top of page
  • Ed Dozier

ON1 Photo Raw 2023 Deblur for Motion Blur Repair

A nice feature included in ON1 Photo Raw 2023 is their “Tack Sharp AI” control, with Deblur. This control has the dual purpose of enhancing image sharpness and repairing image motion blur. There’s always the occasional shot made in really dim conditions where your photo got ruined because of camera shake or subject motion; the “deblur” feature can fix many of these shots.


I’m not going to claim that this tool can transform smeared shots into razor sharp and detailed images, but it can certainly help rescue an irreplaceable photo and give you a second chance. Artificial intelligence (AI) can do some amazing feats that were once considered impossible.




ON1 2023 Using ‘Tack Sharp AI with Deblur’ pixel-level view


The blurred details in the pixel-level view shown above aren’t perfectly recovered, but they’re pretty darn good. This could save shots from the fate of ending up in the trash can.




Nikon Z9, 500mm f/5.6 1/80s ISO 50 repaired photo


The shot above, in which I intentionally used a very slow shutter speed to cause motion blur, was repaired. I used the ‘Tack Sharp AI with Deblur’ tool by itself, just using the default settings (100% on the sliders). You can combine sharpening with the “NoNoise AI” tool, if your shot looks too grainy.




My preferred sharpening settings


I think that the default sharpening settings tend to add a little white border to some image edges, which looks artificial. I happen to prefer setting the sharpen/de-blur slider to a value of around 40 or so for most images. The shot above shows how the view gets split in half, with the left side of the screen showing the ‘before correction’ view. You can drag your image around and you can additionally shift the left/right split to align with a region of interest.


When you’re happy with what the sharpen/de-blur algorithm will do, then simply click the blue “Apply” button to sharpen the photo. If your computer doesn’t have a decent GPU, it will take quite a bit of time to complete the sharpening/de-blur process. My computer has an Nvidia GeForce RTX 3080 GPU, and the sharpen/de-blur operation is nearly instant.


I decided to try some experiments, using a special focus-calibration target; I chose this mundane subject, because it is able to illustrate the capabilities and limitations of the editor AI engine to de-blur photographs. The target squares make it easy to compare de-blurring in vertical versus horizontal orientations.




Moderate horizontal motion blur, uncorrected


In the shot above, I panned the camera left-to-right at a slow shutter speed to force some motion blur. There's a little bit of blur in the vertical direction, but mostly it's horizontal blur.




Corrected motion blur (Tack Sharp AI with deblur) default settings


The pair of shots above (at 100% magnification) show how the de-blur feature had no problems fixing the (mostly horizontal) blur.




Heavy motion blur, horizontal and vertical, uncorrected




Motion blur NOT fixed in horizontal direction!


As shown above, when the blur gets too severe, the de-blur algorithm just gives up and does essentially nothing. The less-severe vertical component of the motion blur was fixed, though. All tools have their limits.


Summary


In an ideal world, you’d never need this tool, and all of your shots are sharp. Reality creeps in, however, and there’s some action that you want to capture near dusk. You really, really hate to delete that smeary photo of Bigfoot.


This de-blurring feature can save the day. Fine surface textures might get lost, but the main image generally gets repaired really well. This tool can mean the difference between losing or saving a shot.


ON1 Photo Raw 2023 has several AI-based editing tools to accomplish a diverse set of tasks; I find it a welcome addition to my photo-editing workflow. In case you were wondering, the ON1 company doesn't pay me anything; I just like to give kudos to companies that offer useful tools to photographers.




488 views
Recent Posts
Archive
bottom of page