AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

MUSE: Unlocking Timestep as Native Task Steering for One-Step Dense Prediction

Shuo Zhou, Zhaoxin Li, Xiujuan Chai

ECCV 2026 · Image generation › Autoregressive and latent generation · approach: Latent diffusion with learned tokenizers · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
assign distinct timesteps to depth and normal tasks instead of an identical timestep (ablation)MUSE-g · NYUv2 depthAbsRel (%): 26.05 -> 9.67Improved
use high timesteps near 999 instead of 599 for all tasks (ablation)MUSE-g · NYUv2 depthAbsRel (%): 27.05 -> 8.95Improved
vary absolute timestep magnitude in the discriminative variant (ablation)MUSE-d · NYUv2 depthAbsRel (%): 6.31, 6.40, 6.48No clear effect
use native sinusoidal timestep embedding instead of Fourier embedding (ablation)MUSE-g · NYUv2 depthAbsRel (%): 10.02 -> 8.95Improved
use timestep key instead of learnable task token (ablation)MUSE-g · NYUv2 depthAbsRel (%): 12.76 -> 8.95Improved
use timestep key instead of learnable task token (ablation)MUSE-g · iBims-1 normalsmean angular error (deg): 20.83 -> 19.67Improved
train depth and normal jointly instead of depth only (ablation)MUSE-g · NYUv2 depthAbsRel (%): 10.70 -> 11.07No clear effect
train depth and normal jointly instead of normal only (ablation)MUSE-g · iBims-1 normalsmean angular error (deg): 20.01 -> 19.66No clear effect
steer with timesteps 995 and 999 instead of unsteered timesteps (ablation)MUSE-g · Latent clusteringSilhouette score (last layer): -0.0166 -> 0.1572Improved
steer with timesteps 995 and 999 instead of unsteered timesteps (ablation)MUSE-g · Latent clusteringCalinski-Harabasz (last layer): 0.89 -> 80.27Improved
remove injected noise (discriminative instead of generative variant)MUSE · NYUv2 depthAbsRel (%): 5.8 -> 5.1Improved
remove injected noise (discriminative instead of generative variant)MUSE · KITTI depthAbsRel (%): 9.8 -> 9.2Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy