Glaze and Nightshade are tools designed to discourage unauthorized scraping of original content by text-to-image models. Glaze works to conceal the stylistic information of your image by applying subtle (usually imperceivable) changes to the image itself. Nightshade can be used to poison ai models by allowing users to insert misleading information about the content of an image.
Link to the University of Chicago’s User Guide
This is my process for creating random but interpretable images that can be used in divination. I use Midjourney via Discord but these principals could potentially apply to other text-to-image models and can be used with Midjourney’s web app as well.
Writing the prompt: Using Austin Osman Spare’s methodology for sigil creation, remove all vowels, repeating letters, spaces, and punctuation from the question, word, phrase, or sentence you seek information on.
Example general div prompt for today’s date: For “July the eleventh twenty twenty-five”, I am left with the string of letters “jlvnthwf”. In this case, I could have also used a string of numbers to represent the date such as “07112025” – the prompt can be really anything as long as the intention is there.
Using Parameters: It is important to use extra parameters in order to achieve the randomness you want in the images – with these applied, Midjourney will also not attempt to re-create text in prompts.
Parameters:
–c 100: encourages different styles and types of images
–weird 3000: encourages more “strangeness” in results
–v 5.2: later Midjourney versions stylize images more aggressively
Today’s Spread:
