ComfyUI Node
TurboDiffusion I2V Sampler
Complete TurboDiffusion I2V inference with dual-expert sampling
TurboDiffusion I2V Sampler
- high_noise_model
- low_noise_model
- conditioning
- vae
- image
- frames
◄num_frames77►
◄num_steps4►
◄resolution480►
◄aspect_ratio16:9►
◄boundary0.9►
◄sigma_max200.00►
◄seed0►
◄use_odefalse►
◄width480►
◄height480►
Categoryvideo/turbodiffusion
Inputs (15)
| Name | Type | Default | Description |
|---|---|---|---|
| high_noise_model | MODEL | High noise expert model from TurboWanModelLoader | |
| low_noise_model | MODEL | Low noise expert model from TurboWanModelLoader | |
| conditioning | CONDITIONING | Text conditioning from CLIPTextEncode | |
| vae | VAE | Wan2.1 VAE from VAELoader | |
| image | IMAGE | Starting image for I2V generation | |
| num_frames | INT | 779–241 | Number of frames to generate (must be 8n+1) |
| num_steps | COMBO | 4 | Number of sampling steps (1-4 for distilled model) |
| resolution | COMBO | 480 | Base resolution or custom |
| aspect_ratio | COMBO | 16:9 | Aspect ratio |
| boundary | FLOAT | 0.90–1 | Timestep boundary for switching from high to low noise model |
| sigma_max | FLOAT | 200.001–1000 | Initial sigma for rCM sampling |
| seed | INT | 00–18446744073709550000 | Random seed |
| use_ode | BOOLEAN | false | Use ODE sampling (sharper but less robust) |
| width | INT | 48064–2048 | Custom width (only used when resolution is 'custom') |
| height | INT | 48064–2048 | Custom height (only used when resolution is 'custom') |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| frames | IMAGE | — |