π AI Interior Design - Structure Preserving
Transform your room's style while preserving the original structure, layout, and architectural elements.
π§ Key Features
This app uses advanced image-to-image and ControlNet models that can:
- β Maintain room dimensions and proportions
- β Preserve architectural features (windows, doors, walls)
- β Keep furniture placement and spatial relationships
- β Transform only the style, colors, and materials
π€ Technology Stack
The app leverages multiple state-of-the-art models in priority order:
- FLUX.1-dev ControlNet Union Pro - Advanced ControlNet with multiple control types
- ControlNet Room Segmentation - Specialized for interior design
- FLUX.1 ControlNet Union - High-quality structure preservation
- Stable Diffusion Depth Img2Img - Uses depth information
- Stable Diffusion v1.5 Img2Img - Reliable transformation
- FLUX.1-dev - General purpose fallback
π¨ Style Presets
- Modern Minimalist - Clean lines, neutral colors, natural light
- Luxury Contemporary - High-end materials, elegant finishes
- Scandinavian Style - Light wood, cozy atmosphere, hygge concept
- Industrial Chic - Exposed brick, metal fixtures, urban aesthetic
- Cozy Traditional - Warm colors, classic patterns, timeless design
π How to Use
- Upload your room photo
- Choose a style preset or add custom prompt
- Adjust transformation strength (lower = more preservation)
- Generate your transformed design!
π‘ Tips for Best Results
- Use well-lit, clear photos of rooms
- Avoid blurry or dark images
- Lower strength values preserve more of the original structure
- Try different style presets to find your favorite
π§ Technical Implementation
This application uses:
- Image-to-image transformation for style transfer
- ControlNet technology for structure preservation
- Smart fallback system across multiple models
- Enhanced prompting with architectural preservation keywords
- Base64 image encoding for API compatibility
- Retry logic with exponential backoff
The models are specifically chosen and configured for their ability to preserve room layout and architectural elements while transforming aesthetic style.
Built with β€οΈ using Gradio and Hugging Face Transformers
Inference Providers NEW
This model isn't deployed by any Inference Provider. π Ask for provider support