Epsilon-VAE: Denoising as Visual Decoding

TL;DR

Epsilon-VAE introduces a novel denoising-based decoding method using diffusion processes, significantly improving image reconstruction and generation efficiency compared to traditional autoencoders.

Contribution

It proposes replacing the traditional decoder with a diffusion process for iterative refinement, enhancing autoencoder performance in visual data compression and generation.

Findings

Achieves high reconstruction quality with diffusion-based decoding.

Enhances downstream generation quality by 22%.

Provides 2.3x inference speedup at higher compression rates.

Abstract

In generative modeling, tokenization simplifies complex data into compact, structured representations, creating a more efficient, learnable space. For high-dimensional visual data, it reduces redundancy and emphasizes key features for high-quality generation. Current visual tokenization methods rely on a traditional autoencoder framework, where the encoder compresses data into latent representations, and the decoder reconstructs the original input. In this work, we offer a new perspective by proposing denoising as decoding, shifting from single-step reconstruction to iterative refinement. Specifically, we replace the decoder with a diffusion process that iteratively refines noise to recover the original image, guided by the latents provided by the encoder. We evaluate our approach by assessing both reconstruction (rFID) and generation quality (FID), comparing it to state-of-the-art autoencoding approaches. By adopting iterative reconstruction through diffusion, our autoencoder, namely Epsilon-VAE, achieves high reconstruction quality, which in turn enhances downstream generation quality by 22% at the same compression rates or provides 2.3x inference speedup through increasing compression rates. We hope this work offers new insights into integrating iterative generation and autoencoding for improved compression and generation.