LSSGen: Leveraging Latent Space Scaling in Flow and Diffusion for Efficient Text to Image Generation

TL;DR

LSSGen introduces a novel latent space scaling method for text-to-image generation that enhances efficiency and image quality without changing model architecture, outperforming traditional pixel-space scaling.

Contribution

It proposes a lightweight latent space upsampler for resolution scaling, improving speed and quality in text-to-image models without modifying core architectures.

Findings

Achieves up to 246% TOPIQ score improvement at similar speeds.

Outperforms conventional pixel-space scaling methods.

Supports flexible multi-resolution generation.

Abstract

Flow matching and diffusion models have shown impressive results in text-to-image generation, producing photorealistic images through an iterative denoising process. A common strategy to speed up synthesis is to perform early denoising at lower resolutions. However, traditional methods that downscale and upscale in pixel space often introduce artifacts and distortions. These issues arise when the upscaled images are re-encoded into the latent space, leading to degraded final image quality. To address this, we propose {\bf Latent Space Scaling Generation (LSSGen)}, a framework that performs resolution scaling directly in the latent space using a lightweight latent upsampler. Without altering the Transformer or U-Net architecture, LSSGen improves both efficiency and visual quality while supporting flexible multi-resolution generation. Our comprehensive evaluation covering text-image alignment and perceptual quality shows that LSSGen significantly outperforms conventional scaling approaches. When generating $1024^2$ images at similar speeds, it achieves up to 246\% TOPIQ score improvement.