Create task black-forest-labs/flux-fill-dev
Official documentation: https://replicate.com/black-forest-labs/flux-fill-dev
/replicate/v1/models/black-forest-labs/flux-fill-dev/predictionsBody Parametersapplication/json
Input content
propertiesImage generation guidance information. Default value: 30. Minimum value: 0, maximum value: 100.
Image to be repaired. Can include alpha masks. If the image width or height is not a multiple of 32, it is scaled to the nearest multiple of 32. If image size exceeds 1440x1440, it will be reduced to fit 1440x1440
Determine the applicable strength of the main LoRA. For basic reasoning, a reasonable result is between 0 and 1. For go_fast, we apply a 1.5x multiplier to this value; typically performance is good when scaling the base value by this multiple. You may still need to experiment to find the best value for your specific LoRA. Default value: 1. Minimum value: -1, maximum value: 3.
A black and white image describing the part of the image that needs to be restored. Black areas will be preserved and white areas will be repaired.
The approximate number of pixels of the generated image. Use match_input to match the size of the input (up to 1440x1440 pixels), default: "1".
Number of denoising steps. The recommended range is 28-50, with fewer steps the lower the output quality but the faster the speed. Default value: 28. Minimum value: 1, maximum value: 50.
The amount of output to generate. Default value: 1. Minimum value: 1, maximum value: 4.
The format of the output image. Default: "webp".
The quality at which the output image is saved, from 0 to 100. 100 is the best quality, 0 is the lowest quality. Has nothing to do with .png output. Default: 80
Prompt to generate image.
Responses
Creation time
Whether sensitive data has been removed
Error message
Resource or task ID
Input content
Run logs
Model name / model ID
Output result
Status
List of resource URLs
Version
