Home / Changelog

Changelog

What's new in Yugma. Every feature, fix, and improvement we ship.

v0.11.0 Latest

AI Materials & Inspector

  • Added 30 PBR material presets organized across 8 categories (metals, stones, woods, fabrics, glass, plastics, organics, special effects)
  • Introduced AI-powered material generation — describe a material in natural language and the system produces roughness, metalness, color, and emissive values
  • Built material inspector that analyzes any object's current material and suggests improvements or alternatives
  • Added material preset grid panel with real-time preview thumbnails
  • Integrated apply_material_preset as a new AI tool so the Director can use presets during scene composition
v0.10.0

Export Everywhere

  • GLB export for full scenes — compatible with Unity, Unreal Engine, Blender, and web viewers
  • High-resolution screenshot capture with transparent background option
  • Embed code generation — drop a live 3D viewer into any webpage with a single snippet
  • Built public embed viewer at /embed/:shareId with "Made with Yugma" watermark
  • Added export options to the viewport toolbar for quick access
v0.9.0

Advanced Collaboration

  • CollabOverlay now renders remote user cursors in 3D space with color-coded avatars
  • 3D comment pins — click anywhere in the scene to leave spatially anchored comments for your team
  • Real-time activity feed showing who changed what, with object-level granularity
  • Comment resolution workflow: mark comments as resolved to keep the feed clean
  • Improved delta sync performance for rooms with 5+ concurrent users
v0.8.0

Text-to-3D Generation

  • Integrated Meshy.ai for text-to-3D mesh generation — describe an object and get a production-quality GLB model
  • Generation job tracking with progress indicators and status polling
  • One-click placement of generated models into the active scene
  • Generated assets are saved to user's personal asset library for reuse across projects
  • Added generation history panel showing all past text-to-3D jobs
v0.7.0

AI Scene Composer

  • Launched the AI Director with 15 structured tools for scene composition (add, update, remove, duplicate, align, distribute, animate, group, and more)
  • Agentic loop architecture — the AI iteratively calls tools up to 8 times per request, referencing IDs of objects it just created for coherent multi-object scenes
  • Built YSL (Yugma Scene Language) serializer at ~45 tokens per object for efficient AI context
  • Reference resolver handles natural language references ("that", "the red sphere", "everything on the left") and maps them to object IDs
  • Creative/Precise mode toggle: temperature 0.7 for artistic exploration, 0.2 for engineering precision
  • Full system prompt with coordinate system, scale references, material recipes, and design principles
v0.6.0

Real-Time Collaboration

  • Room codes for instant multiplayer sessions — share a 6-character code and collaborate in the same scene
  • Live cursors showing each user's viewport position in real time via Firebase Realtime Database
  • Delta sync broadcasts scene changes (add, update, remove) to all connected peers instantly
  • Host/guest model with permission-aware editing
  • Connection state indicators show who is online and active
v0.5.0

Library Panel

  • Primitive library with all supported geometry types: box, sphere, cylinder, plane, torus, cone, capsule, tetrahedron, dodecahedron
  • GLTF/GLB file upload — drag and drop custom 3D models directly into the scene
  • Sketchfab integration for browsing and importing community 3D models
  • Personal asset library with cloud storage for reusing uploaded models across projects
  • Search and filter across all asset sources from a unified panel