Edit Images with JSON + Mask Control
Fibo Edit is Bria's JSON-native structured image editing model. Combine JSON prompts + masks + images for deterministic, controllable edits with minimal drift. No more prompt guessing—get reproducible, production-ready results every time.
What is Fibo Edit?
Fibo Edit (Fibo-Edit) is Bria AI's JSON + Mask + Image structured instruction image editing model. Built for deterministic, controllable, production-ready image editing—no prompt roulette, no unpredictable outputs.
JSON-Native Structured Prompts
Define edits with explicit JSON parameters. Change lighting? Specify it. Swap materials? Define it. No surprises.
Mask-Based Precision Editing
Target specific regions with pixel-perfect masks. Edit what you want, preserve everything else.
Deterministic & Reproducible
Same input → same output. Every time. Debug, iterate, and scale with confidence.
{
"image_url": "https://example.com/input.jpg",
"mask_url": "https://example.com/mask.png",
"prompt": {
"edit_type": "recolor",
"target": "product",
"color": "deep navy blue",
"preserve": ["texture", "lighting"]
},
"guidance_scale": 5,
"num_inference_steps": 50
}
Fibo Edit Live Demo
Experience deterministic AI image editing with JSON + Mask control. Upload your image, define your edit, and see Fibo-Edit in action.
Before & After Gallery
See what Fibo Edit can do: recoloring, relighting, material swaps, background changes, and precise region edits.
Product Recoloring
Change product colors while preserving texture and lighting
Scene Relighting
Transform lighting conditions from day to night
Material Swap
Transform wood to metal, fabric to leather, and more
Background Replacement
Seamlessly swap backgrounds for product photos
Object Replacement
Replace objects with text-based descriptions
Style Transfer
Apply artistic styles while preserving composition
Use Cases by Industry
From e-commerce product photography to creative agencies, Fibo Edit powers production workflows across industries.
E-Commerce Product Photography
Transform product images at scale. Generate color variants, swap backgrounds, and create lifestyle scenes without expensive photoshoots.
Marketing & Creative Agencies
A/B test creative variations at scale. Generate multiple campaign variants with consistent quality and deterministic outputs.
Developers & API Integrators
Build production image editing pipelines with clean APIs, structured schemas, and deterministic outputs.
ComfyUI Workflow Users
Integrate Fibo Edit into your node-based workflows. Official partner nodes with downloadable workflow JSON.
Fibo Edit Key Features
Structured image editing capabilities powered by JSON + Mask control for deterministic, production-ready results.
Recolor with Preservation
Change colors while preserving texture, lighting, and material properties. Perfect for generating product color variants.
AI Relighting
Transform scene lighting conditions. Day to night, studio to outdoor, and everything in between with precise control.
Style Transformation
Apply artistic styles while maintaining composition integrity. From photorealistic to illustrated, with predictable outputs.
Material Swap
Transform textures and materials: wood to metal, fabric to leather, matte to glossy. Deterministic material editing.
Object Replacement
Replace objects with text-based descriptions. Fibo Edit handles the context-aware insertion seamlessly.
Mask-Based Precision
Target specific regions with pixel-perfect masks. Edit exactly what you want, preserve everything else.
How Fibo Edit Works
From input to output in 4 simple steps. JSON-native control for deterministic image editing.
Upload Image
Provide your source image via URL or base64 encoding
Define Mask (Optional)
Create a mask to target specific regions. White = edit, Black = preserve
Write JSON Prompt
Specify your edit with structured JSON parameters for deterministic control
Generate Output
Submit via API, ComfyUI, or demo. Get reproducible, production-ready results
JSON Prompt Structure Explained
Understand Fibo Edit's VGL-style structured prompts. Clear schema, predictable outputs.
Key JSON Fields
Fibo Edit uses structured JSON prompts instead of free-form text. This eliminates ambiguity and enables deterministic outputs.
URL to your source image. Supports common formats (JPEG, PNG, WebP).
URL to mask image. White regions = edit, Black regions = preserve.
Structured edit instructions including edit_type, target, and parameters.
Controls adherence to prompt. Higher = stricter following.
Diffusion steps. More steps = higher quality, longer generation.
// Fibo Edit JSON Prompt Example
{
"image_url": "https://example.com/product.jpg",
"mask_url": "https://example.com/mask.png",
"prompt": {
"edit_type": "recolor",
"target": "product surface",
"new_color": "metallic gold",
"preserve": [
"reflections",
"shadows",
"texture"
]
},
"guidance_scale": 5,
"num_inference_steps": 50,
"seed": 42
}
Mask-Based Editing Guide
Learn how to create effective masks for targeted region editing with Fibo Edit.
Clean Edge Mask
Crisp boundaries around the edit region. White = edit area, pure black = preserved area.
Feathered Edge Mask
Soft gradients at edges for smooth blending between edited and preserved regions.
