This article is part of our AI Phone Face-Off. If you're interested in our other comparisons, check out the links below.
One of the most impressive AI features on phones is the ability to generate images. While there are valid arguments regarding AI and art, there’s no avoiding the fact that phones can create some pretty impressive designs.
With that being said, the quality and ability of AI features like image generation can vary massively between devices. With that in mind, we here at Tom's Guide are testing some of the best phones to see which offers the best experience in every AI feature.
With that in mind, I grabbed my iPhone 15 Pro Max, alongside a Samsung Galaxy S25 and a Google Pixel 9 Pro XL, to see which offered the best image generation. So let's see what these phones managed to create, and who deserves to be crowned the winner.
The rules
This comparison is focused on each phone's native AI image tool, which means I will be exclusively using Apple's Image Playground, Samsung's Drawing Assist and Google Pixel Studio.
Also, while I might be using the last generation of iPhone, there's no difference in the images it produces compared to the iPhone 16 Pro Max or any other Apple Intelligence-compatible iPhone available right now.
In some scenarios, the devices won't be able to create certain images, likely due to some underlying rules on appropriate prompts. In such cases, a phone will be disqualified for that round.
PROMPT: A Skeleton in a suit
For the first prompt, I wanted to see how the devices could handle creating several layers, as well as the bones, which can be complex. However, Pixel Studio refused to generate the image. So the first round is immediately down to only two competitors the iPhone and the Galaxy.
The iPhone's image has the most intricate background, but the actual skeleton is lacking in a lot of detail. At a glance, you can see that the ribcage melds into the coat for some reason, while the skull's teeth are a mess. Meanwhile, the Drawing Assist version has more detail, with noticeable layers and a much more realistic-looking skull.
WINNER: Galaxy S25
PROMPT: Creepy goblins hiding in a forest
For my next image, I wanted to see how the phones handled creating a detailed image while hiding elements within it, rather than having them front and center.
The iPhone seemed to have the most issues with this concept. While the goblin is certainly monstrous, it's neither creepy nor very well hidden. Meanwhile, Google Pixel’s image was certainly creepy, and it managed to add more than one goblin, but again, it seemed to struggle with the hiding aspect.
Drawing Assist's attempt shows off a creepy-looking tree and manages to hide the goblins within its roots. I also really like the use of shading in the entire image, which is an area Samsung seems to excel in.
WINNER: Galaxy S25
PROMPT: Kids on a spaceship
With this prompt, I wanted to see if the AI could place generated characters into a larger element, in this case, a spaceship.
The Galaxy phone opted for stuffed animals on a ship rather than kids. It's stylish and offers some nice shading, but not what I asked for. Apple's Image Playground did work, but only after I tried several different inputs, including a different AI-generated image of a person, to get the desired result. It's certainly the most detailed character in any of the three generated images, but the background is overly confusing and doesn't really sell the idea of a spaceship.
The Pixel phone offered the closest to what I asked, and managed to make an image with a lot of little details. For instance, it appears to have made the kids all look like siblings, while also generating something that resembled a Pixar poster.
WINNER: Google Pixel 9 Pro XL
PROMPT: T-rex on a boat
This one is easily my favourite prompt, as it allows the phones to show off their interpretive ability in how well they can fit the larger dinosaur comfortably into a boat. The Google Pixel, which uses a very lifelike t-rex, but there's a lack of dynamic motion, and the dinosaur just looks placed in the image rather than a part of it.
Meanwhile, the iPhone went for a more cartoony design, with some great-looking water surrounding the boat. But unfortunately, it also shows some common generation mistakes we frequently see in AI art. The dinosaur (which is the least t-rex-looking of the trio) has ended up with four nostrils and two tails.
Finally, we have the Samsung Galaxy, which also went for a more realistic route. However, the big difference is that it managed to make the T-rex feel like part of the image, rather than something imposed over the top. Even if the oars going under the arms looks a bit odd.
WINNER: Galaxy S25
PROMPT: Old man looking sad while his wife laughs
For the final image, I wanted to see how the phones handled creating images with two charaters with different emotions on display. This prompt was the first time that Samsung wouldn’t play ball, with the Galaxy S25 failing to generate anything.
Apple's iPhone is currently limited to only being able to generate a single person into an image, but it managed to give the old man a suitably sad look. However, the hands of the character are completely off, with missing fingers and an odd arm length.
Meanwhile, the Pixel managed to create an image with a subtle play of emotion and a nice blurring effect to help pull focus. However, for some reason, it added what appears to be a pair of metal studs in the old man's forehead.
WINNER: Google Pixel 9 Pro XL
Verdict
For the most part, I would say that none of the phones offer foolproof image generation, at least not yet. However, as we've come to expect from Apple Intelligence, Image Playground is certainly the weakest at this point. However, we should see improvements here when iOS 26 launches, plus it was the only phone to create an image for all five categories.
Google's Pixel Studio makes some good offerings, but it struggles with more complex subjects. However, the recent addition of the ability to generate lifelike people helps it to stand apart from the others.
Therefore, I would have to classify Samsung's Drawing Assist as the winner. Though some of its generated images vary wildly in quality, it was still the most consistent.
However, art is purely subjective, so I am curious about what you think: which one is your favorite, or do you think AI should stay out of art entirely?
More from Tom's Guide
You must confirm your public display name before commenting
Please logout and then login again, you will then be prompted to enter your display name.
