Papers.Data.Code
Paper

GAE: Learning a Geometry-Native Latent Space for 3D-Consistent World Generation

Jiahao Lu, Minghao Yin, Wenbo Hu, Hengyu Liu, Wang Zhao, Sai-Kit Yeung, Ying Shan, Yuan Liu

Replaces appearance-led latents with a compact geometry-native latent from frozen 3D features, then uses a standard conditional flow to generate views and geometry.

arXiv ↗Sep 21, 2026Code ↗

  • cv
  • world generation
  • 3d consistency
  • latent representation
Composite of multiple 3D reconstructions with depth maps showing method outputs across diverse scenes
Composite of multiple 3D reconstructions with depth maps showing method outputs across diverse scenes

What it is

A geometry-native autoencoder compresses 3D foundation-model features into one latent for world generation, improving both video quality and 3D consistency. Instead of adding geometry to appearance latents, it makes the generator evolve a compact latent reparameterized from a geometry foundation model, jointly decodable to RGB, depth, cameras, and point maps.

Key results

  • FVD drops 23.1% on DL3DV vs the strongest competing latent
  • FVD drops 12.7% on RealEstate10K vs the strongest competing latent
  • Camera-trajectory error is roughly halved on RealEstate10K

Shows the latent representation itself is a major bottleneck for 3D-consistent generation, not just the generator architecture.

One useful ML email a week.

More from Sep 21 – Sep 27

The whole week →

Picked Sep 25

  • Dataset
    microsoft/ProgramDistill

    Unlike GUI-agent benchmarks, it evaluates coding agents by restoring masked web-app features with replayable browser traces and executable verifiers across 4,063 tasks.

    Hugging FaceSep 24, 2026
    →
  • Paper
    PACT: From Credit Assignment to Critic Alignment

    PACT gets 72.87% avg on 4 math benchmarks, +8.80 over GRPO and +13.16 over PPO; 67.4% on SWE-bench Verified, +2.0 over GRPO.

    arXivSep 22, 2026
    →
  • Paper
    Recursive self-improvement of AI research agents

    In 8 days it found 7 accepted rewrites, raised grade 0.703→0.778, matched/exceeded a human-built agent on 4 held-out benchmarks, and cut reward hacking 55%→32%.

    arXivSep 22, 2026
    →
  • Repo
    TencentARC/GameHorizon

    Covers 5,000 gameplay hours from 21 AAA titles, 5,000 offline MCQs, 20 online tasks, and benchmarks 47 models in one suite.

    GitHubSep 20, 2026
    →

Picked Sep 24

  • Dataset
    Harland/OmniVChat

    Makes native audio-visual dialogue measurable without text queries, ASR, or captioning, with fine-grained rubric scores shared across eval and RL.

    Hugging FaceSep 18, 2026
    →
  • Repo
    jaredpalmer/kev

    Unlike recent typed-decision repos, it packs many question types into one calibrated, decode-free pass with 2.0× speedup and 4e-6 output parity.

    GitHubSep 17, 2026
    →

Worth your time, not all of your time.

We track new ML work across papers, code, and data, then keep only the pieces worth a closer look. Selected, not collected.

Papers.Data.Code

The Weekly ML Signal

Selected, not collected.

One useful ML email a week: the papers, repos and datasets worth your attention.

Worth your time, not all of your time.