Make the introduction understandable by large public (and change video formats) (#23)

- Make the introduction understandable by large public (and change video formats) (cefb3322111530d555dba8d76990a865e9521b7a)


Co-authored-by: Fabrice TIERCELIN <Fabrice-TIERCELIN@users.noreply.huggingface.co>

app.py

@@ -44,6 +44,12 @@ def animate(
         
     if randomize_seed:
         seed = random.randint(0, max_64_bit_int)
+    
+    if version == "auto"):
+        if 14 < fps_id:
+            version = "svdxt"
+        else:
+            version = "svd"
 
     frames = animate_on_gpu(
         image,
@@ -83,11 +89,11 @@ def animate_on_gpu(
     fps_id: int = 6,
     noise_aug_strength: float = 0.1,
     decoding_t: int = 3,
-    version: str = "auto"
+    version: str = "svdxt"
 ):
     generator = torch.manual_seed(seed)
-    
-    if version == "svdxt" or (14 < fps_id and version != "svd"):
+
+    if version == "svdxt":
         return fps25Pipe(image, decode_chunk_size=decoding_t, generator=generator, motion_bucket_id=motion_bucket_id, noise_aug_strength=noise_aug_strength, num_frames=25).frames[0]
     else:
         return fps14Pipe(image, decode_chunk_size=decoding_t, generator=generator, motion_bucket_id=motion_bucket_id, noise_aug_strength=noise_aug_strength, num_frames=25).frames[0]
@@ -129,9 +135,16 @@ def resize_image(image, output_size=(1024, 576)):
     return cropped_image
 
 with gr.Blocks() as demo:
-  gr.Markdown('''# Community demo for Stable Video Diffusion - Img2Vid - XT ([model](https://huggingface.co/stabilityai/stable-video-diffusion-img2vid-xt), [paper](https://stability.ai/research/stable-video-diffusion-scaling-latent-video-diffusion-models-to-large-datasets), [stability's ui waitlist](https://stability.ai/contact))
-#### Research release ([_non-commercial_](https://huggingface.co/stabilityai/stable-video-diffusion-img2vid-xt/blob/main/LICENSE)): generate `4s` vid from a single image at (`25 frames` at `6 fps`). this demo uses [🧨 diffusers for low VRAM and fast generation](https://huggingface.co/docs/diffusers/main/en/using-diffusers/svd).
-  ''')
+  gr.HTML("""
+    <h1><center>Image-to-Video</center></h1>
+    <big><center>Animate your images into 25 frames of 1024x576 pixels freely, without account, without watermark and download the video</center></big>
+    <br/>
+    
+    <p>
+    This demo is based on <i>Stable Video Diffusion</i> artificial intelligence.
+    No prompt or camera control is handled here. To control motions, rather use <i><a href="https://huggingface.co/spaces/TencentARC/MotionCtrl_SVD">MotionCtrl SVD</a></i>.
+    </p>
+    """)
   with gr.Row():
       with gr.Column():
           image = gr.Image(label="Upload your image", type="pil")
@@ -140,8 +153,8 @@ 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)
-              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)
+              video_format = gr.Radio([["*.mp4", "mp4"], ["*.ogg", "ogg"], ["*.webm", "webm"]], label="Video format for result", info="File extention", value="mp4", interactive=True)
+              frame_format = gr.Radio([["*.png", "png"], ["*.webp", "webp"], ["*.jpeg", "jpeg"], ["*.gif (unanimated)", "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)