Bria Eraser

Bria Eraser — AI Object Removal API

What is Bria Eraser?

Bria Eraser is a production-grade image editing API that removes unwanted objects from photos and fills erased regions with contextually accurate, AI-generated pixels. Powered by Bria's inpainting foundation model, it analyzes surrounding lighting, texture, and composition to reconstruct the scene seamlessly — without disturbing original pixels outside the mask.

The API accepts a source image and a mask URL, then returns a clean, inpainted PNG in a single synchronous call at a flat cost of $0.04 per request.

Key Features

• •Precision Object Removal: Supply a manual mask for pixel-perfect control or use automatic masking for fast, AI-detected region removal.
• •Context-Aware Fill: The model replicates surrounding lighting, texture, and tone when regenerating the erased area — no visible seams.
• •Original Pixel Preservation: Only the masked region is modified; all other pixels remain untouched.
• •Content Moderation: Optional input and output moderation flags for compliance-sensitive pipelines in healthcare, finance, and regulated industries.
• •Flat Per-Inference Pricing: $0.04 per call, regardless of image size or complexity.

Best Use Cases

• •E-Commerce Catalogs: Strip watermarks, cables, dust spots, or stray props from product photography at scale.
• •Marketing Asset Production: Remove outdated logos, seasonal elements, or background distractions from hero images without a reshooting budget.
• •Photo Restoration: Erase scratches, date stamps, or blemishes from archive images.
• •Creative App Integration: Embed object removal into consumer or enterprise photo editors with a single API call.
• •Content Localization: Remove region-specific text or branding from master assets before repurposing for other markets.

Prompt Tips and Output Quality

1. •Use high-resolution source images — the model preserves fine texture detail in the filled region.
2. •For manual masks, paint slightly larger than the object to help the model blend fill edges cleanly.
3. •Use mask_type=automatic for quick edits on well-defined objects like logos or isolated props.
4. •Enable both content moderation flags for public-facing or client-delivery pipelines.
5. •If the fill contains visible artifacts at the mask boundary, re-run with a slightly expanded mask.

Bria Eraser Usage Guide

Bria Eraser removes objects from images with a single API call. Supply a source image, a mask indicating what to remove, and the model returns a seamlessly inpainted result.

Quick Start

{
  "image": "https://your-cdn.com/product.jpg",
  "mask": "https://your-cdn.com/product-mask.png",
  "mask_type": "manual",
  "visual_input_content_moderation": true,
  "visual_output_content_moderation": true
}

Parameter Guide

Use Case Recipes

E-Commerce Object Cleanup

Remove stray props, cables, or dust from product images. Use mask_type: manual and paint a tight mask around the object. Enable both moderation flags for client delivery.

Marketing Asset Refresh

Strip outdated branding or seasonal elements from hero images. For large, well-defined regions, mask_type: automatic saves manual mask creation time.

Archive and Photo Restoration

Remove date stamps, scratches, or text overlays from scanned photos. Use a manual mask for precise, pixel-accurate removal.

Mask Quality Tips

• •Paint the mask slightly larger than the target object to give the model room to blend edges.
• •For thin or complex objects (wires, text), increase mask stroke width by 5–10 px.
• •If results show boundary seams, expand the mask by 10–20 px and re-run.

What image formats does Bria Eraser support?

A: The API accepts publicly accessible image URLs in JPEG or PNG format. High-resolution PNGs produce the cleanest inpainting results.

What is the difference between manual and automatic mask types?

A: Manual mask type uses a mask image you supply, giving pixel-level control over the removal area. Automatic mode lets Bria's segmentation model detect and mask the target object — faster for well-isolated elements.

Does erasing an object affect the rest of the image?

A: No. Bria Eraser only regenerates pixels inside the masked region. All original pixels outside the mask are preserved exactly.

How does content moderation work?

A: Setting visual_input_content_moderation: true scans the source image before processing. Setting visual_output_content_moderation: true screens the result before it is returned — both can be enabled independently.

Is this suitable for batch workflows?

A: Yes. With a flat $0.04-per-inference rate and synchronous response, it integrates cleanly into automated pipelines for bulk catalog or asset processing.