AI Paper Atlas

Research map / Video / Video segmentation and tracking

OneVAE: Joint Discrete and Continuous Optimization Helps Discrete Video VAE Train Better

Yupeng Zhou, Zhen Li, Yuming Chen, Ziheng Ouyang, Ruoyi Du, Daquan Zhou, Bin Fu, Yihao Liu, Peng Gao, Ming-Ming Cheng, Qibin Hou

ECCV 2026 · Video › Video segmentation and tracking · approach: Learned video compression

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 FSQ instead of VQ as quantizer (ablation)continuous VAE latent space · Quantizer on frozen continuous VAE latentPSNR: 22.82 -> 26.46Improved
add progressive training from pretrained continuous VAE (vs FSQ from scratch) (ablation)discrete video VAE · PSNR vs training iterationsconvergence speed: 5x convergence speedupImproved
add progressive training from pretrained continuous VAE (vs FSQ from scratch) (ablation)discrete video VAE · PSNR vs training iterationsPSNR: approximately 1dBImproved
use FSQ instead of SimVQ in progressive training (ablation)progressive discrete VAE · PSNR vs training iterationsPSNR: SimVQ saturates at a lower PSNRImproved
expand decoder with extra layers (ablation)16x16x4 discrete OneVAE · 16x16x4 discrete VAE decoder ablationPSNR: 24.86 -> 25.05Improved
expand decoder with extra layers (ablation)16x16x4 discrete OneVAE · 16x16x4 discrete VAE decoder ablationSSIM: 0.7950 -> 0.8003Improved
add multi-token quantization (two tokens per position) (ablation)discrete OneVAE 8x8x4 · Panda70M 33x512x512 (discrete)PSNR: 30.67 -> 31.80Improved
add multi-token quantization (two tokens per position) (ablation)discrete OneVAE 8x8x4 · Panda70M 33x512x512 (discrete)FVD: 108.93 -> 78.35Improved
add first-frame enhancement (ablation)16x16x4 discrete OneVAE · Panda70M 33x512x512 (discrete)PSNR: 27.40 -> 28.11Improved
add first-frame enhancement (ablation)16x16x4 discrete OneVAE · Panda70M 33x512x512 (discrete)FVD: 336.48 -> 213.04Improved
use OneVAE instead of Cosmos Tokenizer (discrete 8x8x4)discrete video VAE · Panda70M 33x512x512 (discrete)FVD: 82.86 -> 108.93Hurt
add joint discrete and continuous dual-path training (ablation)discrete VAE · latent feature visualizationlatent richnessImproved

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