mirror of https://github.com/Wan-Video/Wan2.1.git synced 2025-11-04 06:15:17 +00:00

Add professional-grade Docker setup for local deployment

This commit introduces comprehensive Docker support for running Wan2.1
video generation models locally with GPU acceleration.

Changes:
- Add Dockerfile with CUDA 12.1 support and optimized layer caching
- Add docker-compose.yml for easy container orchestration
- Add .dockerignore for efficient Docker builds
- Add DOCKER_SETUP.md with detailed setup and troubleshooting guide
- Add DOCKER_QUICKSTART.md for rapid deployment
- Add docker-run.sh helper script for container management
- Update Makefile with Docker management commands

Features:
- Full GPU support with NVIDIA Docker runtime
- Single-GPU and multi-GPU (FSDP + xDiT) configurations
- Memory optimization flags for consumer GPUs (8GB+)
- Gradio web interface support on port 7860
- Volume mounts for models, outputs, and cache
- Comprehensive troubleshooting and optimization guides
- Production-ready security best practices

The Docker setup supports all Wan2.1 models (T2V, I2V, FLF2V, VACE)
and includes both 1.3B (consumer GPU) and 14B (high-end GPU) variants.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

2025-10-26 03:06:50 +00:00

3.8 KiB

Raw Blame History

Wan2.1 Docker Quick Start

Get Wan2.1 running in Docker in 5 minutes!

Prerequisites

Docker 20.10+ installed (Get Docker)
NVIDIA GPU with 8GB+ VRAM (for GPU acceleration)
NVIDIA Docker runtime installed (Install Guide)

Quick Start (3 Steps)

Step 1: Clone and Navigate

git clone https://github.com/Wan-Video/Wan2.1.git
cd Wan2.1

Step 2: Build and Start

Option A: Using the helper script (Recommended)

./docker-run.sh start

Option B: Using Make

make docker-build
make docker-up

Option C: Using Docker Compose directly

docker compose up -d wan2-1

Step 3: Download Models and Run

# Enter the container
./docker-run.sh shell
# OR
make docker-shell
# OR
docker compose exec wan2-1 bash

# Download a model (1.3B for consumer GPUs)
pip install "huggingface_hub[cli]"
huggingface-cli download Wan-AI/Wan2.1-T2V-1.3B --local-dir /app/models/Wan2.1-T2V-1.3B

# Generate your first video!
python generate.py \
  --task t2v-1.3B \
  --size 832*480 \
  --ckpt_dir /app/models/Wan2.1-T2V-1.3B \
  --offload_model True \
  --t5_cpu \
  --sample_shift 8 \
  --sample_guide_scale 6 \
  --prompt "A cute cat playing with a ball of yarn"

# Your video will be in /app/outputs (accessible at ./outputs on your host)

Common Commands

Container Management

# Start container
./docker-run.sh start

# Stop container
./docker-run.sh stop

# Restart container
./docker-run.sh restart

# View logs
./docker-run.sh logs

# Enter shell
./docker-run.sh shell

# Check status
./docker-run.sh status

Using Make Commands

make docker-up        # Start
make docker-down      # Stop
make docker-shell     # Enter shell
make docker-logs      # View logs
make docker-status    # Check status
make help            # Show all commands

Run Gradio Web Interface

# Inside the container
cd gradio
python t2v_14B_singleGPU.py --ckpt_dir /app/models/Wan2.1-T2V-1.3B

# Open browser to: http://localhost:7860

Available Models

Model	VRAM	Resolution	Download Command
T2V-1.3B	8GB+	480P	`huggingface-cli download Wan-AI/Wan2.1-T2V-1.3B --local-dir /app/models/Wan2.1-T2V-1.3B`
T2V-14B	24GB+	720P	`huggingface-cli download Wan-AI/Wan2.1-T2V-14B --local-dir /app/models/Wan2.1-T2V-14B`
I2V-14B-720P	24GB+	720P	`huggingface-cli download Wan-AI/Wan2.1-I2V-14B-720P --local-dir /app/models/Wan2.1-I2V-14B-720P`
I2V-14B-480P	16GB+	480P	`huggingface-cli download Wan-AI/Wan2.1-I2V-14B-480P --local-dir /app/models/Wan2.1-I2V-14B-480P`

Troubleshooting

"CUDA out of memory"

Use the 1.3B model with --offload_model True --t5_cpu
Reduce resolution to 480P

"nvidia-smi not found"

Ensure NVIDIA Docker runtime is installed
Run: docker run --rm --gpus all nvidia/cuda:12.1.1-base-ubuntu22.04 nvidia-smi

Can't access Gradio interface

Check if port 7860 is exposed: docker ps | grep 7860
Try: http://127.0.0.1:7860 instead of localhost

Next Steps

Read the full DOCKER_SETUP.md for advanced configuration
Check the main README.md for model details
Join the Discord community

File Structure

Wan2.1/
├── models/          # Downloaded models (created automatically)
├── outputs/         # Generated videos (accessible from host)
├── cache/           # Model cache
├── Dockerfile       # Docker image definition
├── docker-compose.yml  # Container orchestration
├── docker-run.sh    # Helper script
├── Makefile         # Make commands
└── DOCKER_SETUP.md  # Detailed documentation

Happy Generating! 🎬

3.8 KiB Raw Blame History