ComfyUI Node
FUSE KSampler
A ComfyUI node in Face Enhancement with 22 inputs and 2 outputs.
FUSE KSampler
- model
- vae
- images
- positive
- negative
- mask_optionals
- yolo_optionals
- images
- masks
◄use_cachetrue►
◄seed0►
◄steps20►
◄cfg8.00►
◄sampler_name▾►
◄scheduler▾►
◄denoise0.50►
◄yolo_detector▾►
◄sam_segmenter▾►
◄sam_model_typevit_b►
◄face_id0►
◄face_orderlinear►
◄face_size512►
◄face_padding20►
◄force_squaretrue►
CategoryFace Enhancement
Inputs (22)
| Name | Type | Default | Description |
|---|---|---|---|
| model | MODEL | Base model for sampling. | |
| vae | VAE | VAE model for the sampler. | |
| images | IMAGE | Input image batch for face detection and upscaling. | |
| positive | CONDITIONING | Positive conditioning for the sampler. | |
| negative | CONDITIONING | Negative conditioning for the sampler. | |
| use_cache | BOOLEAN | true | Use internal caching to speed up workflow iteration. |
| seed | INT | 00–18446744073709550000 | Seed for the sampler. Used for determinisitc results with some models. |
| steps | INT | 20 | Number of steps for the sampler. |
| cfg | FLOAT | 8.00 | Classifier-Free Guidance scale for the sampler. |
| sampler_name | COMBO | Choose the sampler to use for sampling. | |
| scheduler | COMBO | Choose the scheduler to use for sampling. | |
| denoise | FLOAT | 0.500–1 | Denoising strength for the sampler. When doing img2img like FUSE, lower values adhere more to the input image (face). |
| yolo_detector | COMBO | Choose the YOLO detector to use for face detection. Must be trained on faces. See: https://github.com/akanametov/yolo-face | |
| sam_segmenter | COMBO | Choose the SAM segmentation model to use for face segmentation. See: https://github.com/facebookresearch/segment-anything?tab=readme-ov-file#model-checkpoints | |
| sam_model_type | COMBO | vit_b | SAM model type |
| face_id | INT | 0 | Index of the face to process in the image (0 is the first face found, -1 is all faces). |
| face_order | COMBO | linear | Order to process detected faces: linear (top-to-bottom, left-to-right), linear_reverse (bottom-to-top, right-to-left), largest_bbox (largest face first), smallest_bbox (smallest face first). |
| face_size | COMBO | 512 | The resolution to sample the face crop at. |
| face_padding | INT | 200–16384 | Padding in pixels (int) to pad the face crop with. |
| force_square | BOOLEAN | true | Force 1:1 square face crops |
| mask_optionalsopt | DICT | Optional masking and blending settings from FUSESamplerMaskOptions node. | |
| yolo_optionalsopt | DICT | Optional YOLO detection settings from FUSEYOLOSettings node. |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| images | IMAGE | — |
| masks | MASK | — |