Update app.py

app.py

@@ -197,9 +197,52 @@ css = """
 .tab-nav button:nth-child(3) { border-top: 3px solid #f7b731; }
 """
 
+# 먼저 text-to-video 함수를 정의합니다.
+@spaces.GPU(duration=300)
+def infer_t2v(prompt, seed=42, randomize_seed=False, width=1024, height=576, guidance_scale=5.0, num_inference_steps=28, 
+              video_steps=50, video_cfg_scale=7.5, video_eta=1.0, video_fps=3, video_length=2):
+    # 텍스트로 이미지 생성
+    image, _, translated_prompt = infer_t2i(prompt, seed, randomize_seed, width, height, guidance_scale, num_inference_steps)
+    
+    # 생성된 이미지로 비디오 생성
+    video_path = infer(np.array(image), translated_prompt, video_steps, video_cfg_scale, video_eta, video_fps, seed, video_length)
+    
+    return video_path, translated_prompt
+
+
 
 with gr.Blocks(analytics_enabled=False, css=css) as dynamicrafter_iface:
-    gr.Markdown("숏폼폼 스튜디오")
+    gr.Markdown("숏폼 스튜디오")
+
+    with gr.Tab(label='Text to Image'):
+        with gr.Column():
+            with gr.Row():
+                t2i_input_text = gr.Text(label='Prompt')
+            with gr.Row():
+                t2i_seed = gr.Slider(label='Seed', minimum=0, maximum=MAX_SEED, step=1, value=42)
+                t2i_randomize_seed = gr.Checkbox(label='Randomize seed', value=False)
+            with gr.Row():
+                t2i_width = gr.Slider(label='Width', minimum=256, maximum=MAX_IMAGE_SIZE, step=64, value=1024)
+                t2i_height = gr.Slider(label='Height', minimum=256, maximum=MAX_IMAGE_SIZE, step=64, value=576)
+            with gr.Row():
+                t2i_guidance_scale = gr.Slider(label='Guidance Scale', minimum=1.0, maximum=20.0, step=0.1, value=5.0)
+                t2i_num_inference_steps = gr.Slider(label='Inference Steps', minimum=1, maximum=100, step=1, value=28)
+#            t2i_generate_btn = gr.Button("Generate")
+#            t2i_output_image = gr.Image(label="Generated Image", elem_id="t2i_output_img")
+#            t2i_output_seed = gr.Number(label="Used Seed", elem_id="t2i_output_seed")
+
+        t2i_generate_btn = gr.Button("Generate")
+        t2i_output_image = gr.Image(label="Generated Image", elem_id="t2i_output_img")
+        t2i_output_seed = gr.Number(label="Used Seed", elem_id="t2i_output_seed")
+        t2i_translated_prompt = gr.Text(label="Translated Prompt (if applicable)", elem_id="t2i_translated_prompt")
+
+    t2i_generate_btn.click(
+        fn=infer_t2i,
+        inputs=[t2i_input_text, t2i_seed, t2i_randomize_seed, t2i_width, t2i_height, t2i_guidance_scale, t2i_num_inference_steps],
+        outputs=[t2i_output_image, t2i_output_seed, t2i_translated_prompt]
+    )
+        
+    
     with gr.Tab(label='Image+Text to Video'):
         with gr.Column():
             with gr.Row():
@@ -233,32 +276,35 @@ with gr.Blocks(analytics_enabled=False, css=css) as dynamicrafter_iface:
         )
 
