ComfyUI Node
PSN Generalized Guidance
Applies Power Law CFG and CFG++ methods for alternative guidance approaches. Pure-logic implementation operating on plain Python lists of floats. In ComfyUI, this would patch a MODEL; here it computes the guided noise vector.
PSN Generalized Guidance
- success
- result_json
- reason
◄cond0.000►
◄uncond0.000►
◄method▾►
◄power1.00►
◄cfg_scale7.5►
◄cond_vector[]►
◄uncond_vector[]►
CategoryParametric/Sampling
Inputs (7)
| Name | Type | Default | Description |
|---|---|---|---|
| cond | FLOAT | 0.000-1000000000–1000000000 | Cond. From -1e+09 to 1e+09. Default 0.0. |
| uncond | FLOAT | 0.000-1000000000–1000000000 | Uncond. From -1e+09 to 1e+09. Default 0.0. |
| method | COMBO | Method. One of 3 choices. | |
| power | FLOAT | 1.000–10 | Power. From 0 to 10. Default 1.0. |
| cfg_scale | FLOAT | 7.50–100 | Cfg scale. From 0 to 100. Default 7.5. |
| cond_vectoropt | STRING | [] | Cond vector. Default []. |
| uncond_vectoropt | STRING | [] | Uncond vector. Default []. |
Outputs (3)
| Name | Type | Description |
|---|---|---|
| success | BOOLEAN | — |
| result_json | STRING | — |
| reason | STRING | — |