BriaReplaceImageBackground - ComfyUI Built-in Node Documentation

This node replaces the background of an image with a new one generated by Bria. The new background can be described with a text prompt or guided by reference images. The subject’s pixels are preserved while the background is generated around them.

Inputs

Common Inputs

Parameter	Description	Data Type	Required	Range
`image`	The input image whose background will be replaced.	IMAGE	Yes
`background`	Describe the new background with a prompt, or guide it with reference images.	DYNAMIC_COMBO	Yes	`"prompt"` `"reference images"`
`original_quality`	Return the input’s exact pixel size instead of scaling the result to about 1 megapixel. A large input then returns a large image. (default: false)	BOOLEAN	No	`true` `false`
`seed`	The same seed usually returns the same background; the automatic prompt refinement can still vary it. (default: 42)	INT	No	0 to 2147483647
`moderation`	Moderation settings. (default: “false”)	DYNAMIC_COMBO	No	`"false"` `"true"`

Prompt Inputs

Shown when background is set to "prompt".

Parameter	Description	Data Type	Required	Range
`prompt`	Description of the new background. A hex color code such as #FF5733 produces a solid color background. Must be at least 1 character long.	STRING	Yes
`mode`	`high_control` follows the prompt most closely, `base` is a balanced default and `fast` trades detail for speed.	COMBO	Yes	`"high_control"` `"base"` `"fast"`
`refine_prompt`	Rewrite the prompt for better results, which also translates non-English prompts. Turn it off to send the prompt exactly as written. (default: true)	BOOLEAN	No	`true` `false`

Reference Images Inputs

Shown when background is set to "reference images".

Parameter	Description	Data Type	Required	Range
`ref_images`	Growable slot: connect 1 to 10 images guiding the new background; they do not need to share a size. Every reference changes the result, so a few consistent ones beat many conflicting ones. A batched input counts once per image.	IMAGE	Yes	1 to 10 images
`enhance_ref_images`	Extra processing of the reference images for better results. (default: true)	BOOLEAN	No	`true` `false`

Moderation Inputs

Shown when moderation is set to "true".

Parameter	Description	Data Type	Required	Range
`prompt_content_moderation`	Enable moderation for prompt content. (default: false)	BOOLEAN	No	`true` `false`
`visual_input_moderation`	Enable moderation for visual input. (default: false)	BOOLEAN	No	`true` `false`
`visual_output_moderation`	Enable moderation for visual output. (default: false)	BOOLEAN	No	`true` `false`

Note: The ref_images input accepts a maximum of 10 images; providing more than 10 returns an error. The prompt field must contain at least 1 character. When original_quality is false, the input image is downscaled to about 1 megapixel before processing.

Outputs

Output Name	Description	Data Type
`image`	The image with the new background.	IMAGE
`refined_prompt`	The prompt Bria generated from, empty on the reference-image path.	STRING

This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub

Source fingerprint (SHA-256): 62c29d61983c9656d2ea2954518404c62d76961c3deba0e10d63e787d0b0b106

​Inputs

​Common Inputs

​Prompt Inputs

​Reference Images Inputs

​Moderation Inputs

​Outputs