Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding

TL;DR

Imagen is a state-of-the-art text-to-image diffusion model that combines large language understanding with high-fidelity image generation, achieving photorealism and superior alignment with text prompts, without training on target datasets.

Contribution

The paper introduces Imagen, a novel diffusion model that leverages large pretrained language models for improved text understanding and image synthesis, setting new benchmarks in photorealism and alignment.

Findings

Achieves a new SOTA FID score of 7.27 on COCO without training on it.

Human raters find Imagen samples comparable to real COCO images.

Outperforms recent models like DALL-E 2 in quality and alignment.

Abstract

We present Imagen, a text-to-image diffusion model with an unprecedented degree of photorealism and a deep level of language understanding. Imagen builds on the power of large transformer language models in understanding text and hinges on the strength of diffusion models in high-fidelity image generation. Our key discovery is that generic large language models (e.g. T5), pretrained on text-only corpora, are surprisingly effective at encoding text for image synthesis: increasing the size of the language model in Imagen boosts both sample fidelity and image-text alignment much more than increasing the size of the image diffusion model. Imagen achieves a new state-of-the-art FID score of 7.27 on the COCO dataset, without ever training on COCO, and human raters find Imagen samples to be on par with the COCO data itself in image-text alignment. To assess text-to-image models in greater depth, we introduce DrawBench, a comprehensive and challenging benchmark for text-to-image models. With DrawBench, we compare Imagen with recent methods including VQ-GAN+CLIP, Latent Diffusion Models, and DALL-E 2, and find that human raters prefer Imagen over other models in side-by-side comparisons, both in terms of sample quality and image-text alignment. See https://imagen.research.google/ for an overview of the results.