Introduction
API specification for Evolve Image Engine. These APIs provide AI-powered Virtual Try-On (VTO) capabilities for e-commerce, enabling garment try-on on person photos, product placement into scene images, and material visualization on surfaces. <br/><br/> Three VTO modes are supported: <strong>Garment VTO</strong> overlays a product garment (tops, bottoms, footwear, or full outfits) onto a person photo, <strong>Scene VTO</strong> places a product into an environment or room photo using a text-based mask prompt, and <strong>Surface VTO</strong> applies materials and textures (fabric, tile, wood, paint, wallpaper, stone) to surfaces in photos using a text-based mask prompt. <br/><br/> All APIs require authentication with a JWT Bearer token which can be obtained by calling the Evolve Platform Generate access token (`/token`) API.
Garment Virtual Try-On
Overlay a product garment onto a person photo. The source image should contain a person, and the garment image should contain the product to try on. The system will generate a composite image with the garment realistically applied to the person. <br/><br/> Images are automatically padded to a supported aspect ratio for processing and cropped back to the original dimensions in the output. The generated result image is returned inline as base64-encoded image data with its MIME type.
Scene Virtual Try-On
Place a product into a scene or environment photo. The source image should contain the scene (e.g., a room, outdoor space), and the product image should contain the item to place. A target area must be specified using a text-based <code>maskPrompt</code> describing what to replace or where to place the product (e.g., 'chair', 'lamp', 'in the space across from the couch'). <br/><br/> Images are automatically padded to a supported aspect ratio for processing and cropped back to the original dimensions in the output. The generated result image is returned inline as base64-encoded image data with its MIME type.
Surface Virtual Try-On
Apply a material or texture to a surface in a scene photo. The source image should contain the scene (e.g., a room, furniture, architectural space), and the material image should contain a swatch or sample of the material to apply. A target surface must be specified using a text-based <code>maskPrompt</code> describing which surface to apply the material to (e.g., 'sofa cushions', 'the floor', 'the wall behind the bed'). <br/><br/> An optional <code>materialClass</code> can be provided to enable material-specific generation for more realistic results. Supported classes are FABRIC, TILE, WOOD, PAINT, WALLPAPER, and STONE. <br/><br/> Images are automatically padded to a supported aspect ratio for processing and cropped back to the original dimensions in the output. The generated result image is returned inline as base64-encoded image data with its MIME type.