Why Samsung's Moongate Shows That We Actually Love Fake Photos

Last week, Samsung got busted for seemingly-fake moon photos taken with its Galaxy phone cameras, which were just too good to be true. This week, nobody seems to care.

Reddit user ibreakphotos noticed that moonshots snapped with the Samsung S20 Ultra looked surprisingly good, considering how tiny the image is in the frame even with a powerful zoom lens. To test it, they downloaded a picture of the Moon, downsized it, blurred the resulting image, and then used a Samsung phone to snap a photo. The result was impossibly good, leading many people to assume that Samsung was just swapping in an existing photo of the Moon, tweaking it, and serving it up as genuine. The reality is more complicated. 

"As a photo archivist with professional expertise in digitizing more than a billion pictures, authenticity is of paramount importance to me. Photography has the unique ability to capture a genuine moment in time, which is why Samsung's use of high-resolution images to fabricate moon pictures is a major letdown. In my opinion, upholding integrity is more valuable than presenting a falsified image," Mitch Goldstone, founder of photo-scanning company ScanMyPhotos, told Lifewire via email.

Reality Distortion

We already know that our phones process camera images heavily. Beauty filters smooth out skin imperfections, night modes use multiple exposures to squeeze details from impenetrable shadows, and the iPhone's Sweater Mode combines many images to get better detail under middling lighting conditions. Other tricks include not taking a photo until all the subject's eyes are open (blink detection), smile detection, sky enhancements, and even—in the case of Google's Magic Eraser—removing distracting objects automatically. 

None of this bothers us. So why does Samsung's Moongate kerfuffle seem worse, somehow? After all, we're taking a picture of the Moon, and getting a picture of the same Moon.

Our tolerance for automatic photo edits and enhancements seems to shift in pace with technology. Making a blue sky bluer, or airbrushing out skin blemishes somehow seems benign, akin to adjusting exposure or white balance, even though those processes may have been counted as "photoshopping" a decade or two back. 

So is using AI to recognize an object, and then applying specific enhancements tailored for that object, going too far? First, let's look at what Samsung is doing.

The Scene Optimizer

In a blog post addressing the issue (an updated English version of the Korean-original posted last year ) Samsung details its "Scene Optimizer" feature. Essentially, this recognizes a number of predetermined scenes and objects, like the Moon, and applies tailor-made processing to enhance those objects. 

In the case of Moon photos taken at higher than 25x zoom, the camera captures more than ten images and combines the results to try to come up with a sharp-detailed image. Samsung doesn't really detail what goes into the AI part of the AI Scene Optimizer, but it seems that it is generating "moony" textures and using them to add detail to the captured image. 

Two years ago, another Redditor, called moonfruitroar, took a picture of the Moon, drew a smiley face on it, and then snapped a photo of their computer monitor with a Samsung S21 Ultra. The phone made the smiley parts all cratery and moon-like, as if it was applying moon-specific enhancements. 

But how far is too far? This Moon enhancement doesn't seem so bad. After all, you took a picture of the Moon, and you get a nice picture of the Moon. But what if someone in your photo is making an ugly face, and your camera recognizes them as one of your friends or family, and swaps in a different, better photo from your camera roll? Maybe even that is ok with you, but it can quickly get out of hand.

"I wouldn't want someone taking photos and, for example, submitting them as evidence in a criminal trial when half the stuff in the image has been AI replaced by some computer's idea of what it probably should have looked like," says technology expert, amateur photographer, and podcaster John Siracusa on his ATP podcast. "It's a complicated topic because we are so accustomed to cameras capturing what we're pretty sure they should be capturing based on what they are seeing."

It is, as they say, complicated. But based on the evidence around us, all we really want are nice photos, and we don't much care how our phones manage to capture—or manufacture—them.