AI Paper Atlas

Research map / Image generation / Text-to-image diffusion

MotionRFT: Unified Reinforcement Fine-Tuning for Text-to-Motion Generation

Xiaofeng Tan, Wanjiang Weng, Hongsong Wang, Fang Zhao, Xin Geng, Liang Wang

TPAMI 2026 · Image generation › Text-to-image diffusion · approach: Preference optimization for diffusion · 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
fine-tune MLD with EasyTune (step-wise reward)MLD · HumanML3DFID: 0.450 -> 0.132Improved
add full MotionRFT (MotionReward rewards) on top of EasyTuneMLD · HumanML3DFID: 0.132 -> 0.101Improved
apply MotionRFT to joint-based ACMDMACMDM · HumanML3DFID: 0.109 -> 0.084Improved
apply MotionRFT to rotation-based HY-MotionHY-Motion · HumanML3DFID: 1.258 -> 0.985Improved
apply MotionRFT to rotation-based HY-MotionHY-Motion · HumanML3DR-Precision Top 1: 0.445 -> 0.499Improved
fine-tune with EasyTune/MotionRFT (semantic alignment sharpens per-prompt consistency)MLD · HumanML3DMultiModality: 2.267 -> 1.621Hurt
use unified MotionReward instead of representation-specific TMR/ReAlign retrievalReAlign · HumanML3D retrieval, rotationAvg. recall: 82.01 -> 85.17Improved
use unified MotionReward instead of per-representation MotionCriticMotionCritic · MotionCritic test, kinematicaccuracy: 85.08 -> 86.09Improved
step-wise EasyTune instead of trajectory-level ReFL-10MLD · HumanML3DFID: 0.207 -> 0.132Improved
EasyTune instead of ReFL-10 on memoryMLD · HumanML3Dpeak memory (GB): 22.10 -> 22.10No clear effect
EasyTune one-step ODE prediction reward on MLD++MLD++ · HumanML3DFID: 0.073 -> 0.073No clear effect
EasyTune on MDM (SDE, noise-aware reward)MDM · HumanML3DFID: 0.489 -> 0.411Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 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