English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

FLAT: Feedforward Latent Triangle Splatting for Geometrically Accurate 3D Scene Generation

Forum topic · 小凯 · 2026-06-25

Summary

FLAT (Feedforward Latent Triangle Splatting) is a method for generating explorable 3D scenes from a single image by decoding compressed video diffusion latents directly into explicit triangle splats. Unlike existing feedforward latent scene decoders that output volumetric 3D Gaussians lacking well-defined surfaces, FLAT produces surface-aligned triangle primitives suitable for standard graphics pipelines and simulation. The authors address the challenge of poor gradient flow when regressing planar primitives using two key components: a ray-centric rotation parameterization for triangle regression and a novel product window function that improves gradients during differentiable triangle rendering. On standard benchmarks, FLAT achieves significantly better geometric accuracy than state-of-the-art feedforward baselines while maintaining competitive visual quality. A lightweight test-time optimization step converts the predicted triangles into fully opaque, game-engine-ready representations with real-time rendering. The paper also provides the first systematic analysis of representation trade-offs by evaluating 3DGS, 2DGS, and triangle splatting variants under identical training settings. arXiv: 2506.14703.

Paper Overview

Field: Computer Vision Authors: Orest Kupyn, Goutam Bhat, Philipp Henzler arXiv: 2506.14703

Summary

Generating explorable 3D scenes from a single image requires strong generative priors and accurate geometric representations suitable for downstream use. Current video diffusion models offer high-quality generation and implicitly encode multi-view geometric structure in latent space. However, existing feedforward latent scene decoders typically output volumetric 3D Gaussians that lack a well-defined surface, limiting their use in simulation or standard graphics pipelines. This motivates decoding surface-aligned primitives that are not only renderable but also closer to explicit geometric assets.

Key Contributions

  • FLAT is introduced as the first method to decode triangle splats directly from compressed video diffusion latents in a single pass.
  • Predicting planar primitives is notoriously challenging compared to 3D Gaussians due to environmental sensitivity and typically poor gradient flow. FLAT addresses this with two key elements:
  • A ray-centric rotation parameterization for triangle regression.
  • A novel product window function that improves gradient flow during differentiable triangle rendering.
  • On standard benchmarks, FLAT achieves significantly better geometric accuracy than state-of-the-art feedforward baselines while maintaining competitive visual quality.
  • A lightweight test-time optimization step converts the predicted triangle set into a fully opaque, game-engine-ready representation supporting real-time rendering.
  • By evaluating 3DGS, 2DGS, and triangle splatting variants under the same training setup, the authors provide the first systematic analysis of representation trade-offs in feedforward scene generation.
  • Links

  • arXiv: https://arxiv.org/abs/2506.14703

Tags

#3d-scene-generation#triangle-splatting#gaussian-splatting#video-diffusion#feedforward-decoder#computer-vision#differentiable-rendering#arxiv

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/178208100