Let’s Enhance is a software that claims to have a learning AI that can hallucinate “missing details to make [a] image look natural.” It launched this October and grabbed the industry’s attention. Now, about two months after its launch, how is the software fairing?
Natasha Lomas from TechCrunch recently posted an in-depth review of Let’s Enhance, tried out the software for herself. She technically tried the software when it first came out, but found the results mixed and hard to judge its usefulness. This time around, she found herself in a situation that she felt the software might be able to help her with. She says that she had “photos [she] wanted to gift as framed prints” but they “didn’t have a great resolution for producing a decent sized print.”
When Lomas ran her photos through the Let’s Enhance, she quickly received back three files labeled ‘Anti-jpeg’, ‘boring’ and ‘magic’ per photo. She writes that the ‘Anti-jpeg’ version of the photos don’t upscale the actual size of the image, “it just appears to lightly polish out some compression artifacts.” She found the fixes to be pretty subtle when comparing to the originals.
As for the “boring and ‘magic’ versions, they upscale the images by 4X. Lomas found the results to be impressive on screen, but she notes that with photos she took for a wedding ran into some challenges. “The photos from the wedding were definitely a lot more challenging. Two of the original shots appear to have been taken with a smartphone camera of fairly standard quality, so while they look nice on a screen their low resolution of just 1024 × 577 meant the max size you could reasonably expect to print them would be a tiddly ~4-inches.” And after Lomas cropped one of the shots to cut out a bystander and upscaled it, the image’s resolution came back as 3408x2304, which is large enough to make a decent print.
However, Lomas noticed with one of her wedding portraits that were close up shots of faces the AI upscaling had made a soft of vaseline screen effect rather than making the images look more finely outlined.
When printing, an upscale crop from Lomas' high resolution image turned out well. “A 20 x 15cm print looked great and had clear and vivid details. A non-professional wouldn’t guess it was just a section from a larger shot. Although whether this was down to the AI’s skilled upscaling or really more a case of having a reasonable resolution to begin with for a print that size is less clear,” she says.
But when it came to her two lower quality Al-upscaled photos, the results weren’t great. They still reflected all the fuzziness of the digital version and strangely darker skin tones had been given a grey-ish hue.
In summary, Lomas found herself “not overly wowed” by the results from Let’s Enhance off screen. She concludes that it may be a useful when you want to increase the size of smartphone photos that are being used for digital use. “it’s not literally going to enhance the details of the original shots if they’re a bit fuzzy or out of focus to begin with,” Lomas says. “The service seems most useful if you need a tool for supersizing low res digital imagery regularly, say for fairly lightweight ecommerce purposes.”