That fake zoom with AI is gross ugh

If I'm taking a picture of something I want it to be real light-to-pixel action not some made up wambo-jambo

I find it kinda scary that this is marketed as "zoom" and "recovering details", when the reality is that it quite literally makes stuff up and hopes you won't notice the difference. You and I know that it's completely fake, but we (or at least I) don't even know how much is faked, and probably 99% of people won't even know that it's fake at all.

How long until someone gets arrested because an AI invented a face that looks like theirs? Hopefully lawyers will be know to throw out evidence like that, but the social media hivemind will happily ruin someone's life based on AI hallucinations.

It becomes misleading to even keep calling it "Zoom".

More like "Interpolation" with a pinch of hallucination. I can see this becoming a thing though, it is after all the mythical 'zoom & enhance' from csi...

> If I'm taking a picture of something I want it to be real light-to-pixel action not some made up wambo-jambo

Then don't take pictures with phones because it's been like that since more than half a decade at this point even on midrange phones.

Digital has never been light-to-pixel.

At a minimum, you have demosaicing, dark frame subtraction, and some form of tone mapping just to produce anything you'd recognise as an photo. Then to produce a half-way acceptable image will involve denoising, sharpening, dewarping, chromatic aberration correction - and that just gets us up to what was normal at the turn of the millennium. Nowadays without automatic bracketing and stacking, digital image stabilisation, rolling shutter reduction, and much more, you're going to have pretty disappointing phone pics.

I suspect you're trying to draw a distinction with the older predictable techniques of turning sensor data into an image when compared to the modern impenetrable ones that can hallucinate. I know what you're getting at, but there's not really a clear point where one becomes the other. You can consider demosaicing and "super-res zoom" as both types of super-resolution technique intended to convert large amounts of raw sensor data into image that's closer to the ground truth. I've even seen some pretty crazy stuff introduced by an old fashioned Lanczos-resampling based demosaicing filter. Albeit, not Ryan Gosling[0].

Of course, if you don't like any of this, you can configure phones to produce RAW output, or even pick up a mirrorless, and take full control of the processing pipeline. I've been out of the photography world for a while so I'm probably out of date now, but I don't think DNGs can even store all of the raw data that is now used by Apple/Google in their image processing pipelines. Certainly, I never had much luck turning those RAW files into anything that looked good. Apple have ProRAW which I think is some sort of hybrid format but I don't really understand it.

[0] https://petapixel.com/2020/08/17/gigapixel-ai-accidentally-a...

Agreed, I was recently looking through photos from my Pixel 5 and realized that it had "helpfully" modified an image of a bunny, making the fur look wavy and adding a weird faded-out eye on the body. On top of that, it applies skin smoothing to every photo, even with the beauty filter disabled. It irked me enough that I'm looking for a new third-party camera app. Despite using exclusively Pixel/Nexus phones for the last 12 years, I might abandon them entirely once this phone dies.

Agreed, not a fan. The world has enough fakery in it already without people accidentally generating even more (I assume quite a lot of casual users will mistake this zoom for zoom in the traditional sense).

It's giving "Samsung fake moon". If generative AI is going to make up details why bother zooming in, you could just ask to make up a whole AI slop picture.

You can see the original image.

I find it cool, it's pretty much the CSI enhance tool.