Mind-Blowing A.I. That Can Add Detail To Even Extreme Blur

The use of artificial intelligence in photo editing is becoming more prevalent with every day that passes. It never fails to impress me, but this has gone one step further, and perhaps, a step too far.

There are a few issues with photographs that cannot be rescued: pure white, blown-out sections of the image with no detail recorded, for instance. Another — arguably worse — issue is missed focus. This can leave your subject anywhere from soft to blurry, rendering it all but useless. Well, there may be something to be done about that now, using an app available for Apple and Android devices: Remini.

This application uses A.I. to generate detail for the parts of the image that need it, but it goes further than you'd imagine. In this video, Unmesh Dinda of PiXimperfect runs some tests on the software to see just how much detail it can bring back. Well, "bring back" is perhaps a little misleading. What it appears to do is use a library of images to extract parts and place them on the image in the correct places. While it's utterly fantastic at that — and it really is singular — it obviously cannot be perfectly accurate with the results, as it has to "make up" the detail, so to speak.

Either way, it's a highly impressive app with a lot of uses, which I am going to pull apart and test to the fullest in a coming review.

Have you tried Remini? What did you think?

Rob Baggs's picture

Robert K Baggs is a professional portrait and commercial photographer, educator, and consultant from England. Robert has a First-Class degree in Philosophy and a Master's by Research. In 2015 Robert's work on plagiarism in photography was published as part of several universities' photography degree syllabuses.

Log in or register to post comments
7 Comments

I'm happy that I never delete rejected pictures. I can already think of some that could benefit from a makeover

It seems that , the usual joke about Deckard telling his computer several times to enhance the photo , in the first Blade Runner , will be no more .

Except by the time the AI has finished making stuff up the final image will be a large hairy Orangutan wearing a green dress and a Fez.

It's amazing how well it actually does...and, quickly.

I remember when Mr. Dinda's videos really dragged, but in the last two years or so they keep getting better and better. He goes at a really fast clip which keeps the pacing fresh and vibrant. And we can always rewind a bit to hear something again if needed.

Topaz Labs used to have a terrific standalone app called AI Clear which did similar work that Remini does, but across an entire photo as opposed to Remini which only does faces. Topaz has since folded AI Clear into its Noise AI app and doesn't seem to work as well, but it's still a good option to consider if you need more than faces affected.

...I must say however that it's a bit comical seeing him au naturel with his mom, and then looking like Max Headroom's cousin in the main production...

I just tried it out... wow!