PIXART-{\delta}: Fast and Controllable Image Generation with Latent Consistency Models

TL;DR

PIXART-{ extdelta} is a rapid, controllable, high-resolution text-to-image synthesis framework that leverages Latent Consistency Models and ControlNet, achieving unprecedented inference speed and efficiency.

Contribution

It introduces PIXART-{ extdelta}, combining LCM and ControlNet for fast, controllable, high-quality image generation at 1024px resolution with efficient training and inference.

Findings

Achieves 0.5s inference time for 1024x1024 images

Enables training on 32GB V100 GPUs within a day

Supports 8-bit inference on 8GB GPUs

Abstract

This technical report introduces PIXART-{\delta}, a text-to-image synthesis framework that integrates the Latent Consistency Model (LCM) and ControlNet into the advanced PIXART-{\alpha} model. PIXART-{\alpha} is recognized for its ability to generate high-quality images of 1024px resolution through a remarkably efficient training process. The integration of LCM in PIXART-{\delta} significantly accelerates the inference speed, enabling the production of high-quality images in just 2-4 steps. Notably, PIXART-{\delta} achieves a breakthrough 0.5 seconds for generating 1024x1024 pixel images, marking a 7x improvement over the PIXART-{\alpha}. Additionally, PIXART-{\delta} is designed to be efficiently trainable on 32GB V100 GPUs within a single day. With its 8-bit inference capability (von Platen et al., 2023), PIXART-{\delta} can synthesize 1024px images within 8GB GPU memory constraints, greatly enhancing its usability and accessibility. Furthermore, incorporating a ControlNet-like module enables fine-grained control over text-to-image diffusion models. We introduce a novel ControlNet-Transformer architecture, specifically tailored for Transformers, achieving explicit controllability alongside high-quality image generation. As a state-of-the-art, open-source image generation model, PIXART-{\delta} offers a promising alternative to the Stable Diffusion family of models, contributing significantly to text-to-image synthesis.