Selective sharpening for Sharpen AI

Simply put the sharpening of the whole image is not suitable unless, only in some cases, the image is a landscape shot.

Pre & output sharpening address general sharpening but I would like to be able too brush in areas that need sharpening. For example, remove skin tones from the process, long exposure water shots etc.


If like AI Clear, Sharpen was made to be an Adjustment as well, it could be masked inside Studio?

1 Like

Yes, but I suspect the processing would be horrendous as I just tested a 6000 x 4000 image, albeit it on an underpowered, unsupported, GT540M and it took 42 minutes to produce a JPEG at 80% quality. Whereas on a supported 1050 it takes 18 minutes.

When I say underpowered it really isn’t so but not satisfactory for Topaz products anymore … Even though the details say Studio ver 1.1.0 it is actually Sharpen.

Topaz Studio Ver: 1.1.0
Operating System: Windows 10 (10.0)
Graphics Hardware: GeForce GT 540M/PCIe/SSE2
OpenGL Driver: 3.3.0 NVIDIA 391.35
CPU RAM (MB): 8173
VRAM [Total, Used] (MB): 1024, 1024
Preview Limit (Pixels): 2089

I suspect Topazlabs have thrown themselves behind AI with a real commitment: it will find out the faults, flaws and frailties in our hardware.
Personally, I have bitten the bullet and future-proofed bigtime.

But that’s just me!l

People on lesser budgets will struggle, I’m sure.

It is interesting that most will have to budget $3000 depending on storage, monitor & CPU with a min 8GB GPU. (Windows & not Mac)

And going on the rate that we have seen an increase in technical requirements over the last 2 years that may even become antiquated.

I understand that AI can do wonderful things but sometimes I believe the process is just not thought through … this is a good example where I can’t ever see wanting to sharpen/stabilize/focus the whole image as I wouldn’t dodge or burn on a whole image.


I suspect professional users of the indispensable range of Topazlabs products will adjust to the monumental challenge - the game-changing challenge - that is AI.

The more amateur/hobby user more likely will find new avenues to express their artistic and creative talents on machines which still hold up good for a whole plethora of plugins - Topaz or not - or they will accept the slower speeds of their systems in the firm belief that Topaz AI is good for them!

I find Topazlabs Products - AI or not - totally indispensable.

But that’s just me :slight_smile:

1 Like

When Sharpen AI is updated to handle RAW files, then it will be much more useful to me. As Don points out, you don’t want to process the whole image - just sections.
So, even if they don’t allow processing of an area selection, at least if it’s a RAW file, I will just process it in PS as a layer and mask the sections I need to use with the original image.

1 Like

I did a test using a 6000 x 4000px jpg file using the Sharpen function in SAI for comparison. My Gigabyte Radeon RX580 GPU took 45 seconds to process and save the file. Stabilize and Focus usually take roughly double the time but I didn’t test it.

The total processing time would be more bearable it were possible to script Sharpen AI so that one could process a batch of images or a single image with different settings. One could then come back and review the saved images when the batch processing was done. Does anyone know if this is currently possible, or if it’s something on the programmer’s ToDo list?


I have started to use Studio to mask the image and then use it to call the Sharpen AI plugin. While it does not speed up the processing since the whole image gets sharpened, it does allow me to restrict where the sharpening will actually be applied when the image is returned to my workflow too.

It wasn’t looking for a comparison because I did state that it was a “underpowered, unsupported” GPU.

People that are thinking of getting a new graphics card can use a comparison or benchmark to give them an idea of what works. For example, they may think they need 8GB of video memory but they don’t (4GB works well) unless they are making huge files which most people won’t need.

This thread is about requesting selective sharpening as an enhancement to Sharpen AI, if you want to offer advice on GPUs start a thread with that propose.

For selective sharpening, I sharpen a copy of the image, and then use Image Layer in Studio with the the original and the copy and using a mask, brush in the areas I want to sharpen. Works good for me.

But the issue will always be that you need to process the whole image rather than simply being able to mask in the areas you want sharpened in the application. Visually it would also be better as you could adjust different areas, e.g. shadows or highlights, with a different strength or indeed use focus on some areas and sharpen or stabilize on other areas.

I totally agree. My workaround is just that - a workaround - not a solution!

1 Like