Gray/Partial Values
Avoid gray mid-tones in masks. Use pure white (255) and pure black (0) for predictable results.
Fibo Edit API Quickstart
Copy-paste examples for curl, JavaScript, and Python. Start editing images in minutes.
# Fibo Edit API Request via fal.ai
curl -X POST "https://fal.run/fal-ai/bria/fibo-edit" \
-H "Authorization: Key $FAL_KEY" \
-H "Content-Type: application/json" \
-d '{
"image_url": "https://example.com/input.jpg",
"mask_url": "https://example.com/mask.png",
"prompt": "change the product color to deep blue",
"guidance_scale": 5,
"num_inference_steps": 50
}'
// Fibo Edit API - JavaScript Example
const response = await fetch("https://fal.run/fal-ai/bria/fibo-edit", {
method: "POST",
headers: {
"Authorization": `Key ${FAL_KEY}`,
"Content-Type": "application/json"
},
body: JSON.stringify({
image_url: "https://example.com/input.jpg",
mask_url: "https://example.com/mask.png",
prompt: "change the product color to deep blue",
guidance_scale: 5,
num_inference_steps: 50
})
});
const result = await response.json();
console.log(result.images[0].url);
# Fibo Edit API - Python Example
import requests
import os
response = requests.post(
"https://fal.run/fal-ai/bria/fibo-edit",
headers={
"Authorization": f"Key {os.environ['FAL_KEY']}",
"Content-Type": "application/json"
},
json={
"image_url": "https://example.com/input.jpg",
"mask_url": "https://example.com/mask.png",
"prompt": "change the product color to deep blue",
"guidance_scale": 5,
"num_inference_steps": 50
}
)
result = response.json()
print(result["images"][0]["url"])
📌 API Endpoints
fal.ai: $0.04/image | Bria Platform: Custom pricing | WaveSpeedAI: Variable rates
⚡ Async Processing
For large batches, use async endpoints with request_id and status_url polling.
🔑 Authentication
Get your API key from fal.ai or Bria Platform. Never expose keys in client-side code.
Fibo Edit API Schema
Complete request and response schema for the Fibo Edit image editing API.
| Field | Type | Required | Description |
|---|---|---|---|
| image_url | string | Yes | URL to the source image (JPEG, PNG, WebP supported) |
| mask_url | string | No | URL to mask image. White=edit, Black=preserve |
| prompt | string | object | Yes | Edit instruction - text or structured JSON object |
| guidance_scale | number | No | Prompt adherence strength (default: 5, range: 1-20) |
| num_inference_steps | integer | No | Diffusion steps (default: 50, recommended: 30-75) |
| seed | integer | No | Random seed for reproducibility |
| output_format | string | No | Output format: "png", "jpeg", "webp" (default: png) |
Async Job Processing
Handle long-running Fibo Edit jobs with async patterns for production workflows.
Submit Request
POST to /queue endpoint with your edit parameters
Get Request ID
Receive request_id and status_url immediately
Poll Status
Check status_url until status = "completed"
Get Result
Retrieve output image URL from completed response
// Async polling pattern for Fibo Edit batch processing
async function pollForResult(statusUrl, maxAttempts = 60) {
for (let i = 0; i < maxAttempts; i++) {
const response = await fetch(statusUrl);
const data = await response.json();
if (data.status === "completed") {
return data.output;
}
if (data.status === "failed") {
throw new Error(data.error);
}
await sleep(2000); // Wait 2 seconds between polls
}
throw new Error("Timeout waiting for result");
}
ComfyUI Partner Nodes
Official Fibo Edit nodes for ComfyUI. Build complex editing pipelines with node-based workflows.
Fibo Edit ComfyUI Integration
Fibo Edit is available as an official ComfyUI Partner Node. Integrate deterministic image editing into your existing workflows, chain with ControlNet, LoRA, IPAdapter, and more.
Downloadable Workflows
Pre-built ComfyUI workflow JSON files for common Fibo Edit use cases.
Fibo Edit on Hugging Face
Download model weights for local deployment and research. 8B parameter model with full documentation.
Local Deployment
Run Fibo Edit locally with full model weights. Ideal for research, private deployments, and custom fine-tuning workflows.
⚠️ Hardware Requirements
- 24GB+ VRAM recommended (RTX 4090, A100)
- CPU offloading available for lower VRAM setups
- ~16GB disk space for model weights
- CUDA 11.8+ / PyTorch 2.0+
Deployment Options
Choose the deployment path that fits your workflow: Cloud API, local inference, or hybrid.
Cloud API
Fastest setup. Pay-per-use via fal.ai or Bria Platform. No infrastructure management.
Local Inference
Full control with open weights from Hugging Face. Best for privacy and customization.
Hybrid / Enterprise
Custom deployment via Bria Platform. AWS SageMaker, Azure AI Foundry support.
