Wavelets Are All You Need for Autoregressive Image Generation

TL;DR

This paper introduces a novel autoregressive image generation method using wavelet-based tokenization combined with a specialized transformer architecture, enabling efficient modeling of image details across multiple resolutions.

Contribution

The work presents a new wavelet-based tokenization scheme and a tailored transformer model for autoregressive image generation, improving detail representation and statistical correlation modeling.

Findings

Effective wavelet tokenization captures multi-resolution image details.

Transformer architecture learns significant statistical correlations.

Experimental results demonstrate successful conditioned image generation.

Abstract

In this paper, we take a new approach to autoregressive image generation that is based on two main ingredients. The first is wavelet image coding, which allows to tokenize the visual details of an image from coarse to fine details by ordering the information starting with the most significant bits of the most significant wavelet coefficients. The second is a variant of a language transformer whose architecture is re-designed and optimized for token sequences in this 'wavelet language'. The transformer learns the significant statistical correlations within a token sequence, which are the manifestations of well-known correlations between the wavelet subbands at various resolutions. We show experimental results with conditioning on the generation process.