ComfyUI Node
HengeFrameRefine
A ComfyUI node in Henge with 19 inputs and 1 output.
HengeFrameRefine
- model
- vae
- sampler
- sigmas
- frames
- clip
- positive
- negative
- IMAGE
◄encoder▾►
◄width1024►
◄height1024►
◄seed0►
◄cfg1.0►
◄denoise0.45►
◄carry0.85►
◄deflicker0.30►
◄color_matchtrue►
◄promptEnhance this image with sharp, crisp details while keeping the content, composition and colors identical.►
◄moss_edit_format►
CategoryHenge
Inputs (19)
| Name | Type | Default | Description |
|---|---|---|---|
| model | MODEL | — | |
| vae | VAE | — | |
| encoder | COMBO | How the current source frame reaches the model: its own edit encoder per frame (needs clip), as a reference latent on positive/negative, or only through the init latent. | |
| sampler | SAMPLER | — | |
| sigmas | SIGMAS | Full schedule; denoise keeps its last part. | |
| frames | IMAGE | Source video frames. | |
| width | INT | 102464–8192 | — |
| height | INT | 102464–8192 | — |
| seed | INT | 00–18446744073709550000 | — |
| cfg | FLOAT | 1.00–30 | — |
| denoise | FLOAT | 0.450.05–1 | — |
| carry | FLOAT | 0.850–1 | How much of the warped previous output seeds each frame where the flow is reliable. |
| deflicker | FLOAT | 0.300–1 | Final pull of reliable regions onto the warped previous output. |
| color_match | BOOLEAN | true | — |
| clipopt | CLIP | — | |
| promptopt | STRING | Enhance this image with sharp, crisp details while keeping the content, composition and colors identical. | — |
| positiveopt | CONDITIONING | — | |
| negativeopt | CONDITIONING | — | |
| moss_edit_formatopt | STRING | MOSS-Image edit layout (TextEncodeMOSSImageEdit edit_format); empty = its first option. |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| IMAGE | IMAGE | — |