---
title: 'Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding'
license: CC-BY-4.0
license_url: https://creativecommons.org/licenses/by/4.0/
license_note: 'paper is arXiv non-exclusive distribution license (verified on arXiv abs page 2026-11-12); this pointer note is our own words and is CC BY 4.0; we link to the paper rather than reproduce it'
source: https://iinventedlight.com/library/notes/2205-11487-imagen.md
source_license: 'arXiv non-exclusive distribution license'
attribution: 'Pointer by iinventedlight (AI-assisted). Source: Chitwan Saharia, William Chan, Saurabh Saxena, and 11 coauthors (listed on the arXiv abs page), Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding (arXiv non-exclusive distribution license). This note links to the paper rather than reproducing it.'
author: 'iinventedlight (AI-assisted)'
date: 2026-11-12
status: published
approved_by: "Cole Withington, 2026-11-12 14:44 PT, Librarian chat widget t51s316: \"Yes, publish all 100 second-list arXiv notes\""
checked_on: 2026-11-12
as_of: 2026-11-12
confidence: medium
untested: 'Pointer note only. Paper not reproduced. Abs page and license checked 2026-11-12.'
category: arxiv
topic_area: 'ai-research: multimodal'
tags: arxiv, multimodal, text-to-image, diffusion
arxiv_id: 2205.11487
alias: 2205.11487-imagen
type: pointer
format: pointer
sources:
  - id: S1
    title: 'Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding'
    publisher: arXiv
    authors: 'Chitwan Saharia, William Chan, Saurabh Saxena, and 11 coauthors (listed on the arXiv abs page)'
    url: https://arxiv.org/abs/2205.11487
    version_read: 'v1 (23 May 2022)'
    accessed: 2026-11-12
    license: 'arXiv non-exclusive distribution license'
---
# Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding

## Summary
[Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding](https://arxiv.org/abs/2205.11487) (2022) introduced Imagen, a text-to-image diffusion system that uses a large frozen language model for text understanding plus a diffusion cascade for image generation. The authors argue that a generic text-only language model, such as T5, is a surprisingly strong text encoder for this task.

What it introduced: a photorealistic text-to-image recipe in which language understanding comes from a large pretrained text model and image quality comes from cascaded diffusion.

What it led to: later text-to-image and text-to-video diffusion systems that keep a strong language encoder in front of a diffusion decoder. Live notes: [2103-00020-clip](https://iinventedlight.com/library/notes/2103-00020-clip/). This week's drafts also include [2102-12092-dalle](https://iinventedlight.com/library/notes/2102-12092-dalle/) and [2204-06125-dalle-2](https://iinventedlight.com/library/notes/2204-06125-dalle-2/).

Use it when you want the original Imagen recipe, alongside the DALL-E line. Where to read it: https://arxiv.org/abs/2205.11487 (v1 (23 May 2022)).

The paper is posted on arXiv under the arXiv non-exclusive distribution license. This note links to it rather than reproducing it. The note itself is CC BY 4.0.

License: this note is CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Pointer by iinventedlight (AI-assisted). Paper license: arXiv non-exclusive distribution; this note links to it and does not reproduce it.

---

Agents: [/agents.md](/agents.md) | [/.well-known/agent-library.json](/.well-known/agent-library.json).
