### Advanced Performance Examples

Source: https://github.com/deepbeepmeep/wan2gp/blob/main/docs/CLI.md

Examples showcasing maximum performance with compilation and preloading, optimization for specific GPUs like RTX 2080Ti with TeaCache, and memory-efficient setups.

```bash
# Maximum performance (requires high-end GPU)
python wgp.py --compile --attention sage2 --profile 3 --preload 2000

# Optimized for RTX 2080Ti
python wgp.py --profile 4 --attention sdpa --teacache 2.0

# Memory-efficient setup
python wgp.py --fp16 --profile 4 --perc-reserved-mem-max 0.3
```

--------------------------------

### Server Configuration Examples

Source: https://github.com/deepbeepmeep/wan2gp/blob/main/docs/CLI.md

Examples for setting up a network-accessible server, creating a shareable URL with a custom theme, and locking the configuration for public use.

```bash
# Network accessible server
python wgp.py --listen --server-port 8080

# Shareable server with custom theme
python wgp.py --share --theme gradio --open-browser

# Locked configuration for public use
python wgp.py --lock-config --share
```

--------------------------------

### Verify PyTorch Installation with ROCm

Source: https://github.com/deepbeepmeep/wan2gp/blob/main/docs/AMD-INSTALLATION.md

Run this command to check if PyTorch is correctly installed with ROCm support and to get GPU details.

```cmd
python -c "import torch; print('PyTorch:', torch.__version__); print('ROCm available:', torch.cuda.is_available()); print('Device:', torch.cuda.get_device_name(0) if torch.cuda.is_available() else 'No GPU')"
```

--------------------------------

### Basic Usage Examples

Source: https://github.com/deepbeepmeep/wan2gp/blob/main/docs/CLI.md

Demonstrates launching the tool with specific model presets, enabling compilation for high performance, or optimizing for low VRAM.

```bash
# Launch with specific model and loras
python wgp.py ----lora-preset mystyle.lset

# High-performance setup with compilation
python wgp.py --compile --attention sage2 --profile 3

# Low VRAM setup
python wgp.py --profile 4 --attention sdpa
```

--------------------------------

### Example LoRA Preset Prompt

Source: https://github.com/deepbeepmeep/wan2gp/blob/main/docs/LORAS.md

A basic LoRA preset prompt structure. Trigger words are typically included as comments starting with '#'.

```text
# Use the keyword "ohnvx" to trigger the lora
A ohnvx character is driving a car through the city
```

--------------------------------

### Example Prompt with WanGP Window Commands

Source: https://github.com/deepbeepmeep/wan2gp/blob/main/docs/PROMPTS.md

A practical example demonstrating the usage of various WanGP window commands within a multi-line prompt.

```text
[/duration=25%] A wide dawn shot of a mountain train station. Steam rolls across the platform, red lanterns sway, and the camera slowly pushes toward a violinist waiting beside a blue suitcase.

[/duration=5s,/overlap=17] The violinist steps onto the train as the platform slides backward. Keep the blue suitcase visible, the lantern reflections in the glass, and a gentle handheld camera rhythm.

[/new_shot,/duration=4s] A sharp cut to inside the dining car at night. The violinist now sits across from a silent chess player in a silver coat while rain lashes the window.

[/duration=30%,/overlap=9] The chess player moves a knight, the train lights flicker, and the blue suitcase opens by itself, revealing a tiny glowing city.
```

--------------------------------

### Quick Start: Initialize API and Submit Task

Source: https://github.com/deepbeepmeep/wan2gp/blob/main/docs/API.md

Demonstrates how to initialize the WanGP API session, define generation settings, submit a task, and process progress, preview, and stream events. It also shows how to retrieve the final job result and handle potential errors.

```python
from pathlib import Path

from shared.api import init

session = init(
    root=Path(r"C:\\WanGP"),
    cli_args=["--attention", "sdpa", "--profile", "4"],
)

settings = {
    "model_type": "ltx2_22B_distilled",
    "prompt": "Cinematic shot of a neon train entering a rainy station",
    "resolution": "1280x704",
    "num_inference_steps": 8,
    "video_length": 97,
    "duration_seconds": 4,
    "force_fps": 24,
}

job = session.submit_task(settings)

for event in job.events.iter(timeout=0.2):
    if event.kind == "progress":
        progress = event.data
        print(progress.phase, progress.progress, progress.current_step, progress.total_steps)
    elif event.kind == "preview":
        preview = event.data
        if preview.image is not None:
            preview.image.save("preview.png")
    elif event.kind == "stream":
        line = event.data
        print(f"[{line.stream}] {line.text}")

result = job.result()
if result.success:
    print(result.generated_files)
else:
    for error in result.errors:
        print(error.message)

```

--------------------------------

### Macro System Syntax and Example

Source: https://github.com/deepbeepmeep/wan2gp/blob/main/docs/PROMPTS.md

Define variables and their values on a single line starting with '!'. Subsequent lines use these variables as templates. WanGP cycles through values, reusing shorter lists cyclically if variable lengths differ.

```text
! {Variable1}="valueA","valueB" : {Variable2}="valueC","valueD"
This is a template using {Variable1} and {Variable2}.
```

--------------------------------

### Install SageAttention

Source: https://github.com/deepbeepmeep/wan2gp/blob/main/docs/AMD-INSTALLATION.md

Install the SageAttention library, which is required for certain attention implementations. Use a version less than 2.

```cmd
pip install "sageattention <2"
```