Bria FIBO 1.5 Image Edit

Bria FIBO 1.5 Image Edit — Image-to-Image Editing API

What is Bria FIBO 1.5 Image Edit?

Bria FIBO 1.5 Image Edit is a JSON-native, image-to-image editing model from the FIBO family, built for precise, repeatable edits in commercial production. Instead of hoping a loose text prompt lands, FIBO Edit turns your instruction into a detailed structured JSON description of the scene, then performs a deterministic edit on your source image. You send one to four images plus a natural-language instruction, and the model returns an edited image together with the structured_prompt it generated — so every result is transparent, auditable, and reproducible.

Because FIBO is trained exclusively on licensed data, its output is safe for commercial use, which makes it a strong fit for brands, agencies, and automated design pipelines that need legal clarity as well as quality.

Key Features

• •Natural-language editing: describe the change in plain English and let the model resolve lighting, composition, and detail.
• •Structured JSON control: pass a structured_instruction for programmatic, pixel-level, reproducible edits.
• •Multi-reference editing: supply up to 4 images to combine a product with a scene, transfer a style, or keep a subject consistent.
• •Native masking: on single-image edits, a mask restricts changes to white regions while black areas stay untouched.
• •Deterministic renders: fix the seed to recreate an exact edit, or change it to explore variants.

Best Use Cases

• •Product photography: drop a product into a new scene or background while preserving its shape, label, and material.
• •E-commerce and ads: relight scenes, change color palettes, and produce on-brand variations at scale.
• •Text and object edits: add or replace text on signage and packaging, swap objects, or add prints to apparel.
• •Creative restyling: convert sketches to photos, restyle images, or transfer the look of one reference onto another.

In testing, a plain studio product shot was recomposed into a premium spa scene — wet stone, eucalyptus, water droplets, soft light — while the bottle and dropper cap identity were preserved exactly, with clean, artifact-free results.

Prompt Tips and Output Quality

• •Name the target scene precisely and add a clause like keep the subject unchanged to lock product or character identity.
• •For repeatable campaigns, keep a fixed seed and reuse the returned structured_instruction, changing only small attributes.
• •Use mask for localized edits on a single image; use multiple images to merge references.
• •The API returns JSON: the edited image is at output.image_url, alongside seed and structured_prompt.

FAQs

Is Bria FIBO 1.5 Image Edit safe for commercial use? Yes. The FIBO family is trained entirely on licensed data, giving enterprises rights-clear, production-ready output.

How do I reproduce an edit exactly? Send the same source images, the mask if used, the structured_instruction from the original response, and the same seed.

How many input images can I use? One to four. Masking is supported only when you supply exactly one image.

What controls actually change the result? image_urls, instruction, mask, structured_instruction, and seed are the real levers on this version.

What format does the API return? A JSON response containing output.image_url, the seed, and a structured_prompt describing the edited scene.