TRELLIS Launches in Foundry Labs, Bringing Generative 3D to More Creators

TRELLIS Launches in Foundry Labs, Bringing Generative 3D to More Creators
Source: Microsoft (screenshot from TRELLIS video)
  • TRELLIS generates editable 3D models from text or image prompts using Microsoft’s new SLAT format.
  • Microsoft’s Asha Sharma announced its release in Azure Foundry Labs, a sandbox for early-stage AI tools.

Microsoft has introduced TRELLIS to Azure AI Foundry Labs, a new generative system for creating high-quality 3D assets from simple inputs. Designed to lower the barrier to spatial content creation, TRELLIS produces detailed, editable models for use in AR, games, simulation, and design tools.

TRELLIS runs on a Structured Latents (SLAT) format that captures geometry and texture in a flexible, compact representation. Trained on 500,000 diverse 3D objects, it supports output types like meshes, radiance fields, and 3D Gaussians, making it compatible across rendering pipelines.

Azure AI Foundry Labs is Microsoft’s experimental R&D sandbox that gives developers, startups, and enterprises early access to emerging AI research projects. Head of AI Platform Products, Asha Sharma, announced TRELLIS’s addition to Foundry Labs on LinkedIn, encouraging creators to explore its prompt-based editing and fast iteration capabilities.


🌀 Tom’s Take:

Generative 3D is advancing fast. Systems like TRELLIS are set to change the game for spatial content development by removing the biggest barrier to entry: 3D asset creation. By pairing generative AI with existing 3D tools, developers and creators can move from concept to prototype faster than ever. This acceleration not only boosts creative momentum, but it may also help teams secure funding earlier to scale projects and assemble full production teams.


Source: LinkedIn - Asha Sharma