Add jelly blob animation video generation setup

[salihcamal2-art]

Overview

Add comprehensive video generation setup for a Pixar-quality jelly blob children's animation using Open-Sora framework.

What's Included

This PR adds complete documentation and tooling to generate a delightful 3D animated scene featuring:

Characters:

• WOBBO: Bright yellow translucent jelly blob kid with big glossy eyes and tiny stubby arms
• DOT: Tiny purple grape-sized baby jelly blob with sparkle particles
• Scene captures DOT giggling "Boo!" and popping in a sparkle burst, with WOBBO reacting in comic shock

Animation Details:

• Pixar-quality 3D render with glossy gelatin surfaces and soft inner glow
• Fast playful pacing with squash-stretch animation effects
• Vibrant saturated candy colors with bright cheerful lighting
• Target: 256×256 or 512×512 video output, 2-4 seconds duration

Files Added

• jelly_blob_prompt.txt - Detailed animation prompt for video generation
• QUICK_START.md - Fast setup guide (5-7 minutes to video)
• GENERATE_VIDEO_README.md - Comprehensive guide with all setup options
• setup_wsl.sh - Automated WSL installation script for Windows users
• run_docker.sh - Docker container setup and execution
• download_models.sh - Pretrained model downloader
• Dockerfile.opensora - Docker configuration for containerized generation
• generate_jelly_blob_video.py - Python helper for generation
• Jelly_Blob_Video_Generation.ipynb - Google Colab notebook (ready-to-run)

Generation Options

Multiple ways to generate the video:

1. Google Colab (Recommended for fastest start - 5 min, free GPU)
2. Windows WSL - Full setup with CUDA support
3. Docker - Containerized environment with GPU support
4. Local Python - Direct installation on Linux/Mac

Key Features

✅ Multiple setup options - Choose method that works for your environment
✅ Complete documentation - Step-by-step guides with troubleshooting
✅ Ready-to-run notebooks - Google Colab notebook for instant generation
✅ Hardware requirements - Clear specs and time estimates
✅ Model management - Helper scripts for downloading pretrained models

Time Estimates

• Colab: ~5-7 minutes total (install + generate + download)
• WSL: ~30 minutes setup + 3 minutes generation
• Docker: ~15 minutes setup + 3 minutes generation
• Output: 256×256 video (~2 sec) or 512×512 (~4 sec)

Testing

To test the setup:

# Option 1: Google Colab (Recommended)
- Open Colab notebook at: https://colab.research.google.com
- Upload Jelly_Blob_Video_Generation.ipynb
- Run all cells in order

# Option 2: WSL
bash setup_wsl.sh
python scripts/inference.py --config configs/opensora/inference/16x256x256.py --prompt-path jelly_blob_prompt.txt

# Option 3: Docker
bash run_docker.sh

Notes

• Includes detailed prompt text tuned for Open-Sora's text-to-video capabilities
• Models are downloaded on-demand (no pre-bundled weights)
• Generated videos automatically saved to outputs/jelly_blob_video/
• Supports both MP4 and GIF output formats