Meissonic: Revitalizing Masked Generative Transformers for Efficient High-Resolution Text-to-Image Synthesis

TL;DR

Meissonic advances masked image modeling for text-to-image synthesis, achieving high-resolution, high-quality images comparable to diffusion models through architectural innovations and optimized training strategies.

Contribution

Introduces Meissonic, a novel masked image modeling approach with architectural and training improvements that surpass existing methods in high-resolution text-to-image synthesis.

Findings

Matches or exceeds SDXL performance in image quality.

Produces $1024 imes 1024$ resolution images.

Demonstrates superior efficiency and fidelity in experiments.

Abstract

We present Meissonic, which elevates non-autoregressive masked image modeling (MIM) text-to-image to a level comparable with state-of-the-art diffusion models like SDXL. By incorporating a comprehensive suite of architectural innovations, advanced positional encoding strategies, and optimized sampling conditions, Meissonic substantially improves MIM's performance and efficiency. Additionally, we leverage high-quality training data, integrate micro-conditions informed by human preference scores, and employ feature compression layers to further enhance image fidelity and resolution. Our model not only matches but often exceeds the performance of existing models like SDXL in generating high-quality, high-resolution images. Extensive experiments validate Meissonic's capabilities, demonstrating its potential as a new standard in text-to-image synthesis. We release a model checkpoint capable of producing $1024 \times 1024$ resolution images.