Samsung's latest image-generating AI tool is a bit too good

Samsung likes us (and its shareholders) knowing that its new mobile phone is the best mobile phone AI-Ed ever in history, and Folding 6 I test is a new tool called "Skice to Image". Draw a rough sketch on the image or Tom notebook that uses the AI ​​generation to convert it into images. When Samsung announced it for disobedience, I put on it, just another ai -ting -bet that's really good. Great, it worries me a little.

The use of a sketch note is very harmless: you draw some content, highlight the display, and then select "3D cartoon" and "illustration" from a few styles to get graffiti for more detailed things. Your image will be sent to the cloud and after a few minutes you will see some options. The results are often cute and fun based on my two year old's requests. Sometimes you will see a teddy bear with too many arms, but nothing serious.

Using sketches to represent images is where things get weird. I'm the world's worst artist and this tool transformed my very simple sketches into photorealistic images. AI-generated elements are convincingly incorporated into photos—scaled and matched to their surroundings in such a way that they're hard to spot as fakes. That's how I got to the bee problem. I took a photo of a dock just south of downtown Seattle with some flowers in the foreground. Since they are close to the camera and my focus was far away, they are a bit blurry. I drew the world's worst sketch of a bee on one of these flowers, thinking that the AI ​​would insert a picture of the focal bee and easily pass it off as a fake. Wrong!

Ai Bite is blurry, just like the flowers they landed. If I do not know the origin of this bee, if I roll the photo on Instagram, I don't think twice. In my opinion, the photographer took the picture at the right moment, or was waiting nearby for a bee to fly into the picture - these things require skill and patience. But in reality it is not. In fact, I'm not even sure I'd see the "AI Generated Content" watermark in the corner of the image.

I haven't messed around much with sketch-to-image this past week, and the results aren't always "fuzzy" good. They often feature AI-generated art features, such as text written in a strange-looking language or strange textures that don't look right. At first glance, it looks convincing, but if you look at it for more than a second, you'll notice something is wrong. Sometimes the content itself is emitted -I think no one thinks that I see a large pirate ship anchoring in the Gulf of Eliova or I see a huge orange cat at the crossroads of Siba. Even if the picture is so strange, no one will make a mistake to be real and they look very realistic.