Meta's LSRM wins ECCV 2026 Best Paper Honorable Mention, scales 3D reconstruction with sparse attention
rsasaki0109 · x · 2026-09-30
Meta Reality Labs introduced LSRM (Large Sparse Reconstruction Model), which earned a Best Paper Honorable Mention at ECCV 2026, studying how scaling transformer context windows improves feed-forward 3D reconstruction.
Key finding: expanding the context window — 20× more active object tokens and over 2× more image tokens than prior SOTA — markedly closes the gap with dense-view optimization in fine-grained texture and appearance, enabling high-fidelity 3D object reconstruction and inverse rendering.
Three technical contributions:
- Coarse-to-fine pipeline that predicts sparse high-resolution residuals to focus compute on informative regions;
- 3D-aware spatial routing establishing accurate 2D-3D correspondences via explicit geometric distances instead of attention scores;
- Block-aware sequence parallelism with an All-gather-KV protocol to balance dynamic sparse workloads across GPUs.
On novel-view synthesis benchmarks LSRM delivers >2.4 dB higher PSNR and >40% lower LPIPS than prior SOTA; extended to inverse rendering, its LPIPS matches or beats SOTA dense-view optimization. Paper and code are available.
Related event: Meta's LSRM Earns ECCV 2026 Best Paper Honorable Mention(2 posts)→
More from Multimodal
- Hand-editing 0.2% of weights plus 100 CPU-minutes turns an English TTS model into an 18MB Sanskrit one — kastnerkyle · 2026-11-16
- Zyphra open-sources textnorm tool, lifting TTS structured-data accuracy by up to 22 points — kastnerkyle · 2026-11-16
- Testing Qwen Image 2.1 and Minimax H3 on anime-to-photo conversion and film photo restoration — Ant_6431 · 2026-11-16
- Second take of a 'Cubee run' video generated with Minimax H3 — JahJedi · 2026-11-16
- Non-artist builds 14-chapter hand-painted storybook game with Fable 5.2 in 4 hours — jradoff · 2026-11-16
- Subject LoRAs on Minimax H3 Image learn poor likenesses while identical datasets excel on Qwen 2.1 — Ad_Organic · 2026-11-16