Upload 4 files

README.md

@@ -1,21 +1,21 @@
 ---
-title: SUPIR Image Upscaler
+title: FramePack/HunyuanVideo
+emoji: 🎥
+colorFrom: pink
+colorTo: gray
 sdk: gradio
-emoji: 📷
-sdk_version: 4.38.1
-app_file: app.py
-license: mit
-colorFrom: blue
-colorTo: pink
+sdk_version: 5.29.1
+app_file: app_endframe.py
+license: apache-2.0
+short_description: Text-to-Video/Image-to-Video/Video extender
 tags:
-  - Upscaling
-  - Restoring
-  - Image-to-Image
-  - Image-2-Image
-  - Img-to-Img
-  - Img-2-Img
+  - Image-to-Video
+  - Image-2-Video
+  - Img-to-Vid
+  - Img-2-Vid
   - language models
   - LLMs
-short_description: Restore blurred or small images with prompt
 suggested_hardware: zero-a10g
----
+---
+
+Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

app_endframe.py

@@ -496,7 +496,7 @@ def worker(input_video, end_frame, end_frame_weight, prompt, n_prompt, seed, bat
 
                 # 20250509 pftq: Dynamic frame allocation like original num_clean_frames, fix split error
                 available_frames = video_latents.shape[2] if is_start_of_video else history_latents.shape[2]
-                if is_start_of_video:
+                if is_start_of_video and False:
                     effective_clean_frames = 1 # avoid jumpcuts from input video
                 else:
                     effective_clean_frames = max(0, num_clean_frames - 1) if num_clean_frames > 1 else 1
@@ -632,8 +632,6 @@ def worker(input_video, end_frame, end_frame_weight, prompt, n_prompt, seed, bat
                     break
 
             history_pixels = torch.cat([input_video_pixels, history_pixels], dim=2)
-            #overlapped_frames = latent_window_size * 4 - 3
-            #history_pixels = soft_append_bcthw(input_video_pixels, history_pixels, overlapped_frames)
 
             output_filename = os.path.join(outputs_folder, f'{job_id}_final.mp4')
             save_bcthw_as_mp4(history_pixels, output_filename, fps=fps, crf=mp4_crf)

requirements.txt

@@ -1,41 +1,24 @@
-pydantic==2.10.6
-fastapi==0.115.8
-gradio_imageslider==0.0.20
-gradio_client==1.7.0
-numpy==1.26.4
-requests==2.32.3
-sentencepiece==0.2.0
-tokenizers==0.19.1
-torchvision==0.22.0
-uvicorn==0.30.1
-wandb==0.17.4
-httpx==0.27.0
-transformers==4.42.4
-accelerate==0.32.1
-scikit-learn==1.5.1
-einops==0.8.0
-einops-exts==0.0.4
-timm==1.0.7
-openai-clip==1.0.1
-fsspec==2024.6.1
-kornia==0.7.3
-matplotlib==3.9.1
-ninja==1.11.1.1
-omegaconf==2.3.0
-opencv-python==4.11.0.86
-pandas==2.3.0
-pillow==11.2.1
-pytorch-lightning==2.5.1.post0
-PyYAML==6.0.2
-scipy==1.15.3
-tqdm==4.67.1
-triton==3.3.0
-urllib3==2.4.0
-webdataset==0.2.111
-xformers==0.0.30
-facexlib==0.3.0
-k-diffusion==0.1.1.post1
+accelerate==1.6.0
 diffusers==0.33.1
-pillow-heif==0.22.0
-
-open-clip-torch==2.24.0
+transformers==4.46.2
+sentencepiece==0.2.0
+pillow==11.1.0
+av==12.1.0
+numpy==1.26.2
+scipy==1.12.0
+requests==2.31.0
+torchsde==0.2.6
+torch>=2.0.0
+torchvision
+torchaudio
+einops
+opencv-contrib-python
+safetensors
+huggingface_hub
+spaces
+decord
+imageio_ffmpeg
+sageattention
+xformers==0.0.29.post3
+bitsandbytes
+pillow-heif==0.22.0