Upscaling Video to 4K Doesn't Make It Sharper—It's Software Guessing the Extra Pixels

Click the "4K" option in your conversion software and the image does get bigger, with lines that look sharper. But that doesn't mean you've somehow captured more detail. Most of those extra pixels are values a computer calculated as a best guess—not something the camera actually recorded at the time. This is where upscaling gets misunderstood the most.

Let's get the numbers straight first. What people generally call 4K refers to the Ultra HD standard, at 3,840 x 2,160 pixels; 1080p is 1,920 x 1,080. Do the math and upscaling 1080p to 4K means adding three new pixels next to every original one, quadrupling the total pixel count. The film industry has its own "true 4K" or DCI 4K spec, at 4,096 x 2,160, which isn't quite the same as the Ultra HD found on consumer displays. As for analog-era standard-definition video, vertical line count sits at 480 or 576, with actual width varying depending on the broadcast region.

Traditional upscaling software relies on interpolation: it samples the color and brightness around a given pixel and calculates what the new pixels in between should look like. Done crudely, this might just mean copying neighboring pixels; more advanced algorithms try to preserve edges and avoid jagged artifacts. But no matter how refined the process, interpolation can never conjure up real texture that wasn't there to begin with, and the result usually ends up looking softer or blurrier overall. Layer on sharpening or noise reduction to compensate, and if the balance isn't right, the image can end up looking unnatural—a dilemma that's awkward no matter which way you adjust it.

The Cost of Restoring From Real Film: The Star Trek: The Next Generation Case

If the original source material is good enough, the ceiling for upscaling quality goes up considerably. When CBS commissioned a team in 2011 to restore Star Trek: The Next Generation, they didn't work from the old digital masters—they went back to the 35mm film negatives shot at the time, material described as being roughly equivalent to 20-megapixel quality. The project drew on 25,000 reels of original film, combined with visual effects and CGI touch-ups, and took over three years to bring all seven seasons from their original broadcast-TV quality up to 1080p Blu-ray specs. According to one producer involved in the project, the whole effort cost more than $12 million.

Ironically, the analog-to-digital conversion actually worked in the restoration's favor here: the film itself had captured far more detail than what the television broadcasts of the time ever used, so the restoration team was essentially digging up information that already existed but had simply gone unused. But if the best material you have is already a digital master file locked at a fixed resolution, the situation is completely different—every pixel added on top is pure guesswork, with no real detail left to dig up.

Home Upscaling Relies on Interpolation or AI—and They Fill In the Gaps Differently

Obviously nobody's spending millions of dollars to restore a home video. The realistic options are traditional interpolation algorithms or the AI upscaling that's become increasingly common in recent years. Both generate the new pixels needed to fill out a 4K frame, but they do it differently. AI models are trained on huge numbers of paired low-resolution and high-resolution images, learning to recognize what common visual features "should" look like—essentially predicting what a sharper version of the image would be, rather than simply interpolating pixel by pixel. The results usually look more pleasing and convincing, but they're still not an accurate reconstruction of reality, and can sometimes produce strange artifacts that interpolation wouldn't. These models keep improving too, so results that look great today might seem crude a few years down the road.

If you're dealing with older footage, there's also the issue of interlacing to consider: video designed for old CRT screens split each frame into odd and even sets of horizontal lines that refresh separately, a method modern displays don't use—so the footage needs to be de-interlaced before upscaling. As for whether a 480i NTSC video can be upscaled to 4K and still look watchable, honestly, there's no guarantee.

One more thing worth remembering: compression rate. Video is often compressed at a lower bitrate to save space, meaning less data is retained per second of footage. Upscaling only enlarges the frame size—it doesn't reverse artifacts like blocking or color banding caused by compression. At the end of the day, how well upscaling software performs still comes down to how much usable information was in the source material to begin with.