Samsung explains its ‘fake’ Moon photos
The company says its AI enhances details, but critics claim it's adding info that wasn't there.
Samsung is hoping to talk its way out of a controversy over its camera processing technology. The company has shared an explanation of the Moon photo detection system it has used since the Galaxy S21. If you have Scene Optimizer turned on, AI detects when you’re taking a clear photo of the Moon at 25X zoom or above. The tech lowers the brightness, captures multiple frames (to produce a bright, low-noise picture) and uses a neural network to enhance the detail using a high-resolution reference image for comparison.
You can turn Scene Optimizer off. Samsung also notes that this won’t work if you either take a snapshot of the obscured Moon or use an image that clearly wasn’t taken on Earth. The Moon is tidally locked to the planet, so you’ll always see the same lunar surface unless you go to space.
The defense comes after Reddit user Breakphotos alleged that Samsung was faking Moon images by adding detail that wasn’t present in the raw scene. To make the case, Breakphotos even snapped pictures of blurry, low-resolution images on a computer screen — there’s no info the phone could recover from the shot. Even with blown-out exposure, the device appeared to add info that simply wasn’t there.
This isn’t an outright fake. Samsung is using the actual shot as a baseline. However, its algorithms are clearly going to an extreme by producing photos that don’t represent what you get through the lens. The company appears to be aware of this, too, as it says it’s refining Scene Optimizer to “reduce any potential confusion” between taking photos of the actual Moon and mere images of it.
This isn’t the first time a phone manufacturer has received criticism for manipulating photo output, of course. Some brands have had beauty modes that mask perceived body and skin imperfections to create unrealistic portraits. However, Samsung is effectively claiming its phones can take technically impossible photos — you may buy a Galaxy S23 Ultra under the misguided impression that someone’s sharp, crisp lunar image reflects what the phone can physically produce.