 
-    with gr.Tab(label='Text to Image'):
+    with gr.Tab(label='Text to Video'):
         with gr.Column():
+            t2v_input_text = gr.Text(label='Prompt')
             with gr.Row():
-                t2i_input_text = gr.Text(label='Prompt')
+                t2v_seed = gr.Slider(label='Seed', minimum=0, maximum=MAX_SEED, step=1, value=42)
+                t2v_randomize_seed = gr.Checkbox(label='Randomize seed', value=False)
             with gr.Row():
-                t2i_seed = gr.Slider(label='Seed', minimum=0, maximum=MAX_SEED, step=1, value=42)
-                t2i_randomize_seed = gr.Checkbox(label='Randomize seed', value=False)
+                t2v_width = gr.Slider(label='Width', minimum=256, maximum=MAX_IMAGE_SIZE, step=64, value=1024)
+                t2v_height = gr.Slider(label='Height', minimum=256, maximum=MAX_IMAGE_SIZE, step=64, value=576)
             with gr.Row():
-                t2i_width = gr.Slider(label='Width', minimum=256, maximum=MAX_IMAGE_SIZE, step=64, value=1024)
-                t2i_height = gr.Slider(label='Height', minimum=256, maximum=MAX_IMAGE_SIZE, step=64, value=576)
+                t2v_guidance_scale = gr.Slider(label='Image Guidance Scale', minimum=1.0, maximum=20.0, step=0.1, value=5.0)
+                t2v_num_inference_steps = gr.Slider(label='Image Inference Steps', minimum=1, maximum=100, step=1, value=28)
             with gr.Row():
-                t2i_guidance_scale = gr.Slider(label='Guidance Scale', minimum=1.0, maximum=20.0, step=0.1, value=5.0)
-                t2i_num_inference_steps = gr.Slider(label='Inference Steps', minimum=1, maximum=100, step=1, value=28)
-#            t2i_generate_btn = gr.Button("Generate")
-#            t2i_output_image = gr.Image(label="Generated Image", elem_id="t2i_output_img")
-#            t2i_output_seed = gr.Number(label="Used Seed", elem_id="t2i_output_seed")
+                t2v_video_steps = gr.Slider(label='Video Steps', minimum=1, maximum=50, step=1, value=30)
+                t2v_video_cfg_scale = gr.Slider(label='Video CFG Scale', minimum=1.0, maximum=15.0, step=0.5, value=3.5)
+            with gr.Row():
+                t2v_video_eta = gr.Slider(label='Video ETA', minimum=0.0, maximum=1.0, step=0.1, value=1.0)
+                t2v_video_fps = gr.Slider(label='Video FPS', minimum=5, maximum=20, step=1, value=8)
+            t2v_video_length = gr.Slider(label='Video Length (seconds)', minimum=2, maximum=8, step=1, value=2)
+            
+            t2v_generate_btn = gr.Button("Generate Video")
+            t2v_output_video = gr.Video(label="Generated Video", elem_id="t2v_output_vid", autoplay=True, show_share_button=True)
+            t2v_translated_prompt = gr.Text(label="Translated Prompt (if applicable)", elem_id="t2v_translated_prompt")
 
-        t2i_generate_btn = gr.Button("Generate")
-        t2i_output_image = gr.Image(label="Generated Image", elem_id="t2i_output_img")
-        t2i_output_seed = gr.Number(label="Used Seed", elem_id="t2i_output_seed")
-        t2i_translated_prompt = gr.Text(label="Translated Prompt (if applicable)", elem_id="t2i_translated_prompt")
+        t2v_generate_btn.click(
+            fn=infer_t2v,
+            inputs=[t2v_input_text, t2v_seed, t2v_randomize_seed, t2v_width, t2v_height, t2v_guidance_scale, t2v_num_inference_steps,
+                    t2v_video_steps, t2v_video_cfg_scale, t2v_video_eta, t2v_video_fps, t2v_video_length],
+            outputs=[t2v_output_video, t2v_translated_prompt]
+        )
 
-    t2i_generate_btn.click(
-        fn=infer_t2i,
-        inputs=[t2i_input_text, t2i_seed, t2i_randomize_seed, t2i_width, t2i_height, t2i_guidance_scale, t2i_num_inference_steps],
-        outputs=[t2i_output_image, t2i_output_seed, t2i_translated_prompt]
-    )
-    
 dynamicrafter_iface.queue(max_size=12).launch(show_api=True)