Vision & Multimodal Oct 2025 · #16 most cited · 221 citations

Diffusion Transformers with Representation Autoencoders

Boyang Zheng et al.

arXiv:2510.11690

In short

Most diffusion transformers still compress images with an old VAE. This paper swaps in a frozen pretrained representation encoder (such as DINO) plus a trained decoder, and solves the problems of diffusing in its high-dimensional space, reaching state-of-the-art ImageNet image quality with faster training.

Why it matters

It argues the generation and understanding sides of vision should share one latent space.

Read first

The 4 Field Guide ideas this paper leans on.

Neural Networks · Standard Autoencoder An unsupervised neural network that learns to compress data into a latent representation and reconstruct it, useful for dimensionality reduction. Neural Networks · Standard Latent Space A compressed, learned representation space where similar data points are close together, used in autoencoders and VAEs. Vision & Multimodal · Landmark Diffusion Model A generative model that learns to denoise data, achieving state-of-the-art image generation (Stable Diffusion, DALL-E 2). Neural Networks · Landmark Representation Learning Learning useful features or representations of data automatically, rather than hand-crafting them.

Starting from scratch? The full route 11 ideas · basics first

Machine Learning ✓ understood
Building systems that learn patterns from data instead of following hand-written rules, getting better at a task as they see more examples.
Neural Network ✓ understood
A computational model inspired by biological neural networks, consisting of interconnected nodes (neurons) organized in layers that process information through weighted connections.
Unsupervised Learning ✓ understood
Learning from unlabeled data to discover hidden patterns, structures, or relationships without explicit target outputs.
Autoencoder · read first ✓ understood
An unsupervised neural network that learns to compress data into a latent representation and reconstruct it, useful for dimensionality reduction.
Dataset ✓ understood
A collection of data examples used for training, validating, or testing machine learning models.
Feature ✓ understood
A single measurable property of an example, such as a house's floor area or how many links an email contains, used as an input to a model.
Deep Learning ✓ understood
A subset of machine learning that uses neural networks with multiple layers (deep neural networks) to learn hierarchical representations of data.
Representation Learning · read first ✓ understood
Learning useful features or representations of data automatically, rather than hand-crafting them.
Embedding ✓ understood
A list of numbers (a vector) that represents a word, sentence, image or other item, learned so that similar items end up close together.
Latent Space · read first ✓ understood
A compressed, learned representation space where similar data points are close together, used in autoencoders and VAEs.
Diffusion Model · read first ✓ understood
A generative model that learns to denoise data, achieving state-of-the-art image generation (Stable Diffusion, DALL-E 2).

In the frontier

Rank: #16 of 100
Citations: 221; as of Aug 9, 2026
Published: Oct 2025

Topics: Image generation and editing , Model architecture , Efficiency and serving

Selection: 1kpapers.com by Together AI, most-cited as of Aug 9, 2026

Nearby papers

Canon · 2020 Denoising Diffusion Probabilistic Models DDPM is the recipe behind Stable Diffusion, DALL·E, Midjourney and today’s video models. Canon · 2021 High-Resolution Image Synthesis with Latent Diffusion Models It’s the architecture behind Stable Diffusion, which put high-quality text-to-image generation on ordinary GPUs. Frontier · #2 DINOv3 A single frozen backbone that beats specialised models on dense vision tasks makes self-supervised vision a practical default. Frontier · #4 Qwen-Image Technical Report Legible, editable text in generated images was a long-standing weakness; this open model largely fixes it.