StabilityAI launches DreamWeaver models tailored for UI and high-fidelity mockups

AI · 5 min read

DreamWeaver models accept a combination of prompts, wireframes, and brand assets and return layered mockups (SVG + raster) with labeled components and CSS hints. The family includes versions optimized for marketing pages, admin dashboards, and mobile apps, each trained with curated UI datasets and design-system annotations.

StabilityAI provides a ready-to-run SDK with export filters for Figma and Sketch, as well as a web-based painter for quick iterations. The company highlights granular style controls—type scale presets, spacing systems, and iconography families—that let design leads enforce visual consistency.

Concerns about copyright and asset reuse are met with filters and provenance metadata that indicate which training sources influenced outputs. Designers note that DreamWeaver’s outputs are often high-quality starting points, but they require polish for accessibility and edge-case interactions.