Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

MacRumors

macrumors bot
Original poster
Apr 12, 2001
67,490
37,777


As part of Apple Intelligence, Apple has unveiled a new Image Playground feature that allows users to create generative images on the fly using a range of concepts like themes, costumes, accessories, places, and more.

Apple-WWDC24-Apple-Intelligence-Image-Playground-240610.jpg

With Image Playground, users can quickly create playful images, choosing from three styles: Animation, Illustration, or Sketch. Image Playground is straightforward to use, built into apps like Messages, and also available as a standalone app.

Previews are created on-device, and your generative images are accessible in Messages, Pages, Freeform, and even a dedicated Image Playground app. The feature is part of "Apple Intelligence," a deeply integrated, personalized AI experience for Apple devices that uses cutting-edge generative artificial intelligence to enhance user experiences across iPhone, iPad, and Mac.

Article Link: Apple Unveils 'Image Playground' On-Device AI Image Generator
 
Last edited:
Why a limit to only three styles of generated images? That's backwards compared to what we've been able to play with for the last couple of years, needlessly.
 
  • Like
Reactions: fullauto
I have seen apples image model and it's pretty bad compared to Dalle3, SD, ideogram. Gemini have not checked lately because its deficiency and ...politics?....were disabling. Hopefully Apple have cleaned it up since last time they released a public version. It had fairly bad comprehension and composition. I didnt see them display any text but only Ideogram sometimes Dalle3 and some other models like SD3 can reliably do that.

The real win here is that apple seems to have incorporated picture to picture and outside of Stable diffusion and Ideogram thats not widely available. Transforming your photos into new images is a huge win depending on how deeply they are using the information about the people in your photos. If I have 10,000 photos of my son that should be enough to create a model that can consistently put him in any context and that is amazing because with SD that means training a LORA, developing images which isn't bad but isn't easy for a beginner. IF this works out of the box that will be an absolute huge win. Excited to see how it turns out.
 
Maybe it's a way for Apple to prevent/reduce the proliferation of AI deepfakes
Nah, it must be more than that. Half the fun of imagine generation has been the ability to say stuff like "in the style of/reminiscent of/photo realistic" etc. I can see them shying away from giving it the ability to copy styles of real artists, for legal reasons.
 
Nah, it must be more than that. Half the fun of imagine generation has been the ability to say stuff like "in the style of/reminiscent of/photo realistic" etc. I can see them shying away from giving it the ability to copy styles of real artists, for legal reasons.
100% this is to avoid the barrage of lawsuits.
what i want to know is how they trained their models...
where was the image training data sourced from ?
where did they get all the language models data from ?
 
100% this is to avoid the barrage of lawsuits.
what i want to know is how they trained their models...
where was the image training data sourced from ?
where did they get all the language models data from ?
I'm curious too... they'll have had to be very careful.
 
can't wait to see how much this tool is gimped based on apple's puritan ideology
For such a puritan company, who gave us fine woven to replace leather (*facepalm*), I'm surprised they allowed fishing, AKA fish-torture, to be shown as family friendly 😂
 
  • Like
Reactions: haruhiko
This will be a game changer for Keynote/Powerpoint. Hopefully no more endless searching on Google for the right image, which turns out to be low res.
I absolutely agree. To me that was one of the most amazing features. Letting me draw a sketch in Keynote, and then have it created will be freaking amazing, if it works properly!
No photo realistic option at launch is disappointing
Indeed. But I understand why at least. Apple will have a fight to begin with, to combat generated penises, racism, etc.
Again, something that my kids will get a kick out of. Wonder how this stuff will work with ChatGPT integration.
Also my first thought. My daughter will have a party with this app.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.