Update app.py

app.py

@@ -444,16 +444,41 @@ def process(
     input_image, prompt, t2v=False, n_prompt="", seed=31337,
     total_second_length=60, latent_window_size=9, steps=25,
     cfg=1.0, gs=10.0, rs=0.0, gpu_memory_preservation=6,
-    use_teacache=True, mp4_crf=16
+    use_teacache=True, mp4_crf=16, quality_radio="640x360", aspect_ratio="1:1"
 ):
     global stream
+    quality_map = {
+        "360p": (640, 360),
+        "480p": (854, 480),
+        "540p": (960, 540),
+        "720p": (1280, 720),
+        "640x360": (640, 360),  # fallback for default
+    }
+    # Aspect ratio map: (width, height)
+    aspect_map = {
+        "1:1": (1, 1),
+        "3:4": (3, 4),
+        "4:3": (4, 3),
+        "16:9": (16, 9),
+        "9:16": (9, 16),
+    }
+    selected_quality = quality_map.get(quality_radio, (640, 360))
+    base_width, base_height = selected_quality
+
     if t2v:
-        default_height, default_width = 640, 640
-        input_image = np.ones((default_height, default_width, 3), dtype=np.uint8) * 255
-        print("Using blank white image for text-to-video mode")
+        # Use aspect ratio to determine final width/height
+        ar_w, ar_h = aspect_map.get(aspect_ratio, (1, 1))
+        if ar_w >= ar_h:
+            target_height = base_height
+            target_width = int(round(target_height * ar_w / ar_h))
+        else:
+            target_width = base_width
+            target_height = int(round(target_width * ar_h / ar_w))
+        input_image = np.ones((target_height, target_width, 3), dtype=np.uint8) * 255
+        print(f"Using blank white image for text-to-video mode, {target_width}x{target_height} ({aspect_ratio})")
     else:
+        target_width, target_height = selected_quality
         if isinstance(input_image, dict) and "composite" in input_image:
-            # Handle uploaded image with alpha channel
             composite_rgba_uint8 = input_image["composite"]
             rgb_uint8 = composite_rgba_uint8[:, :, :3]
             mask_uint8 = composite_rgba_uint8[:, :, 3]
@@ -467,7 +492,6 @@ def process(
         elif input_image is None:
             raise ValueError("Please provide an input image or enable Text to Video mode")
         else:
-            # Handle regular RGB image
             input_image = input_image.astype(np.uint8)
 
     yield None, None, '', '', gr.update(interactive=False), gr.update(interactive=True)
@@ -668,10 +692,10 @@ with block:
             with gr.Group():
                 total_second_length = gr.Slider(
                     label="Duration (Seconds)",
-                    minimum=0.01,
-                    maximum=15,
+                    minimum=1,
+                    maximum=10,
                     value=2,
-                    step=0.1,
+                    step=1,
                     info='Length of generated video'
                 )
                 steps = gr.Slider(
@@ -687,16 +711,22 @@ with block:
                     minimum=1.0,
                     maximum=32.0,
                     value=10.0,
-                    step=1,
+                    step=0.1,
                     info='8-12 recommended'
                 )
-                mp4_crf = gr.Slider(
-                    label="Video Quality",
-                    minimum=0,
-                    maximum=51,
-                    value=16,
-                    step=1,
-                    info='15-20 recommended'
+                quality_radio = gr.Radio(
+                    label="Video Quality (Resolution)",
+                    choices=["360p", "480p", "540p", "720p"],
+                    value="640x360",
+                    info="Choose output video resolution"
+                )
+                # Aspect ratio dropdown, hidden by default
+                aspect_ratio = gr.Dropdown(
+                    label="Aspect Ratio",
+                    choices=["1:1", "3:4", "4:3", "16:9", "9:16"],
+                    value="1:1",
+                    visible=False,
+                    info="Only applies to Text to Video mode"
                 )
             
             gr.Markdown("### Advanced")
@@ -793,7 +823,8 @@ with block:
         input_image, prompt, t2v, n_prompt, seed,
         total_second_length, latent_window_size,
         steps, cfg, gs, rs, gpu_memory_preservation,
-        use_teacache, mp4_crf
+        use_teacache, 16,  # mp4_crf default
+        quality_radio, aspect_ratio
     ]
     
     start_button.click(
@@ -816,4 +847,14 @@ with block:
         queue=False
     )
 
+    # Show/hide aspect ratio dropdown based on t2v checkbox
+    def show_aspect_ratio(t2v_checked):
+        return gr.update(visible=bool(t2v_checked))
+    t2v.change(
+        fn=show_aspect_ratio,
+        inputs=[t2v],
+        outputs=[aspect_ratio],
+        queue=False
+    )
+
 block.launch(share=True)