lucataco/demofusion-enhance

Image to Image enhancer using DemoFusion

Capabilities

Reference Images Negative Prompt Seed

Cost

Community model (estimated from hardware time)

Input Parameters

Name	Type	Description	Default	Constraints
`image` *	string (uri)	Input image	`—`	—
`auto_prompt`	boolean	Select to use auto-generated CLIP prompt instead of using the above custom prompt	`false`	—
`cosine_scale_1`	number	Control the strength of skip-residual	`3`	—
`cosine_scale_2`	number	Control the strength of dilated sampling	`1`	—
`cosine_scale_3`	number	Control the strength of the Gaussian filter	`1`	—
`guidance_scale`	number	Scale for classifier-free guidance	`8.5`	min: 1, max: 50
`multi_decoder`	boolean	Use multiple decoders	`false`	—
`negative_prompt`	string	Input Negative Prompt	`"blurry, ugly, duplicate, poorly drawn, deformed, mosaic"`	—
`num_inference_steps`	integer	Number of denoising steps	`40`	min: 1, max: 500
`prompt`	string	Input prompt	`"A high resolution photo"`	—
`scale`	integer	Scale factor for input image	`2`	1 2 4
`seed`	integer	Random seed. Leave blank to randomize the seed	`—`	—
`sigma`	number	The standard value of the Gaussian filter	`0.8`	—
`stride`	integer	The stride of moving local patches	`64`	—
`view_batch_size`	integer	The batch size for multiple denoising paths	`16`	—

image required string

Input image

auto_prompt boolean

Select to use auto-generated CLIP prompt instead of using the above custom prompt

Default: false

cosine_scale_1 number

Control the strength of skip-residual

Default: 3

cosine_scale_2 number

Control the strength of dilated sampling

Default: 1

cosine_scale_3 number

Control the strength of the Gaussian filter

Default: 1

guidance_scale number

Scale for classifier-free guidance

Default: 8.5 min: 1, max: 50

multi_decoder boolean

Use multiple decoders

Default: false

negative_prompt string

Input Negative Prompt

Default: "blurry, ugly, duplicate, poorly drawn, deformed, mosaic"

num_inference_steps integer

Number of denoising steps

Default: 40 min: 1, max: 500

prompt string

Input prompt

Default: "A high resolution photo"

scale integer

Scale factor for input image

Default: 2

1 2 4

seed integer

Random seed. Leave blank to randomize the seed

sigma number

The standard value of the Gaussian filter

Default: 0.8

stride integer

The stride of moving local patches

Default: 64

view_batch_size integer

The batch size for multiple denoising paths

Default: 16

Version: 5bcfe11066c8 Updated: 6/8/2026 11.2K runs