<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Gwern.net — AI: Image Generation (GANs, Diffusion, DALL-E)</title>
    <link>https://gwern.net</link>
    <description>Pages from gwern.net in the 'AI: Image Generation (GANs, Diffusion, DALL-E)' section</description>
    <lastBuildDate>Sat, 07 Mar 2026 22:52:09 +0000</lastBuildDate>
    <generator>GwernRSSBuilder/1.0</generator>
    <atom:link href="https://gwern.net/feed.xml" rel="self" type="application/rss+xml"/>
    <item>
      <title>Danbooru2021: A Large-Scale Crowdsourced &amp; Tagged Anime Illustration Dataset · Gwern.net</title>
      <link>https://gwern.net/danbooru2021</link>
      <guid isPermaLink="true">https://gwern.net/danbooru2021</guid>
      <description>Danbooru2021 is a large-scale anime image database with 4.9m+ images annotated with 162m+ tags; it can be useful for machine learning purposes such as image recognition and generation.</description>
      <author>Gwern</author>
      <pubDate>Tue, 15 Dec 2015 00:00:00 +0000</pubDate>
      <dc:date>2025-02-01T00:00:00Z</dc:date>
      <category>essay</category>
      <category>ai/anime/danbooru</category>
    </item>
    <item>
      <title>Dropcap Generation With AI · Gwern.net</title>
      <link>https://gwern.net/dropcap</link>
      <guid isPermaLink="true">https://gwern.net/dropcap</guid>
      <description>We develop AI image generation workflows for webpage dropcap typography, creating PNGs &amp;amp; SVGs using image generators. As demos, we create new Gwern.net logos and several custom FLOSS dropcap sets, including felines, Gene​ Wolfe horror fiction, and neural-net-inspired dropcaps.</description>
      <author>Gwern</author>
      <pubDate>Sun, 15 Oct 2023 00:00:00 +0000</pubDate>
      <dc:date>2024-10-30T00:00:00Z</dc:date>
      <category>essay</category>
      <category>ai/nn/diffusion/midjourney/dropcap</category>
      <category>ai/nn/transformer/gpt/dall-e/3</category>
      <category>cs/css</category>
      <category>cs/js</category>
    </item>
    <item>
      <title>Midjourney Personalization Flaws, by Gwern · Gwern.net</title>
      <link>https://gwern.net/blog/2024/personalization</link>
      <guid isPermaLink="true">https://gwern.net/blog/2024/personalization</guid>
      <description>Flaws in Midjourney’s new personalization feature: inefficient, noisy, and crude… but still useful and recommended.</description>
      <author>Gwern</author>
      <pubDate>Mon, 22 Jul 2024 00:00:00 +0000</pubDate>
      <dc:date>2024-07-23T00:00:00Z</dc:date>
      <category>essay</category>
    </item>
    <item>
      <title>Making Anime Faces With StyleGAN · Gwern.net</title>
      <link>https://gwern.net/face</link>
      <guid isPermaLink="true">https://gwern.net/face</guid>
      <description>A tutorial explaining how to train and generate high-quality anime faces with StyleGAN 1+2 neural networks, and tips/scripts for effective StyleGAN use.</description>
      <author>Gwern</author>
      <pubDate>Mon, 04 Feb 2019 00:00:00 +0000</pubDate>
      <dc:date>2022-10-19T00:00:00Z</dc:date>
      <category>essay</category>
      <category>ai/anime/danbooru</category>
      <category>ai/nn/gan/stylegan/anime</category>
      <category>cs/python</category>
      <category>tutorial</category>
    </item>
    <item>
      <title>GANs Didn’t Fail, They Were Abandoned, by Gwern Branwen · Gwern.net</title>
      <link>https://gwern.net/gan</link>
      <guid isPermaLink="true">https://gwern.net/gan</guid>
      <description>Diffusion models supposedly beat GANs because they scale better and stabler. That is unproven, and false. GANs should be revisited.</description>
      <author>Gwern Branwen</author>
      <pubDate>Tue, 04 Oct 2022 00:00:00 +0000</pubDate>
      <dc:date>2022-10-15T00:00:00Z</dc:date>
      <category>essay</category>
      <category>ai/nn/gan</category>
      <category>ai/scaling</category>
    </item>
    <item>
      <title>ThisWaifuDoesNotExist.net, by Gwern · Gwern.net</title>
      <link>https://gwern.net/blog/2019/twdne-website</link>
      <guid isPermaLink="true">https://gwern.net/blog/2019/twdne-website</guid>
      <description>N/A</description>
      <author>Gwern</author>
      <pubDate>Tue, 19 Feb 2019 00:00:00 +0000</pubDate>
      <dc:date>2022-05-05T00:00:00Z</dc:date>
      <category>essay</category>
    </item>
    <item>
      <title>Making Anime With BigGAN · Gwern.net</title>
      <link>https://gwern.net/biggan</link>
      <guid isPermaLink="true">https://gwern.net/biggan</guid>
      <description>Experiments in using BigGAN to generate anime faces and whole anime images; semi-successful.</description>
      <author>Gwern</author>
      <pubDate>Mon, 04 Feb 2019 00:00:00 +0000</pubDate>
      <dc:date>2021-01-29T00:00:00Z</dc:date>
      <category>essay</category>
      <category>ai/anime/danbooru</category>
      <category>ai/nn/gan/biggan</category>
      <category>cs/python</category>
      <category>tutorial</category>
    </item>
    <item>
      <title>Anime Neural Net Graveyard · Gwern.net</title>
      <link>https://gwern.net/face-graveyard</link>
      <guid isPermaLink="true">https://gwern.net/face-graveyard</guid>
      <description>Post-mortems of failed neural network experiments in generating anime images, pre-StyleGAN/BigGAN.</description>
      <author>Gwern</author>
      <pubDate>Mon, 04 Feb 2019 00:00:00 +0000</pubDate>
      <dc:date>2021-01-29T00:00:00Z</dc:date>
      <category>essay</category>
      <category>ai/anime/danbooru</category>
      <category>ai/nn/gan/biggan</category>
      <category>ai/nn/gan/stylegan/progan</category>
      <category>ai/nn/vae</category>
    </item>
    <item>
      <title>Anime Crop Datasets: Faces, Figures, &amp; Hands, by Gwern Branwen, Arfafax, Shawn Presser, Anonymous, Danbooru Community · Gwern.net</title>
      <link>https://gwern.net/crop</link>
      <guid isPermaLink="true">https://gwern.net/crop</guid>
      <description>Description of 3 anime datasets for machine learning based on Danbooru: cropped anime faces, whole-single-character crops, and hand crops (with hand detection model).</description>
      <author>Gwern Branwen, Arfafax, Shawn Presser, Anonymous, Danbooru Community</author>
      <pubDate>Sun, 10 May 2020 00:00:00 +0000</pubDate>
      <dc:date>2020-08-05T00:00:00Z</dc:date>
      <category>essay</category>
      <category>ai/anime/danbooru</category>
      <category>ai/dataset</category>
      <category>ai/nn/gan/data-augmentation</category>
      <category>ai/nn/gan/stylegan/anime</category>
      <category>dataset</category>
    </item>
  </channel>
</rss>
