Type: huggingface.image_to_image.FluxFill

Namespace: huggingface.image_to_image

Description

Performs image inpainting/filling using FLUX Fill models with support for GGUF quantization. image, inpainting, fill, flux, quantization, mask

Use cases:
- Fill masked regions in images with high-quality content
- Remove unwanted objects from images
- Complete missing parts of images
- Memory-efficient inpainting using GGUF quantization
- High-quality image editing with FLUX models

Properties

Property Type Description Default
model hf.inpainting The FLUX Fill model to use for image inpainting. {'type': 'hf.inpainting', 'repo_id': 'black-forest-labs/FLUX.1-Fill-dev', 'path': None, 'variant': None, 'allow_patterns': None, 'ignore_patterns': None}
prompt str A text prompt describing what should fill the masked area. a white paper cup
image image The input image to fill/inpaint {'type': 'image', 'uri': '', 'asset_id': None, 'data': None}
mask_image image The mask image indicating areas to be filled (white areas will be filled) {'type': 'image', 'uri': '', 'asset_id': None, 'data': None}
height int The height of the generated image. 1024
width int The width of the generated image. 1024
guidance_scale float Guidance scale for generation. Higher values follow the prompt more closely 30.0
num_inference_steps int Number of denoising steps 50
max_sequence_length int Maximum sequence length for the prompt. 512
seed int Seed for the random number generator. Use -1 for a random seed. -1
enable_cpu_offload bool Enable CPU offload to reduce VRAM usage. True

Outputs

Output Type Description
output image  

Metadata

Browse other nodes in the huggingface.image_to_image namespace.