FLUX.2

Provided by FalLearn More

Customizable multi-reference editing with surgical control over transformation quality and prompt adherence. FLUX.2 combines up to 10 reference images with adjustable inference steps and guidance tuning for context-aware edits. Whether compositing multiple product shots, refining typography in existing designs, or blending style references, FLUX.2 delivers professional editing with granular cost-quality control—dial up steps for final deliverables, adjust guidance for creative vs. literal interpretations.

FLUX.2 extends its configurable generation pipeline to image editing, understanding relationships between multiple input images while giving you direct control over the transformation process. Combine reference images, specify exact edits through natural language, and tune generation parameters to balance speed with output quality.

Preview

Inputs

Prompt
text

The prompt to generate an image from. You can reference colors directly in the prompt (e.g. #FF0000), or directly reference fields (e.g. @scene, @style, @camera). You can also use the following JSON example to format your JSON prompt.

Loading...

optional: true
Images
image
Accepts multiple

Images to edit or create variations from. You can use a maximum of 3 images for [dev] model, 9 for [pro], and 10 for [flex] model to combine their features, or use a single image to edit or create variations from.

optional: true
Image Size
dropdown

The size of the generated image.

default: squareAccepts: Square (1:1), Portrait 4:3, Portrait 16:9, Landscape 4:3, Landscape 16:9
Model Type
dropdown

The type of model to use.

default: devAccepts: Dev, Pro, Flex
Steps
number

The number of inference steps to perform.

default: 28minimum: 10maximum: 50
Seed
seed

The same seed and prompt will output the same image every time.

default: 2916
CFG Scale
number

Classifier Free Guidance controls how closely the generated image adheres to the text prompt you provide. Accepts values between 0 and 20.

default: 2.5minimum: 0maximum: 20
Expand Prompt
toggle

Whether to expand the prompt using the model's own capabilities

default: true
Block NSFW
toggle

Prevents the model from generating images that portray nudity, violence, and other unsafe content.

default: false

Outputs

Generated Image
image

The generated image from the model.