ComfyUI Node
DDRK Omega Auto (one-click)
One-click DDRK sampling. Connect model, prompts and latent; the node recognises the model (SD 1.5, SDXL, SD3, Flux, Qwen-Image, Chroma, Wan and more) and chooses steps, CFG, schedule and integrator itself. The 'settings' output says exactly what it chose. Set steps or CFG above 0 to override them.
DDRK Omega Auto (one-click)
- model
- positive
- negative
- latent_image
- latent
- settings
◄seed0►
◄qualitybalanced►
◄model_typeauto►
◄steps0►
◄cfg0.0►
◄denoise1.00►
Categorysampling/DDRK Omega
Inputs (10)
| Name | Type | Default | Description |
|---|---|---|---|
| model | MODEL | — | |
| positive | CONDITIONING | — | |
| negative | CONDITIONING | — | |
| latent_image | LATENT | — | |
| seed | INT | 00–18446744073709550000 | — |
| quality | COMBO | balanced | fast: ~0.6x the steps. balanced: the model's usual step count. best: EDM (SD/SDXL) 1.5x the steps; Flow Matching (Flux, SD3, Qwen...) adds a second pass at 1.33x resolution, ~1.7x the time and more VRAM. |
| model_type | COMBO | auto | Pick 'turbo / lightning / few-step' for Turbo, Lightning, Hyper, LCM, DMD or schnell-style models and LoRAs: 4-8 steps at CFG 1. 'auto' recognises few-step base models (Flux schnell, Z-Image Turbo) but cannot see a turbo LoRA or finetune. |
| steps | INT | 00–200 | 0 = automatic for this model and quality. Anything above 0 is used as is. |
| cfg | FLOAT | 0.00–30 | 0 = automatic for this model (e.g. SDXL 6, SD3 4.5, Flux 1). Anything above 0 is used as is. |
| denoiseopt | FLOAT | 1.000.01–1 | 1.0 for text-to-image. Below 1.0 for img2img: how much of the input latent is redrawn. |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| latent | LATENT | The sampled latent - connect it to VAE Decode. |
| settings | STRING | What the node chose: model, steps, CFG, integrator, schedule. |