{"id":8080,"date":"2023-02-07T14:55:47","date_gmt":"2023-02-07T21:55:47","guid":{"rendered":"https:\/\/mattfife.com\/?p=8080"},"modified":"2023-02-07T15:05:26","modified_gmt":"2023-02-07T22:05:26","slug":"stable-diffusion-2-0-was-well","status":"publish","type":"post","link":"https:\/\/mattfife.com\/?p=8080","title":{"rendered":"Stable diffusion 2.0 was…well…"},"content":{"rendered":"\n

Stable Diffusion\u00a02.0\u00a0seems to have been a step backwards<\/a> in capabilities and quality. Many people went back to\u00a0v1.5<\/a>\u00a0for their business.<\/p>\n\n\n\n

The difficulty in 2.0 was in part caused by:<\/p>\n\n\n\n

    \n
  1. Using a new language model that is trained from scratch<\/li>\n\n\n\n
  2. The training dataset was heavily censored with a NSFW filter<\/li>\n<\/ol>\n\n\n\n

    The second part would have been fine, but the filter was quite inclusive and has removed substantial amount of good-quality data. 2.1 promised to bring the good data back.<\/p>\n\n\n\n

    Installing Stable Diffusion 2.1<\/h2>\n\n\n\n

    If you’re interested in trying Stable Diffusion 2.1, use this tutorial to installing and use 2.1 models<\/a> in AUTOMATIC1111<\/a> GUI, so you can make your judgement by using it.<\/p>\n\n\n\n

    You might also try this tutorial by TingTing<\/a><\/p>\n\n\n\n

    \n