Bria FIBO Edit Gemini Prompt to JSON

This modular pipeline block uses Gemini to turn a plain-language image-edit instruction and its reference image or images into the structured JSON consumed by briaai/Fibo-Edit-1.5-base and briaai/Fibo-Edit-1.5-turbo.

Set your Gemini API key before loading the block:

export GEMINI_API_KEY="your-api-key"

Usage

from diffusers.modular_pipelines import ModularPipelineBlocks
from PIL import Image

block = ModularPipelineBlocks.from_pretrained(
    "briaai/FIBO-edit-gemini-prompt-to-JSON",
    trust_remote_code=True,
).init_pipeline()

# Single reference
image = Image.open("source.png")
output = block(image=image, prompt="Change the car color to green")
json_prompt = output.values["json_prompt"]

# Multiple ordered references
references = [Image.open("subject.png"), Image.open("setting.png")]
output = block(
    image=references,
    prompt="Place the subject from <image_1> in the setting from <image_2>",
)
json_prompt = output.values["json_prompt"]

# Masked edit (one reference only)
mask = Image.open("mask.png")
output = block(image=image, mask_image=mask, prompt="Replace the masked region with flowers")
json_prompt = output.values["json_prompt"]
Downloads last month
113
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support