Upload folder using huggingface_hub

README.md

@@ -1,12 +1,12 @@
 ---
-title: Stable Video Diffusion Upscale
-emoji: 🦀
-colorFrom: indigo
+title: Stable Video Diffusion with Upscaler
+emoji: 📺
+colorFrom: gray
 colorTo: green
 sdk: gradio
-sdk_version: 4.40.0
+sdk_version: 4.37.2
 app_file: app.py
 pinned: false
----
-
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+license: other
+disable_embedding: true
+---

app.py

@@ -0,0 +1,60 @@
+import gradio as gr
+import os
+from glob import glob
+from diffusers.utils import load_image
+import spaces
+from panna.pipeline import PipelineSVDUpscale
+
+
+model = PipelineSVDUpscale(upscaler="instruct_ir")
+example_files = []
+root_url = "https://huggingface.co/spaces/multimodalart/stable-video-diffusion/resolve/main/images"
+examples = ["disaster_meme.png", "distracted_meme.png", "hide_meme.png", "success_meme.png", "willy_meme.png", "wink_meme.png"]
+for example in examples:
+    load_image(f"{root_url}/{example}").save(example)
+tmp_output_dir = "outputs"
+os.makedirs(tmp_output_dir, exist_ok=True)
+title = ("# [Stable Video Diffusion](ttps://huggingface.co/stabilityai/stable-video-diffusion-img2vid-xt)\n"
+         "The demo is part of [panna](https://github.com/abacws-abacus/panna) project.")
+
+
+@spaces.GPU(duration=120)
+def infer(init_image, upscaler_prompt, num_frames, motion_bucket_id, noise_aug_strength, decode_chunk_size, fps, seed):
+    base_count = len(glob(os.path.join(tmp_output_dir, "*.mp4")))
+    video_path = os.path.join(tmp_output_dir, f"{base_count:06d}.mp4")
+    model(
+        init_image,
+        output_path=video_path,
+        prompt=upscaler_prompt,
+        num_frames=num_frames,
+        motion_bucket_id=motion_bucket_id,
+        noise_aug_strength=noise_aug_strength,
+        decode_chunk_size=decode_chunk_size,
+        fps=fps,
+        seed=seed
+    )
+    return video_path
+
+
+with gr.Blocks() as demo:
+    gr.Markdown(title)
+    with gr.Row():
+        with gr.Column():
+            image = gr.Image(label="Upload your image", type="pil")
+            run_button = gr.Button("Generate")
+        video = gr.Video()
+    with gr.Accordion("Advanced options", open=False):
+        upscaler_prompt = gr.Text(label="Prompt for upscaler", show_label=False, max_lines=1, placeholder="Enter your prompt", container=False)
+        seed = gr.Slider(label="Seed", minimum=0, maximum=1_000_000, step=1, value=0)
+        num_frames = gr.Slider(label="Number of frames", minimum=1, maximum=100, step=1, value=25)
+        motion_bucket_id = gr.Slider(label="Motion bucket id", minimum=1, maximum=255, step=1, value=127)
+        noise_aug_strength = gr.Slider(label="Noise strength", minimum=0, maximum=1, step=0.01, value=0.02)
+        fps = gr.Slider(label="Frames per second", minimum=5, maximum=30, step=1, value=6)
+        decode_chunk_size = gr.Slider(label="Decode chunk size", minimum=1, maximum=10, step=1, value=2)
+    run_button.click(
+        fn=infer,
+        inputs=[image, upscaler_prompt, num_frames, motion_bucket_id, noise_aug_strength, decode_chunk_size, fps, seed],
+        outputs=[video]
+    )
+    gr.Examples(examples=examples, inputs=image)
+demo.launch()

requirements.txt

@@ -0,0 +1,8 @@
+https://gradio-builds.s3.amazonaws.com/756e3431d65172df986a7e335dce8136206a293a/gradio-4.7.1-py3-none-any.whl
+git+https://github.com/huggingface/diffusers.git
+transformers
+accelerate
+safetensors
+opencv-python
+uuid
+panna>=0.0.2