Google Arts & Culture released “Say What You See “in November as a fun little game that lets you learn about the “art of image prompting” and makes for good practice.
This Google Arts & Culture experiment shows an AI-generated image on the left, and you have to write a prompt, up to 120 characters, that generates a close match. This process takes up to 30 seconds and can show you works from the Arts & Culture library after the fact. Google does not specify which model it’s using. (Imagen 2 was announced in December with “significantly improved image quality” that is “photorealistic, high-resolution, [and] aesthetically pleasing.”)
In Level 1, you “need a 50% match to pass and each level will get more difficult” (60% in Level 2, 70% in Level 3…). Users get “three attempts per image to pass a visual similarity threshold,” with Google providing tips after each image to “craft your prompt,” like “include information about the medium, subject, and context,” as well as specifying art period and materials or shapes and textures.
You can also play this game on mobile, with that link opening the Arts & Culture app if you have it installed. This experience is “available in select countries and will be rolling out to more in the coming months.”
After the game, you can write actual image prompts using the Search Generative Experience (SGE) if you’re enrolled in Google Search Labs.
More Google Arts & Culture:
FTC: We use income earning auto affiliate links. More.
