GitHub radar
Fibo 2.0: an AI image generator with exact object placement
Fibo 2.0 is BRIA's image model that follows a detailed description instead of a vague prompt, placing objects exactly where you tell it to.
It's an AI image-generation model: instead of a short prompt, it reads a detailed description — light, camera angle, every object — and reproduces the exact count and placement you ask for.
When it helps
Useful for product shots, mockups, and repeatable templates where accuracy matters. Not for selling finished images, since the license covers personal use only.
Pros
- Places the exact number and position of objects, not just roughly
- First version in the family to both edit and generate from scratch
- Ranks 1st for editing and 2nd for generation among open models
Cons
- License covers personal use only — you can't sell the images
- Needs a GPU with 12GB VRAM or a Mac with 16GB+ memory
How to set it up — step by step
- 1Open your AI coding agent (Claude Code, Codex) in your project folder.
- 2Send the agent the text from the block below — it will install ComfyUI (a free app for running these models) and download Fibo 2.0.
- 3Tell the agent to grab the lightweight version for your GPU (12GB+ VRAM) or Mac (16GB+ memory).
- 4Ask it to generate an image with an exact number and placement of objects — say, five apples and one pear.
- 5Check the result: count the objects in the image and compare to what you asked for.
Text for your agent
Copy this and send it to your agent — Claude Code, Codex, any of them:
Set up ComfyUI for me and load the Fibo 2.0 model from https://huggingface.co/Comfy-Org/Fibo-2.0 — pick the lightweight version for my GPU and generate an image with an exact, specified number and arrangement of objects, then show me the result.
You need a GPU with 12 GB of VRAM or a Mac with 16 GB of memory or more — the lightweight version of the ComfyUI package takes around 14 GB on disk.
Open on Hugging Face▌ More finds