Bria Background Generation
Bria Background Generation replaces or generates realistic backgrounds from a text prompt or a reference image while preserving the foreground subject. It’s trained exclusively on licensed data for safe, low-risk commercial use across ads, e-commerce, social, portraits, and product photos.
✨ Highlights
- Dual input: drive backgrounds by prompt or match style/composition with a reference image.
- Subject preservation: clean edges and minimal color bleed around people and products.
- Style range: photorealistic, illustration, anime—swap backgrounds without touching the subject.
- Production-ready: predictable results that slot into real design pipelines.
- Licensed training data: built for compliant commercial deployment.
🧩 Parameters
- 
image* (required)
Foreground/subject image (URL or upload). Missing image will fail. 
- 
prompt* (required)
Background description only (scene, lighting, style, materials, mood).
Example: “Epic anime city at night, neon signs, volumetric fog, blue-purple rim light.” 
🚀 How to Use
- Provide image (required) — a clear, unobstructed subject works best.
- Write prompt (required) — describe the background only: environment, lighting, style.
- Click Run (e.g., $0.04) and download the generated image.
- Keep the same image and tweak the prompt to iterate styles while preserving composition.
💰 Pricing
🧠 Prompting Tips (background-focused)
- Separate roles: keep the subject in the image; keep the background in the prompt.
- Lead with light: phrases like golden hour, studio softbox, neon backlight, volumetric fog quickly lift realism.
- Style anchors: add photorealistic, cinematic, anime, watercolor, or minimalist to control look.
- Materials & space: e.g., concrete rooftop, wet asphalt, marble lobby, skylight, depth of field.
- Consistent campaigns: reuse the same image and a prompt template; adjust color/weather for quick variants.