_execute_video(*args, **kwargs):
 def gpu_wrapped_execute_image(*args, **kwargs):
     return gradio_pipeline.execute_image(*args, **kwargs)
 
-def is_square_video(video_path):
-    video = cv2.VideoCapture(video_path)
-
-    width = int(video.get(cv2.CAP_PROP_FRAME_WIDTH))
-    height = int(video.get(cv2.CAP_PROP_FRAME_HEIGHT))
-
-    video.release()
-    if width != height:
-        raise gr.Error("Error: the video does not have a square aspect ratio. We currently only support square videos")
-
-    return gr.update(visible=True)
+# Removida a exigência de vídeo quadrado.
+# Mantemos uma função neutra só para não quebrar a lógica se quiser reutilizar depois.
+def accept_any_video(video_path):
+    return gr.update()
 
 # assets
 title_md = "assets/gradio_title.md"
@@ -68,9 +60,9 @@ data_examples = [
     [osp.join(example_portrait_dir, "s7.jpg"), osp.join(example_video_dir, "d19.mp4"), True, True, True, True],
     [osp.join(example_portrait_dir, "s22.jpg"), osp.join(example_video_dir, "d0.mp4"), True, True, True, True],
 ]
+
 #################### interface logic ####################
 
-# Define components first
 eye_retargeting_slider = gr.Slider(minimum=0, maximum=0.8, step=0.01, label="target eyes-open ratio")
 lip_retargeting_slider = gr.Slider(minimum=0, maximum=0.8, step=0.01, label="target lip-open ratio")
 retargeting_input_image = gr.Image(type="filepath")
@@ -82,6 +74,7 @@ output_video_concat = gr.Video()
 with gr.Blocks(theme=gr.themes.Soft()) as demo:
     gr.HTML(load_description(title_md))
     gr.Markdown(load_description("assets/gradio_description_upload.md"))
+
     with gr.Row():
         with gr.Accordion(open=True, label="Source Portrait"):
             image_input = gr.Image(type="filepath")
@@ -98,6 +91,7 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
                 inputs=[image_input],
                 cache_examples=False,
             )
+
         with gr.Accordion(open=True, label="Driving Video"):
             video_input = gr.Video()
             gr.Examples(
@@ -111,6 +105,7 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
                 inputs=[video_input],
                 cache_examples=False,
             )
+
     with gr.Row():
         with gr.Accordion(open=False, label="Animation Instructions and Options"):
             gr.Markdown(load_description("assets/gradio_description_animation.md"))
@@ -118,12 +113,18 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
                 flag_relative_input = gr.Checkbox(value=True, label="relative motion")
                 flag_do_crop_input = gr.Checkbox(value=True, label="do crop")
                 flag_remap_input = gr.Checkbox(value=True, label="paste-back")
+
     gr.Markdown(load_description("assets/gradio_description_animate_clear.md"))
+
     with gr.Row():
         with gr.Column():
             process_button_animation = gr.Button("🚀 Animate", variant="primary")
         with gr.Column():
-            process_button_reset = gr.ClearButton([image_input, video_input, output_video, output_video_concat], value="🧹 Clear")
+            process_button_reset = gr.ClearButton(
+                [image_input, video_input, output_video, output_video_concat],
+                value="🧹 Clear"
+            )
+
     with gr.Row():
         with gr.Column():
             with gr.Accordion(open=True, label="The animated video in the original image space"):
@@ -131,9 +132,10 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
         with gr.Column():
             with gr.Accordion(open=True, label="The animated video"):
                 output_video_concat.render()
+
     with gr.Row():
-        # Examples
         gr.Markdown("## You could also choose the examples below by one click ⬇️")
+
     with gr.Row():
         gr.Examples(
             examples=data_examples,
@@ -145,14 +147,17 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
                 flag_do_crop_input,
                 flag_remap_input
             ],
-            outputs=[output_image, output_image_paste_back],
+            outputs=[output_video, output_video_concat],
             examples_per_page=6,
             cache_examples=False,
         )
+
     gr.Markdown(load_description("assets/gradio_description_retargeting.md"), visible=True)
+
     with gr.Row(visible=True):
         eye_retargeting_slider.render()
         lip_retargeting_slider.render()
+
     with gr.Row(visible=True):
         process_button_retargeting = gr.Button("🚗 Retargeting", variant="primary")
         process_button_reset_retargeting = gr.ClearButton(
@@ -165,6 +170,7 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
             ],
             value="🧹 Clear"
         )
+
     with gr.Row(visible=True):
         with gr.Column():
             with gr.Accordion(open=True, label="Retargeting Input"):
@@ -188,14 +194,14 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
         with gr.Column():
             with gr.Accordion(open=True, label="Paste-back Result"):
                 output_image_paste_back.render()
-    # binding functions for buttons
+
     process_button_retargeting.click(
-        # fn=gradio_pipeline.execute_image,
         fn=gpu_wrapped_execute_image,
         inputs=[eye_retargeting_slider, lip_retargeting_slider, retargeting_input_image, flag_do_crop_input],
         outputs=[output_image, output_image_paste_back],
         show_progress=True
     )
+
     process_button_animation.click(
         fn=gpu_wrapped_execute_video,
         inputs=[
@@ -208,13 +214,10 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
         outputs=[output_video, output_video_concat],
         show_progress=True
     )
-    # image_input.change(
-    #     fn=gradio_pipeline.prepare_retargeting,
-    #     inputs=image_input,
-    #     outputs=[eye_retargeting_slider, lip_retargeting_slider, retargeting_input_image]
-    # )
+
+    # Sem validação obrigando vídeo quadrado
     video_input.upload(
-        fn=is_square_video,
+        fn=accept_any_video,
         inputs=video_input,
         outputs=video_input
     )
@@ -222,4 +225,4 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
 demo.launch(
     server_name="0.0.0.0",
     server_port=7860
-)
+)