Performance & Cost
Fibo Edit performance metrics and cost considerations for production planning.
Fibo Edit vs Alternatives
See how Fibo Edit's structured editing approach compares to traditional AI image editors.
| Feature | Fibo Edit | Adobe Firefly | Stable Diffusion Inpaint | Midjourney |
|---|---|---|---|---|
| Edit Control | JSON Structured ✓ | Text Prompt | Text Prompt | Text Prompt |
| Deterministic Output | Yes ✓ | No | Partial (seed) | No |
| Mask Support | Native ✓ | Yes | Yes | Limited |
| API Access | Yes ✓ | Yes | Yes | No Public API |
| Open Weights | Yes (HF) ✓ | No | Yes | No |
| Licensed Training Data | 100% Licensed ✓ | Yes | Mixed/Unknown | Unknown |
| Commercial Safe | Yes ✓ | Yes | Check License | Subscription |
Solving "Prompt Drift"
Traditional text prompts produce unpredictable results. Fibo Edit's JSON structure eliminates guesswork.
The Problem: Traditional Prompts
- Same prompt produces different results each run
- Vague language leads to unintended changes
- "Make it bluer" affects unexpected areas
- No way to specify exactly what to preserve
- Hard to debug why edits failed
- Production workflows become unreliable
The Solution: Fibo Edit JSON
- Deterministic: same input = same output
- Explicit fields define exactly what changes
- "preserve": ["texture", "lighting"] keeps context
- Mask defines edit boundaries pixel-perfect
- Debug by inspecting JSON parameters
- Reproducible results for production at scale
Edit Templates Library
Common editing templates ready to use. Copy, customize, and deploy.
Product Recolor
Change product colors while preserving texture
Scene Relight
Transform lighting conditions
Background Swap
Replace backgrounds seamlessly
Material Change
Wood to metal, fabric to leather
Style Transfer
Apply artistic styles to images
Object Replace
Swap objects with text descriptions
Troubleshooting Common Issues
Solutions to the most common Fibo Edit errors and issues.
Ensure your mask uses pure white (255, 255, 255) for edit regions and pure black (0, 0, 0) for preserved areas. Avoid gray values. The mask must be the same resolution as your input image. PNG format with no compression artifacts works best.
Try increasing guidance_scale (7-12) for stricter prompt following. Be more specific in your JSON prompt about what to change AND what to preserve. Use the "preserve" array to explicitly list attributes that shouldn't change.
Fibo Edit works best with images between 512-1024px on each side. Very large images may be automatically resized. For best results, resize images to ~1024px before processing, then upscale the output if needed.
For production use, implement async polling patterns. Reduce num_inference_steps to 30 for faster (but slightly lower quality) results. Consider using fal.ai's queue endpoints for batch processing.
Ensure your input image is in sRGB color space. Use specific color names or hex codes in your prompt. For exact color matching, provide reference images when possible or use more descriptive color terms.
Release Notes & Changelog
Track Fibo Edit updates, new features, and improvements.
Fibo Edit Initial Release
Launch of Fibo Edit with JSON-native structured prompts, mask-based editing, and open weights on Hugging Face.
ComfyUI Partner Nodes
Official ComfyUI integration launched with partner nodes. Available on Comfy Cloud for instant testing.
Official Resources
Links to official documentation, repositories, and platforms for Fibo Edit.
Frequently Asked Questions
Common questions about Fibo Edit, licensing, and usage.
What is Fibo Edit?
Fibo Edit is Bria AI's 8B parameter image editing model that uses JSON-native structured prompts combined with masks for deterministic, controllable image editing.
Can I use Fibo Edit for commercial projects?
Yes. Fibo Edit is trained on 100% licensed data, making it suitable for commercial use. Check Bria's licensing terms for specific deployment requirements.
What platforms support Fibo Edit?
Fibo Edit is available via fal.ai API, Bria Platform API, ComfyUI Partner Nodes, and Hugging Face open weights for local deployment.
How is Fibo Edit different from Stable Diffusion inpainting?
Fibo Edit uses structured JSON prompts instead of free-form text, enabling deterministic outputs. Same input always produces the same result—no "prompt drift."
What hardware do I need for local deployment?
24GB+ VRAM is recommended (RTX 4090, A100). CPU offloading is available for lower VRAM setups but will be slower.
How much does API access cost?
fal.ai charges approximately $0.04 per image. Bria Platform offers custom pricing for enterprise. Local deployment has no per-image cost.
Community & Support
Connect with other Fibo Edit users, get help, and share your workflows.
Tutorials & Guides
Deep-dive tutorials, use-case guides, and comparison articles.
Getting Started with Fibo Edit JSON Prompts
Complete beginner's guide to structured image editing
E-Commerce Product Photo Automation
Batch process 1000s of product images with Fibo Edit
Fibo Edit vs Adobe Firefly: Which to Choose?
Detailed feature comparison for production workflows