Topaz Sharpen AI update
Topaz Sharpen AI V2
Image sharpening software
Topaz has announced a significant update to the Sharpen AI software. The update includes batch processing capabilities.
Sharpen AI also adds the ability to apply or remove (mask) the sharpening treatment selectively to different parts of an image. It’s free for existing users.
The Mac version is reviewed here, but the Windows version works just the same.
Topaz has an initial sale for the new version. Sharpen AI is on sale for $59.99 (25% off) until the third of April 2020.
See all of Keith’s Topaz reviews and articles for more
Topaz Sharpen AI
All digital images require some degree of sharpening at some point in their journey from camera sensor to print or screen. For a perfectly steady camera with a static subject and a top performance lens, this may be negligible. However with real lenses, moving subjects and many photographers, there will be types of blur in an photo. It might be lens softness or camera movement – something I’m always aware of given my liking for hand held photography. There are lens issues too, where aberrations can creep in away from the centre of the image.
For this reason I’ve always paid careful attention to aspects of image sharpening and long realised that it is rarely the optimal solution to apply it to a whole image, irrespective of image content. For people who suggest that an overall sharpening is required at some point in a workflow, I just say: “Noise in clear areas of sky”
The new software appeared as an update for my existing version of Sharpen AI
The release notes for the V2 update:
- Added ability to selectively mask out sharpness
- Added ability to process batches of images
- Updated AI models
- The right panel has been redesigned to improve clarity
- Save dialog has been redesigned
- Improvements to processing speed
- Added option for default save as prefix
- Fixed Photoshop outputting default settings
- Fixed Lightroom template installation on Mac
Once installed, you need to login to your Topaz account to activate the software
It’s always worth a quick look at the startup splash-screen sequence with an update, since it shows things you might not otherwise notice.
The software opens up in standalone mode, but I’ll be using it here with Photoshop CS6, which is still my basic image processing software.
A quick tour of the software preferences is also in order.
An example image and why Sharpen AI works well
I’ll use this photo of the church at Staunton Harold as an example. It was taken when I was testing the Samyang 24mm tilt/shift lens a few years ago.
The lens is an economical entry point to using tilt/shift lenses. However, at full upwards shift, as i’ve used here to correct for converging verticals, the image quality drops off, even if you use smaller apertures (f/11 for example). The lens is shifted up (aka rise) meaning that the top of the frame is very much at the edge of the lenses image circle.
We’re used to image quality falling off towards the edge of the frame – it’s sometimes corrected for with software ‘profiles’ for a lens. However, the loss of detail is symmetrical about the centre of the frame, which is no longer the case for a shifted image.
This means that the softness and amount of sharpening varies over the frame in an unpredictable manner (the lens doesn’t supply shift data for our image’s EXIF data) – In some ways similar to my observations that sharpening is not a ‘same for all’ function.
It’s important to note that I’m only addressing lens aberrations such as coma/astigmatism/chromatic here, not geometrical distortions (see my Samyang 24mm T/S review for more about this).
Whilst in Photoshop, Sharpen AI is available in the ‘Filter’ menu. It opens up for you to work on, but does NOT return the results as a new layer. You need to duplicate your background layer before opening the filter. This will return the sharpened result in a new layer that you can easily mask with the underlying layer if you only want parts of the image sharpened. Sharpen AI 2 does include masking (see later) if you’re using an editor that doesn’t support layers.
What if you forget to duplicate the layer first? Well, duplicate the image, then in the original step back in the edit history to before applying sharpen AI. Now drag your duplicated image onto the original (holding shift will align) and it will appear as a sharpened layer over your unsharpened source image, ready for masking – yes, I still have to do this every so often…
There are not many controls for the filter, but there are three basic types of sharpening. See more about this in my original review
You can fine tune settings if need be, but my first move is to turn off the ‘auto update preview’ setting, since a lot of calculations are going through your graphics card even to update the zoomed preview view.
The auto setting are pretty good in analysing the image (basic sharpen here).
The view above (click to zoom, as with many images here) shows the split view, where you can move the dividing line left/right to show before/after views.
Or, a side by side view may be easier (and is a smaller preview to generate).
This side by side view shot is using the more agressive ‘Stabilize’ correction.
Why choose this?
It happens that the radial distortion of this lens towards the edges of the field looks a bit like camera shake – well, it looks like camera shake to the camera processing software. The software is really good at dealing with real camera shake, and treats this area of my image as if there is movement.
A look at before/after versions of the image shows the degree of ‘fixing’
A few notes on what you’re seeing here.
- This is a 200% zoomed view of a 21MP image – expect to see some slight ‘hardness’ and pixelation of the resulting image.
- The original RAW file was processed with all sharpening turned off – this reduces artefacts in the sharpened image due to bits of ‘double sharpening’.
- This was the ‘auto’ setting – I made no attempt to fine tune the settings.
Anyway, here’s the shot after processing – which takes a minute or two, depending on how strong you’ve set the sharpening and the size of the original image.
Other ways of using Sharpen AI
If you’re using the software and don’t have masking capabilities available, then it’s possible to set a mask area for which parts of the image are going to be sharpened.
This is ‘painted’ onto the image preview, before processing. If you are using this, I’d suggest starting with a few smallish images so as to get the hang of the masking tools, without having to wait too long to check results.
If you’ve a whole folder of images that need fixing, then the standalone app now has a batch mode. As ever, test a few images on their own to make sure you’re happy with settings, before leaving your computer running for a long time going through a folder of several hundred images.
Thoughts on using Sharpen AI
The software is currently my first choice in sharpening images when using tilt/shift lenses, since it is completely unphased by varying levels of softening and distortion in different parts of the frame. The camera shake correction not only works very well with slight movement (I like shooting hand-held) but fixes some distortions such as you get with cheaper tilt/shift lenses such as the Samyang T/S 24mm
It’s also rather good with some of my adapted medium format lenses, whether tilted, shifted or not. See my look at an M645->RF adapter for more.
Any questions? – please email or use the comments below…
Never miss a new article or review - Sign up for our occasional (ad-free) Newsletter
Enjoyed this article?
All the latest articles/reviews and photo news items appear on Keith's Photo blog
We've a whole section of the site devoted to Digital Black and White photography and printing. It covers all of Keith's specialist articles and reviews.
For All about using tilt and shift - articles/reviews about tilt/shift lenses
Articles below by Keith (Google's picks for matching this page)