Meta's LSRM Wins ECCV 2026 Honorable Mention, Beats 3D Reconstruction SOTA by 2.4 dB
rsasaki0109 · x · 2026-09-30
LSRM from Meta Reality Labs received an ECCV 2026 Best Paper Honorable Mention. It scales object-centric 3D reconstruction with a Sparse Transformer that handles 20× more 3D object tokens and over 2× more image tokens than prior methods, combined with a coarse-to-fine pipeline and 3D-aware spatial routing to efficiently recover finer geometry and textures.
Results: >2.4 dB PSNR improvement and >40% LPIPS gain over previous SOTA, with inverse-rendering quality matching or beating optimization-based SOTA methods.
Related event: Meta's LSRM Earns ECCV 2026 Best Paper Honorable Mention(2 posts)→
More from Multimodal
- VideoLoop rewrites bounded working memory, hits 88.3% on VideoMME long video — Jinfa Huang · 2026-09-30
- I used Codex to make a 15-second animation — and still had to give it editing notes — naridubs · 2026-09-30
- Flux 3 nails four-way split-screen images of one event from four angles — umesh_ai · 2026-09-30
- MageTrail 2.8B booru finetune costs $593 so far, hits limits of 41k-image dataset — Turbulent-Bass-649 · 2026-09-30
- Filmmakers jam with AI video generation wait times to shoot a music duet with Luma — mrjonfinger · 2026-09-30
- Adobe Research Shows Adversarial Post-Training Restores Missing High-Frequency Detail in Pixel Diffusion — adobe-research · 2026-09-30