Update app.py

app.py

@@ -161,6 +161,12 @@ def is_square_video(video_path):
     return gr.update(visible=True)
 
 
+def txt_to_driving_video(input_text):
+    audio_outpath = gpu_wrapped_elevenlabs_pipeline_generate_voice(text)
+    video_outpath = gpu_wrapped_stf_pipeline_execute(audio_outpath)
+    return video_outpath
+
+
 
 # assets
 title_md = "assets/gradio_title.md"
@@ -193,162 +199,161 @@ driving_video_path=gr.Video()
 with gr.Blocks(theme=gr.themes.Soft()) as demo:
     #gr.HTML(load_description(title_md))
 
-    with gr.Tabs():
-        with gr.Tab("Text to LipSync"):
-            gr.Markdown("# Text to LipSync")
+    
+    gr.Markdown("# Talk-GEN by ESTsoft")
+    gr.Markdown("## Text to talking video generation")
 
-            gr.Markdown("### 1. Text to audio")
-            with gr.Row():
-                #with gr.Column():
-                script_txt = gr.Text()
-                audio_gen_button = gr.Button("Audio generation", variant="primary")
-                # with gr.Column():
-                #     txt2video_gen_button = gr.Button("txt2video generation", variant="primary")
-
-                #with gr.Column():
-                    #audio_gen_button = gr.Button("Audio generation", variant="primary")
-            with gr.Row():
-                    output_audio = gr.Audio(label="Generated audio", type="filepath")
-            
-            
-            gr.Markdown("### 2. Audio to Driving-Video")
-            with gr.Row():
-                    #audio_path_component = gr.Textbox(label="Input", value="assets/examples/driving/test_aud.mp3")
-                    video_gen_button = gr.Button("Audio to Video generation", variant="primary")
-            # with gr.Row():
-            #         #a2v_output = gr.Video()
-            #         driving_video_path.render()
-
-                
-            gr.Markdown("### 3. Image to Talking Video")
-            #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")
-                    gr.Examples(
-                        examples=[
-                            #[osp.join(example_portrait_dir, "01.webp")],
-                            [osp.join(example_portrait_dir, "02.webp")],
-                            [osp.join(example_portrait_dir, "03.jpg")],
-                            [osp.join(example_portrait_dir, "04.jpg")],
-                            [osp.join(example_portrait_dir, "05.jpg")],
-                            [osp.join(example_portrait_dir, "06.jpg")],
-                            [osp.join(example_portrait_dir, "07.jpg")],
-                            [osp.join(example_portrait_dir, "08.jpg")],
-                        ],
-                        inputs=[image_input],
-                        cache_examples=False,
-                    )
-                with gr.Accordion(open=True, label="Driving Video"):
-                    video_input = gr.Video()
-                    gr.Examples(
-                        examples=[
-                            [osp.join(example_video_dir, "d0.mp4")],
-                            [osp.join(example_video_dir, "d18.mp4")],
-                            [osp.join(example_video_dir, "d19.mp4")],
-                            [osp.join(example_video_dir, "d14_trim.mp4")],
-                            [osp.join(example_video_dir, "d6_trim.mp4")],
-                        ],
-                        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"))
-                    with gr.Row():
-                        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")
-            with gr.Row():
-                with gr.Column():
-                    with gr.Accordion(open=True, label="The animated video in the original image space"):
-                        output_video.render()
-                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,
-            #         fn=gpu_wrapped_execute_video,
-            #         inputs=[
-            #             image_input,
-            #             video_input,
-            #             flag_relative_input,
-            #             flag_do_crop_input,
-            #             flag_remap_input
-            #         ],
-            #         outputs=[output_image, output_image_paste_back],
-            #         examples_per_page=6,
-            #         cache_examples=False,
-            #     )
+    #gr.Markdown("### 1. Text to audio")
+    gr.Markdown("### 1. Text to Driving-Video")
+    with gr.Row():
         
