Stable Diffusion

About

Stable Diffusion is a neural network AI that, in addition to generating images based on a textual prompt, can also create images based on existing images. In September 2022, the network achieved virality online as it was used to generate images based on well-known memes, such as Pepe the Frog.

History

On August 10th, 2022, Stability AI^[1] made a limited release for Stable Diffusion,^[2] a text-to-image model able to generate pictures based on a provided textual or visual (image) prompt. On August 22nd, Stability AI^[3] made a full public release of the model, allowing both commercial and non-commercial usage and expressing hope that the model would be used in an "ethical, moral and legal manner" (demonstration video shown below).

Features

In addition to generating images from a text prompt, a feature also available in previously released DALL-E 2 and Midjourney models, Stable Diffusion is also able to generate pictures from a visual prompt (another image). The user must provide a textual prompt to direct the AI, and the degree to which the new image should look like the original can be tuned (original Pepe the Frog image shown below, left; images generated by Stable Diffusion based on it shown below, right).

Use in Memes

Following the release of the model, it gained virality on 4chan's /g/ technology board where multiple daily threads exploring its capabilities were created starting in late August 2022.^[4][5][6]

Beginning as early as September 8th, 2022, users on 4chan^[7] started using the image prompt feature to create images based on iconic memes such as Pepe the Frog, Daily Dose, Girls Laughing and many others. For example, on September 8th, two 4chan^[8][9] users posted Daily Dose images generated by the AI in a /g/ thread (shown below, left and center). On September 11th, a user posted a Booba meme generated by the AI (shown below, right).

These and other meme recreations posted on /g/ and other 4chan boards saw viral spread online in the following days. For example, on September 15th, Facebook^[10] page Burds ARE Real reposted three memes generated by Stable Diffusion, with the post gaining over 390 reactions and 70 shares.

Similar Projects

DALL-E 2

DALL-E 2 is an AI system that can create realistic images and art from a description in natural language. Unveiled by OpenAI as an improved version of DALL-E in April 2022, the AI system gained significant attention as it was used to create art and photorealistic, humorous images. In June 2022, a knock-off free version of DALL-E, DALL-E mini, went viral on social media.

Crayion / DALL-E Mini

DALL-E mini is an AI system that can create images from a description in natural language which is a knock-off, less advanced but openly available version of DALL-E by OpenAI. The project went viral on Twitter in early June 2022 as it was used to generate humorous images, such as characters in unusual settings or characters eating various foods.

Artflow.ai

Artflow.ai is a website that hosts an artificial intelligence program that allows its users to generate digital faces based on short, textual input. Interest in the website increased notably in August 2021.

Various Examples

Search Interest

External References