AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

Latent Diffusion Models with Masked AutoEncoders

Junho Lee, Jeongwoo Shin, Hyungwook Choi, Joonseok Lee

ICCV 2025 · Image generation › Autoregressive and latent generation · approach: Latent diffusion with learned tokenizers · 6 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
add masking loss L_M to MSE autoencoder (ablation)VMAE · ImageNet-1K generation ablationgFID: 12.92 -> 8.92Improved
add latent regularizer L_reg (ablation)VMAE · ImageNet-1K generation ablationgFID: 8.92 -> 6.32Improved
add perceptual loss L_P (ablation)VMAE · ImageNet-1K generation ablationgFID: 6.32 -> 5.98Improved
add masking loss L_M (ablation)VMAE · ImageNet-1K generation ablationrFID: 6.21 -> 4.66Improved
add latent regularizer L_reg (ablation)VMAE · ImageNet-1K generation ablationPSNR: 32.01 -> 31.14Hurt
add perceptual loss L_P (ablation)VMAE · ImageNet-1K generation ablationrFID: 1.59 -> 0.89Improved
add masking loss L_M (ablation)VMAE · ImageNet-1K generation ablationPSNR: 32.18 -> 32.01Hurt
replace SD-VAE with VMAE autoencoderLDM (DiT-B/1) · ImageNet-1K generationgFID: 6.49 -> 5.98Improved
replace SD-VAE with VMAE autoencoderLDM (DiT-B/1) · CelebA-HQ generationgFID: 9.00 -> 7.61Improved
use probabilistic instead of deterministic autoencoder (SD-VAE vs AE)LDM (DiT-B/1) · ImageNet-1K generationgFID: 12.92 -> 6.49Improved
use VAE with KL only (no adversarial/perceptual loss)LDM (DiT-B/1) · ImageNet-1K generationgFID: 34.60 vs 12.92 (AE)Hurt
use VMAE in place of SD-VAE for efficiencyautoencoder · Model efficiencytraining time (hr): 24 -> 9Improved

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