Diffusion:Stable Diffusion
Developed by Stability AI, Stable Diffusion, or simply SD, is an open-source generative text-to-image model based on the diffusion technique. While there have been three major versions since 2022, the most frequently used remained the 1.x series until the new XL line was created in 2023. Despite 3.0 being released a year later, XL remains dominant within the community due to the former suffering poor performance.
Releases
[edit source]| Version number | Release date |
|---|---|
| 1.1 | June 2022 |
| 1.2 | June 2022 |
| 1.3 | June 2022 |
| 1.4 | August 2022 |
| 1.5 | October 2022 |
| 2.0 | November 2022 |
| 2.1 | December 2022 |
| XL 1.0 | July 2023 |
| XL Turbo | November 2023 |
| Cascade | February 2024 |
Checkpoints
[edit source]Many popular models have been based on SD’s base, known as Checkpoints, which typically target anime and furry aesthetics. Checkpoints often serve as base models in their own right, due to their specific target audience. YiffyMix and IndigoFurry series are one of the most popular within the Furry Fandom.
Open Source Dilemma
[edit source]Like OpenAI’s pre-3.0 GPT models, Diffusion’s open-source nature and methods used to create such models means that, unlike DALL-E or Midjourney, it is here to stay. The above-mentioned Checkpoints wouldn’t exist had it not been for their open nature. While we understand artists’ concerns, as the founder of this wiki, Tony Bark, is one himself, there is not much he can do. He argued that OpenAI opened Pandora’s box a long time ago with GPT, and their attempt to close the source 3.0 for “safety concerns,” while noble, was too little too late.
Despite Bark’s own creativity, his art skills, while evolving every day, only go so far at the moment, and he needs help. He uses generative art for visualizing ideas and creating concept art that would have previously taken years.