lucataco/ip-adapter-faceid

(Research only) IP-Adapter-FaceID can generate various style images conditioned on a face with only text prompts

Capabilities

1:1 4:3 3:4 16:9 9:16 2:3 3:2 Negative Prompt Seed

Community model (estimated from hardware time)

Name	Type	Description	Default	Constraints
`face_image` *	string (uri)	Input face image	`—`	—
`agree_to_research_only`	boolean	You must agree to use this model only for research. It is not for commercial use.	`false`	—
`height`	integer	Height of output image	`1024`	—
`negative_prompt`	string	Input Negative Prompt	`"monochrome, lowres, bad anatomy, worst quality, low quality, blurry"`	—
`num_inference_steps`	integer	Number of denoising steps	`30`	min: 1, max: 200
`num_outputs`	integer	Number of images to output	`1`	min: 1, max: 4
`prompt`	string	Input prompt	`"photo of a woman in red dress in a garden"`	—
`seed`	integer	Random seed. Leave blank to randomize the seed	`—`	—
`width`	integer	Width of output image	`1024`	—

face_image required string

Input face image

agree_to_research_only boolean

You must agree to use this model only for research. It is not for commercial use.

Default: false

height integer

Height of output image

Default: 1024

negative_prompt string

Input Negative Prompt

Default: "monochrome, lowres, bad anatomy, worst quality, low quality, blurry"

num_inference_steps integer

Number of denoising steps

Default: 30 min: 1, max: 200

num_outputs integer

Number of images to output

Default: 1 min: 1, max: 4

prompt string

Input prompt

Default: "photo of a woman in red dress in a garden"

seed integer

Random seed. Leave blank to randomize the seed

width integer

Width of output image

Default: 1024

Version: fb81ef963e74 Updated: 2/26/2026 32.2K runs