Spaces:
Build error
Build error
Update README.md
Browse files
README.md
CHANGED
|
@@ -1,84 +1,60 @@
|
|
| 1 |
-
|
| 2 |
-
|
| 3 |
-
|
| 4 |
-
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
|
| 15 |
-
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
|
| 19 |
-
|
| 20 |
-
-
|
| 21 |
-
|
| 22 |
-
|
| 23 |
-
|
| 24 |
-
### 4. **Key Improvements Made**
|
| 25 |
-
|
| 26 |
-
#### โ
**Fixed Issues:**
|
| 27 |
-
- Proper requirements.txt formatting (each dependency on new line)
|
| 28 |
-
- Added missing dependencies (opencv, xformers, etc.)
|
| 29 |
-
- Used `@spaces.GPU` decorator for proper GPU allocation
|
| 30 |
-
- Simplified model downloading and setup
|
| 31 |
-
- Added proper error handling and timeouts
|
| 32 |
-
- Reduced frame limits for HF Spaces constraints
|
| 33 |
-
|
| 34 |
-
#### โ
**HF Spaces Optimizations:**
|
| 35 |
-
- Pre-download model weights efficiently
|
| 36 |
-
- Use temporary directories properly
|
| 37 |
-
- Clean up resources after generation
|
| 38 |
-
- Progress indicators for user feedback
|
| 39 |
-
- GPU memory management
|
| 40 |
|
| 41 |
-
|
| 42 |
-
- **First run**: Will take 2-3 minutes to download model (~10GB)
|
| 43 |
-
- **Subsequent runs**: Much faster (model cached)
|
| 44 |
-
- **Recommended**: 25-75 frames for reasonable generation time
|
| 45 |
-
- **Input images**: Works best with landscapes/scenes, max 512px
|
| 46 |
|
| 47 |
-
|
| 48 |
|
| 49 |
-
|
| 50 |
-
- Reduce number of frames (try 25-50)
|
| 51 |
-
- Use smaller input images
|
| 52 |
-
- Restart the Space
|
| 53 |
|
| 54 |
-
|
| 55 |
-
|
| 56 |
-
-
|
| 57 |
-
-
|
|
|
|
| 58 |
|
| 59 |
-
|
| 60 |
-
- This is normal! Video generation is compute-intensive
|
| 61 |
-
- A100 GPU should handle 50 frames in ~2-3 minutes
|
| 62 |
-
- Consider reducing frames for faster results
|
| 63 |
|
| 64 |
-
|
| 65 |
-
- **
|
| 66 |
-
- **25 frames
|
| 67 |
-
- **
|
| 68 |
-
- **75+ frames**: 4-5+ minutes generation
|
| 69 |
|
| 70 |
-
|
| 71 |
|
| 72 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 73 |
|
| 74 |
-
|
| 75 |
|
| 76 |
-
**
|
| 77 |
-
-
|
| 78 |
-
-
|
| 79 |
-
-
|
| 80 |
-
- Fine-grained scene control and evolution
|
| 81 |
|
| 82 |
---
|
| 83 |
|
| 84 |
-
*
|
|
|
|
| 1 |
+
---
|
| 2 |
+
title: Matrix-Game-2.0
|
| 3 |
+
emoji: ๐ฎ
|
| 4 |
+
colorFrom: purple
|
| 5 |
+
colorTo: blue
|
| 6 |
+
sdk: gradio
|
| 7 |
+
sdk_version: 4.42.0
|
| 8 |
+
app_file: app.py
|
| 9 |
+
suggested_hardware: a100-large
|
| 10 |
+
suggested_storage: large
|
| 11 |
+
python_version: 3.10
|
| 12 |
+
short_description: Interactive World Model for Real-Time Video Generation
|
| 13 |
+
tags:
|
| 14 |
+
- video-generation
|
| 15 |
+
- diffusion
|
| 16 |
+
- interactive
|
| 17 |
+
- world-model
|
| 18 |
+
- matrix-game
|
| 19 |
+
models:
|
| 20 |
+
- Skywork/Matrix-Game-2.0
|
| 21 |
+
pinned: false
|
| 22 |
+
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 23 |
|
| 24 |
+
# ๐ฎ Matrix-Game-2.0
|
|
|
|
|
|
|
|
|
|
|
|
|
| 25 |
|
| 26 |
+
An interactive world foundation model for real-time video generation. Upload an image and generate dynamic video content with controllable parameters!
|
| 27 |
|
| 28 |
+
## ๐ How to Use
|
|
|
|
|
|
|
|
|
|
| 29 |
|
| 30 |
+
1. **Upload an Image**: Choose a landscape or scene image (works best)
|
| 31 |
+
2. **Set Parameters**:
|
| 32 |
+
- **Frames**: 25-100 (lower = faster generation)
|
| 33 |
+
- **Seed**: Any number for reproducibility
|
| 34 |
+
3. **Generate**: Click the button and wait 2-5 minutes
|
| 35 |
|
| 36 |
+
## ๐ก Tips
|
|
|
|
|
|
|
|
|
|
| 37 |
|
| 38 |
+
- **First run** takes longer (downloads model weights ~10GB)
|
| 39 |
+
- **Best inputs**: Clear, well-lit outdoor scenes or landscapes
|
| 40 |
+
- **Frame recommendations**: 25-50 frames for faster results
|
| 41 |
+
- **Hardware**: Requires GPU acceleration (A100 recommended)
|
|
|
|
| 42 |
|
| 43 |
+
## โ๏ธ Technical Details
|
| 44 |
|
| 45 |
+
Built on the Matrix-Game-2.0 model by SkyworkAI, this Space provides:
|
| 46 |
+
- Real-time video generation up to 25 FPS
|
| 47 |
+
- Interactive control through diffusion-based synthesis
|
| 48 |
+
- Long video generation capabilities
|
| 49 |
+
- Fine-grained scene evolution
|
| 50 |
|
| 51 |
+
## ๐ง Model Information
|
| 52 |
|
| 53 |
+
- **Base Model**: Skywork/Matrix-Game-2.0
|
| 54 |
+
- **Architecture**: Auto-regressive diffusion
|
| 55 |
+
- **Input**: Single image + parameters
|
| 56 |
+
- **Output**: MP4 video (configurable length)
|
|
|
|
| 57 |
|
| 58 |
---
|
| 59 |
|
| 60 |
+
*Powered by Hugging Face Spaces ๐ค*
|