FLUX.1 [dev] Fill with LoRAs
FLUX.1 Fill: Seamless image inpainting with LoRA personalization.
A tradução deste card ainda não está disponível; parte do conteúdo aparece em inglês.
FLUX.1 [dev] Fill with LoRAs is an image generation model specialized in inpainting—replacing or extending parts of an image with AI-generated content. It integrates LoRA adapters, allowing you to apply specific styles, characters, or objects you've trained into the fill process.
Use it when you need to remove unwanted objects, change backgrounds, or add new elements while keeping the rest of the image intact. The model works best with a mask that defines the area to fill, and you can guide the result with a text prompt.
What makes FLUX.1 Fill distinctive is its ability to blend LoRA personalization with inpainting, so you can insert a custom character or style into any scene seamlessly. It's ideal for iterative creative work where consistency and control matter.
- Removing objects or people from photos
- Changing backgrounds while preserving foreground
- Adding custom characters or elements via LoRA
- Extending images beyond their original borders
- Restoring damaged or missing parts of images
Fills match the surrounding image in lighting, texture, and perspective.
Applies custom styles or characters directly into the filled area.
Produces detailed, coherent results even in complex scenes.
- Use a clean, high-contrast mask to define the fill area clearly.
- Keep your text prompt focused on what should appear in the filled region.
- Experiment with different LoRA weights to balance style and realism.
- For best results, start with a lower guidance scale and increase if needed.
- May struggle with very large or complex fill regions.
- Requires a precise mask for best results.
- LoRA adapters need separate training and may not generalize to all contexts.
A mystical forest with glowing mushrooms and a stream, extending the original photo to the right
A cyberpunk street scene at night, neon signs, with my custom LoRA character standing under a streetlamp
Restore the missing corner of an old painting, matching the oil paint texture and 19th century style
A number that sets the starting point for generation; same seed gives reproducible results.
Number of denoising steps; more steps can improve detail but take longer.
Softens the edges of the fill area for smoother blending.
The width-to-height ratio of the output image, like 16:9 or 1:1.
Controls how closely the output follows your prompt; higher values mean stricter adherence.
Os preços exibidos são o que você paga na Muitto. Sem surpresas.
What is inpainting?
Inpainting is the process of filling in a selected area of an image with new AI-generated content.
How do I use a LoRA with this model?
Load a LoRA adapter before generating, and it will influence the style or subject of the filled area.
Can I fill multiple areas at once?
Yes, you can provide a mask with multiple disconnected regions, but results may vary.