Stephen Schenck / Android Authority
TL;DR Google Gemini used to accept input letting you specify the shape, or aspect ratio, of images it generates.
Last month, this stopped working, and Gemini will only create new pictures that are square.
Google says that it’s working on a fix, and there are some workaround you can use until then.
If you’ve used modern AI chatbots enough, this probably is going to sound very familiar: You construct a prompt containing one or more very specific restrictions, and then watch as the result you get just utterly ignores those limitations.
I recently had just such an experience, trying to use Gemini to cheat at Blue Prince solve some word puzzles, and even when directing that I wanted only 6-letter answers, still got a bunch of 7- and 8-letter responses. Normally, when you point this out to an AI system, it revises its output accordingly, eventually giving you want you wanted in the first place. But what’s currently going on with Gemini and image generation is proving to be nothing short of maddening.
Don’t want to miss the best from Android Authority? Set us as a favorite source in Google Discover to never miss our latest exclusive reports, expert analysis, and much more.
to never miss our latest exclusive reports, expert analysis, and much more. You can also set us as a preferred source in Google Search by clicking the button below.
Gemini got a big upgrade when it comes to creating and editing images a few weeks back with Nano Banana, and it’s definitely very impressive. But right around the same time, it seems, Gemini has started getting awfully fussy about the shape of the pictures it generates.
Google’s image synthesis tools have long defaulted to making pictures with a square (1:1) aspect ratio. In the past, though, we’ve been able to successfully steer that in other directions — namely widescreen (16:9) output — just by telling the system that’s what we wanted.
... continue reading