AI Paper Atlas

Research map / Video / Video generation

MotionStreamer: Streaming Motion Generation via Diffusion-based Autoregressive Model in Causal Latent Space

Lixing Xiao, Shunlin Lu, Huaijin Pi, Ke Fan, Liang Pan, Yueer Zhou, Ziyong Feng, Xiaowei Zhou, Sida Peng, Jingbo Wang

ICCV 2025 · Video › Video generation · approach: Text-to-motion diffusion · 69 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
use continuous Causal TAE latents instead of VQ-VAE tokens (ablation)MotionStreamer · HumanML3D testFID: 13.226 -> 11.790Improved
use causal VAE (Causal TAE) instead of non-causal VAE (ablation)MotionStreamer · HumanML3D testFID: 19.902 -> 11.790Improved
use compressed KL-regularized latent instead of plain autoencoder (ablation)MotionStreamer · HumanML3D testFID: 43.828 -> 11.790Improved
add QK normalization in the AR transformer (ablation)MotionStreamer · HumanML3D testFID: 12.324 -> 11.790Improved
add Two-Forward training (ablation)MotionStreamer · HumanML3D testFID: 12.944 -> 11.790Improved
add diffusion head instead of MSE loss regression (ablation)MotionStreamer · HumanML3D testFID: 59.296 -> 11.790Improved
use T5-XXL text encoder instead of CLIP (ablation)MotionStreamer · HumanML3D testFID: 14.234 -> 11.790Improved
use root loss weight lambda 7.0 instead of 5.0 or 9.0 (ablation)Causal TAE · HumanML3D testreconstruction FID: 0.696 -> 0.661Improved
use Causal TAE with hidden size 1024 instead of 512 at latent dim 16 (ablation)Causal TAE · HumanML3D testgeneration FID: 14.096 -> 11.790Improved
increase latent dimension (larger dimension, less compression) (ablation)Causal TAE · HumanML3D testreconstruction vs generationMixed
use CFG scale 4.0 over no guidance (ablation)MotionStreamer · HumanML3D testgeneration qualityImproved
use continuous latent AR model instead of VQ-LLaMA discrete variantMotionStreamer · BABEL subsequenceR@3: 0.383 -> 0.568Improved

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