-            process_button_animation.click(
-                fn=gpu_wrapped_execute_video,
-                inputs=[
-                    image_input,
-                    video_input,
-                    flag_relative_input,
-                    flag_do_crop_input,
-                    flag_remap_input
+        script_txt = gr.Text()
+        # audio_gen_button = gr.Button("Audio generation", variant="primary")
+        # with gr.Column():
+        #     txt2video_gen_button = gr.Button("txt2video generation", variant="primary")
+        txt2video_gen_button = gr.Button("txt2video generation", variant="primary")
+
+        #with gr.Column():
+            #audio_gen_button = gr.Button("Audio generation", variant="primary")
+    # with gr.Row():
+    #         output_audio = gr.Audio(label="Generated audio", type="filepath")
+    
+    
+    # gr.Markdown("### 2. Audio to Driving-Video")
+    # with gr.Row():
+    #         #audio_path_component = gr.Textbox(label="Input", value="assets/examples/driving/test_aud.mp3")
+    #         video_gen_button = gr.Button("Audio to Video generation", variant="primary")
+    # with gr.Row():
+    #         #a2v_output = gr.Video()
+    #         driving_video_path.render()
+
+        
+    gr.Markdown("### 2. Image to Talking-Video with Driving-Video")
+    #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")
+            gr.Examples(
+                examples=[
+                    #[osp.join(example_portrait_dir, "01.webp")],
+                    [osp.join(example_portrait_dir, "02.webp")],
+                    [osp.join(example_portrait_dir, "03.jpg")],
+                    [osp.join(example_portrait_dir, "04.jpg")],
+                    [osp.join(example_portrait_dir, "05.jpg")],
+                    [osp.join(example_portrait_dir, "06.jpg")],
+                    [osp.join(example_portrait_dir, "07.jpg")],
+                    [osp.join(example_portrait_dir, "08.jpg")],
                 ],
-                outputs=[output_video, output_video_concat],
-                show_progress=True
+                inputs=[image_input],
+                cache_examples=False,
             )
-            # txt2video_gen_button.click(
-            #     fn=txt_to_driving_video,
-            #     inputs=[
-            #         script_txt
-            #     ],
-            #     outputs=[video_input],
-            #     show_progress=True
-            # )
-            audio_gen_button.click(
-                fn=gpu_wrapped_elevenlabs_pipeline_generate_voice,
-                inputs=[
-                    script_txt
+        with gr.Accordion(open=True, label="Driving Video"):
+            video_input = gr.Video()
+            gr.Examples(
+                examples=[
+                    [osp.join(example_video_dir, "d0.mp4")],
+                    [osp.join(example_video_dir, "d18.mp4")],
+                    [osp.join(example_video_dir, "d19.mp4")],
+                    [osp.join(example_video_dir, "d14_trim.mp4")],
+                    [osp.join(example_video_dir, "d6_trim.mp4")],
                 ],
-                outputs=[output_audio],
-                show_progress=True
+                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"))
+            with gr.Row():
+                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")
+    with gr.Row():
+        with gr.Column():
+            with gr.Accordion(open=True, label="The animated video in the original image space"):
+                output_video.render()
+        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,
+    #         fn=gpu_wrapped_execute_video,
+    #         inputs=[
+    #             image_input,
+    #             video_input,
+    #             flag_relative_input,
+    #             flag_do_crop_input,
+    #             flag_remap_input
+    #         ],
+    #         outputs=[output_image, output_image_paste_back],
+    #         examples_per_page=6,
+    #         cache_examples=False,
+    #     )
+
+    process_button_animation.click(
+        fn=gpu_wrapped_execute_video,
+        inputs=[
+            image_input,
+            video_input,
+            flag_relative_input,
+            flag_do_crop_input,
+            flag_remap_input
+        ],
+        outputs=[output_video, output_video_concat],
+        show_progress=True
+    )
+    txt2video_gen_button.click(
+        fn=txt_to_driving_video,
+        inputs=[
+            script_txt
+        ],
+        outputs=[video_input],
+        show_progress=True
+    )
+    # audio_gen_button.click(
+    #     fn=gpu_wrapped_elevenlabs_pipeline_generate_voice,
+    #     inputs=[
+    #         script_txt
+    #     ],
+    #     outputs=[output_audio],
+    #     show_progress=True
+    # )
+
+    # video_gen_button.click(
+    #     fn=gpu_wrapped_stf_pipeline_execute,
+    #     inputs=[
+    #         output_audio
+    #         #audio_path_component
+    #     ],
+    #     outputs=[
+    #         video_input
+    #         #driving_video_path
+    #     ],
+    #     show_progress=True
+    # )
 
-            video_gen_button.click(
-                fn=gpu_wrapped_stf_pipeline_execute,
-                inputs=[
-                    output_audio
-                    #audio_path_component
-                ],
-                outputs=[
-                    video_input
-                    #driving_video_path
-                ],
-                show_progress=True
-            )
 
+    
+    # image_input.change(
+    #     fn=gradio_pipeline.prepare_retargeting,
+    #     inputs=image_input,
+    #     outputs=[eye_retargeting_slider, lip_retargeting_slider, retargeting_input_image]
+    # )
+    video_input.upload(
+        fn=is_square_video,
+        inputs=video_input,
+        outputs=video_input
+    )
 
-            
-            # image_input.change(
-            #     fn=gradio_pipeline.prepare_retargeting,
-            #     inputs=image_input,
-            #     outputs=[eye_retargeting_slider, lip_retargeting_slider, retargeting_input_image]
-            # )
-            video_input.upload(
-                fn=is_square_video,
-                inputs=video_input,
-                outputs=video_input
-            )
-        
-        # # 세 번째 탭: Flux 개발용 탭
-        # with gr.Tab("FLUX Image"):
-        #     flux_demo = create_flux_tab(image_input)  # Flux 개발용 탭 생성
 
 demo.launch(
     server_port=args.server_port,