AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

Latent Diffusion Model without Variational Autoencoder

Minglei Shi, Haolin Wang, Wenzhao Zheng, Ziyang Yuan, Xiaoshi Wu, Xintao Wang, Pengfei Wan, Jie Zhou, Jiwen Lu

ICLR 2026 · Image generation › Autoregressive and latent generation · approach: Latent diffusion with learned tokenizers · 72 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 SVG feature space instead of VAE latents for diffusionSiT-XL · ImageNet 256 few-step (5 steps, w/ CFG)gFID: 29.48 -> 9.03Improved
use SVG feature space instead of VA-VAE latentsSiT-XL · ImageNet 256 (10 steps, w/o CFG)gFID: 17.41 -> 9.39Improved
use SVG at 25 steps vs SiT (SD) at 250 stepsB/L model scale · ImageNet FID-50K w/ CFGgFID: 13.40 -> 11.49 (B); 6.03 -> 5.96 (L)Improved
add Residual Encoder to frozen DINOv3 (ablation)SVG autoencoder · ImageNet reconstruction (40 epochs)rFID: 1.17 -> 0.78Improved
add distribution alignment of residual features to DINO statistics (ablation)SVG (with Residual Encoder) · ImageNet gFID w/ CFGgFID: 9.03 -> 6.11Improved
add distribution alignment (ablation)SVG autoencoder · ImageNet reconstruction (40 epochs)rFID: 0.78 -> 0.65Improved
use SVG encoder instead of frozen DINOv3 for downstream tasksDINOv3 features · ImageNet-1K / ADE20K / NYUv2Top-1 / mIoU / RMSE: 81.71 -> 81.80 Top-1; 46.37 -> 46.51 mIoUNo clear effect
use shallower DINOv3 layer as feature space (ablation)SVG-B diffusion · ImageNet gFID w/o CFG 200K stepsgFID: 27.00 (layer 11) -> 73.45 (layer 4)Hurt
use SVG feature space for diffusion with 80 epochs vs VA-VAESiT-XL 25 steps · ImageNet 256 w/o CFGgFID: 7.29 -> 6.57Improved
use MAE or SigLIP2 instead of DINOv3 as base encoder (ablation)unified feature space · ImageNet reconstruction (5 epochs)rFID: 1.87 -> 1.69 (MAE); 1.87 -> 4.05 (SigLIP2)Mixed

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