Video format

app.py

@@ -31,6 +31,7 @@ def animate(
     fps_id: int = 6,
     noise_aug_strength: float = 0.1,
     decoding_t: int = 3,
+    video_format: str = "mp4",
     frame_format: str = "webp",
     version: str = "auto",
     output_folder: str = "outputs",
@@ -44,18 +45,16 @@ def animate(
     frames = animate_on_gpu(
         image,
         seed,
-        randomize_seed,
         motion_bucket_id,
         fps_id,
         noise_aug_strength,
         decoding_t,
-        frame_format,
         version
     )
     
     os.makedirs(output_folder, exist_ok=True)
-    base_count = len(glob(os.path.join(output_folder, "*.mp4")))
-    video_path = os.path.join(output_folder, f"{base_count:06d}.mp4")
+    base_count = len(glob(os.path.join(output_folder, "*." + video_format)))
+    video_path = os.path.join(output_folder, f"{base_count:06d}." + video_format)
 
     export_to_video(frames, video_path, fps=fps_id)
     
@@ -65,12 +64,10 @@ def animate(
 def animate_on_gpu(
     image: Image,
     seed: Optional[int] = 42,
-    randomize_seed: bool = True,
     motion_bucket_id: int = 127,
     fps_id: int = 6,
     noise_aug_strength: float = 0.1,
     decoding_t: int = 3,
-    frame_format: str = "webp",
     version: str = "auto"
 ):
     generator = torch.manual_seed(seed)
@@ -128,7 +125,8 @@ with gr.Blocks() as demo:
               motion_bucket_id = gr.Slider(label="Motion bucket id", info="Controls how much motion to add/remove from the image", value=127, minimum=1, maximum=255)
               noise_aug_strength = gr.Slider(label="Noise strength", info="The noise to add", value=0.1, minimum=0, maximum=1, step=0.1)
               decoding_t = gr.Slider(label="Decoding", info="Number of frames decoded at a time; this eats more VRAM; reduce if necessary", value=3, minimum=1, maximum=5, step=1)
-              frame_format = gr.Radio([["*.png", "png"], ["*.webp", "webp"], ["*.jpeg", "jpeg"], ["*.gif", "gif"], ["*.bmp", "bmp"]], label="Image format for result", info="File extention", value="webp", interactive=True)
+              video_format = gr.Radio([["*.mp4", "mp4"], ["*.avi", "avi"]], label="Video format for result", info="File extention", value="mp4", interactive=True)
+              frame_format = gr.Radio([["*.png", "png"], ["*.webp", "webp"], ["*.jpeg", "jpeg"], ["*.gif", "gif"], ["*.bmp", "bmp"]], label="Image format for frames", info="File extention", value="webp", interactive=True)
               version = gr.Radio([["Auto", "auto"], ["🏃🏻‍♀️ SVD (trained on 14 f/s)", "svd"], ["🏃🏻‍♀️💨 SVD-XT (trained on 25 f/s)", "svdxt"]], label="Model", info="Trained model", value="auto", interactive=True)
               seed = gr.Slider(label="Seed", value=42, randomize=True, minimum=0, maximum=max_64_bit_int, step=1)
               randomize_seed = gr.Checkbox(label="Randomize seed", value=True)
@@ -141,15 +139,15 @@ with gr.Blocks() as demo:
           gallery = gr.Gallery(label="Generated frames", visible=False)
       
   image.upload(fn=resize_image, inputs=image, outputs=image, queue=False)
-  generate_btn.click(fn=animate, inputs=[image, seed, randomize_seed, motion_bucket_id, fps_id, noise_aug_strength, decoding_t, frame_format, version], outputs=[video, download_button, gallery, seed], api_name="video")
+  generate_btn.click(fn=animate, inputs=[image, seed, randomize_seed, motion_bucket_id, fps_id, noise_aug_strength, decoding_t, video_format, frame_format, version], outputs=[video, download_button, gallery, seed], api_name="video")
     
   gr.Examples(
     examples=[
-        ["Examples/Fire.webp", 42, True, 127, 25, 0.1, 3, "png", "auto"],
-        ["Examples/Water.png", 42, True, 127, 25, 0.1, 3, "png", "auto"],
-        ["Examples/Town.jpeg", 42, True, 127, 25, 0.1, 3, "png", "auto"]
+        ["Examples/Fire.webp", 42, True, 127, 25, 0.1, 3, "mp4", "png", "auto"],
+        ["Examples/Water.png", 42, True, 127, 25, 0.1, 3, "mp4", "png", "auto"],
+        ["Examples/Town.jpeg", 42, True, 127, 25, 0.1, 3, "mp4", "png", "auto"]
     ],
-    inputs=[image, seed, randomize_seed, motion_bucket_id, fps_id, noise_aug_strength, decoding_t, frame_format, version],
+    inputs=[image, seed, randomize_seed, motion_bucket_id, fps_id, noise_aug_strength, decoding_t, video_format, frame_format, version],
     outputs=[video, download_button, gallery, seed],
     fn=animate,
     run_on_click=True,