Make-A-Scene

Overview

Make-A-Scene is a multimodal generative AI method from Meta that lets users describe and illustrate their vision through both text input and freeform sketches. It combines a text prompt with a spatial layout from a rough sketch to produce an image that follows the user's composition intent.

Best for

Best for
Artists, designers, and researchers who want to guide AI image generation with sketch-based layout control.

Use cases

• Generate an image from a text description while controlling object placement with a simple sketch
• Iteratively refine an AI-generated scene by editing the sketch layout
• Explore creative concepts by combining written prompts with hand-drawn scene layouts

Notes

Make-A-Scene is a multimodal generative AI method from Meta that lets users describe and illustrate their vision through both text input and freeform sketches. It combines a text prompt with a spatial layout from a rough sketch to produce an image that follows the user’s composition intent.

Use cases

• Generate an image from a text description while controlling object placement with a simple sketch
• Iteratively refine an AI-generated scene by editing the sketch layout
• Explore creative concepts by combining written prompts with hand-drawn scene layouts

Pros

• Gives users direct control over spatial composition not possible with text-only models
• Supports both text and sketch input for flexible creative expression
• Open research release enables community experimentation and customization

Cons

• Not a polished end-user product; requires technical knowledge to run or adapt
• Output quality and consistency may lag behind commercial image generators
• Limited to research release with no official support or updates

Indexed from awesome-generative-ai and enriched against its public facts.