GLIDE: Towards Photorealistic Image Generation and Editing with Text-Guided Diffusion Models

TL;DR

This paper introduces GLIDE, a diffusion-based model for photorealistic image generation and editing guided by text prompts, demonstrating superior human-rated quality and capabilities in image inpainting.

Contribution

The paper presents a new diffusion model with classifier-free guidance for text-to-image synthesis and editing, outperforming previous models like DALL-E in photorealism and caption similarity.

Findings

Classifier-free guidance yields more photorealistic images.

GLIDE outperforms DALL-E in human evaluations.

Model can be fine-tuned for image inpainting and editing.

Abstract

Diffusion models have recently been shown to generate high-quality synthetic images, especially when paired with a guidance technique to trade off diversity for fidelity. We explore diffusion models for the problem of text-conditional image synthesis and compare two different guidance strategies: CLIP guidance and classifier-free guidance. We find that the latter is preferred by human evaluators for both photorealism and caption similarity, and often produces photorealistic samples. Samples from a 3.5 billion parameter text-conditional diffusion model using classifier-free guidance are favored by human evaluators to those from DALL-E, even when the latter uses expensive CLIP reranking. Additionally, we find that our models can be fine-tuned to perform image inpainting, enabling powerful text-driven image editing. We train a smaller model on a filtered dataset and release the code and weights at https://github.com/openai/glide-text2im.