Photoshop's new AI features look like magic
Adobe has announced a new AI feature for Photoshop that is available in beta today and, frankly, is one of the most remarkable demonstrations of AI's potential as a tool I've yet seen. The feature, called Generative Fill, is powered by Adobe's Firefly AI models and allows you to insert new objects and alter pictures with text prompts. Really, though, you should just watch what happens in the announcement video.
As someone who's too cheap to fork out for Photoshop but routinely struggles to express his creative visions with MS Paint, this is the kind of thing that makes you think twice. Obviously Adobe is going to do its best to make the technology look good in a showcase video, but this thing is already in the wild and people have been using it to do things like inserting hippos into family photos.
Adobe designer Scott Belsky made the following points about the software:
- We’re leveraging context from your Photoshop file to optimize your prompt.
- Firefly is trained on licensed / non-copywrited content (no scraping), so enterprises can use.
- Content credentials added for media provenance.
"It’s an exciting era for creativity," said Belsky. "This technology enables more people to feel confident creating while also saving time and unlocking more cycles of discovery and possibility for pros." On Adobe's website there are further details about the feature and its future ambitions for its functionality, including the creation of bespoke 3D objects.
Plenty of senior tech figures have been at pains in recent times to emphasise the nature of various AI software as creative tools rather than one-stop-shops that will be producing the movies and games of the future. That's obviously the way this is all going, but I've never seen a more stark or beautiful illustration of AI's potential as a real time creative tool. And if this thing can help me work out which layer I'm working on without any fuss then hell, maybe the AI future's brighter than I thought.