# Apple — Luce: Relightable Gaussians for 3D Asset Generation

- Company: Apple (apple.com)
- Announced: 2026-08-26
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://machinelearning.apple.com/research/relightable-gaussians-3d-generation
- Record: https://forck.live/items/10308-luce-relightable-gaussians-for-3d-asset-generation
- Subject: Machine Learning Research

Apple researchers propose Luce, a 3D representation that unifies geometry and physically based rendering (PBR) materials within a voxelized multimodal Gaussian cloud. A rectified-flow transformer generates this representation from a single image, and the latent decodes into relightable PBR Gaussians and an optional textured mesh. On the Toys4K benchmark, Luce achieves state-of-the-art single-image-to-3D generation, improving FID by 28% over the strongest baseline.

## Evidence

Verbatim from https://machinelearning.apple.com/research/relightable-gaussians-3d-generation:

> We propose Luce, a 3D representation that unifies geometry and PBR materials within a voxelized multimodal Gaussian cloud, using dedicated Gaussian primitives for each modality.

---

Record: https://forck.live/items/10308-luce-relightable-gaussians-for-3d-asset-generation
Catalogue: https://forck.live/llms.txt
Current issue: https://forck.live/feed.md