Finally got to spend some time with the new Photoshop 2026 and the new Generative Fill using Nano Banana. As expected, the results are a mixed bag.
On one hand: Given the amount of time spent versus the results generated, it’s hard to ignore the possibilities. I can do in minutes what used to take hours. You can visualize a concept much, much faster. You can swiftly make sweeping changes to clothes, hair, accessories and backgrounds. If you have a photo that’s “almost what we want”, you can quickly generate your way to “yeah, that’ll work”. I took a photo of the CN tower and in about 15 minutes I had transformed it into a pretty interesting poster design.
BUT– what gets generated can be unpredictable at best and completely insane at worst. The AI can’t quite understand the full context of what you’re trying to do, and it kind of just does what it wants. This leads to generations that are “good enough” but not exactly what you’re asking for. For instance, I really wanted my alien monster to have tentacles grabbing the tower and kind of ‘swallowing’ it. I just could not get the generative fill to play along. After about 5 failed generations I got this sort of ‘floating’ alien monster. Good enough!
And it’s EXPENSIVE! Adobe charges “between 10 and 60 credits per generation (you don’t actually know until you click the button…I can’t seem to find a pattern of when it charges 10 versus 60. But in about 15 minutes I had managed to use over 100 credits. That’s only about 65 cents, but using this for a whole day could end up costing you over $20 in AI credits. ON TOP of the $80/month you’re paying for Adobe CC.
Do we NEED this new feature? No. We’ve been compositing and layering in Photoshop since “the 1900’s”. But for a few pennies you can potentially save yourself hours of work for virtually the same quality. I think in the end shrinking deadlines and the speed of commerce is going to lead a lot of designers to be more “Prompt Engineer” than “Production Artist”.
If we become dependent on these tools, getting the exact visuals you want may unobtainable for the next generation of designers who haven’t spent thousands of hours masking and blending layers. If we become an industry of “good enough”, are people going to continue to engage with us? If you want “good enough”, you’ll can just hop on ChatGPT and generate your own poster.
Graphic Designers are always focused on quality and precision. AI kind of does an okay version of whatever it wants, but fast and cheap. You still need to know what looks good, and that takes skill and knowledge. But in a world of people more distracted than ever… maybe good enough is good enough? Time will tell. I’m pretty sure “Production Artist” is going the way of “Prepress Technician”.
