Deconvolution sharpening or a toolset for removing blurry / out of focus elements (e.g. topaz sharpen ai / luminar supersharp ai)

Dear team,

i tested several sharpening tools in the last week. It seems that tools using deconvolution sharpening or some other methods can help to fix technically bad phiotos which are a bit unsharp due to focus issues or too low shutter speed. In former times I just deleted the pictures. With these programs I was able to keep some of them.
It would be nice to have such toolset in the nik tools to rescue blurry / out of focus pics.

thank you in advance


DxO products are not intended as repair tools for defective images and that is unlikely to change anytime soon. If you don’t own it already, you should consider using Topaz AI sharpening for that purpose.



isn’t that what Nik Presharpener does - uses deconvolution

they are as defective as the ones for which DP/DPXD are used

I am not sure I understand what you are suggesting. DP and DPXD are not designed or intended for use on defective images. Perhaps your definition of a defective image is different than mine. My definition of defective in this case are issues caused by photographer error which is generally the reason for blurry or out of focus images, or sometimes due to sub par equipment.


image can be underexposed because of user error or genuine situation, image might lack shapness because of user error or genuine situation

just like AI/ML based “NR” inventing details there is nothing wrong w/ having deconvolution based (where btw there is no AI/ML - just the math … so deconvolution != any AI/ML inventing) resolution enhancement - and I think this is what DxO already using even w/o referring to “deconvolution” directly

sure and I have a different definition so we can leave it at that

I have no problem with differences in definition. Let me restate and say that DxO products are not designed to repair defective images caused by human error. However, they sometimes are able to improve poorly captured images to some degree. In my experience, blurry or out of focus images are generally a result of human error.


i did try topaz AI, which need to install 42 module to be efficient (wasn’t impressed by this) and the sad thing is that using trial version doesn’t allow you to save, only a visualization of what you’d get, but they will offer you to buy it though if you’d like to save the work.

This topic is a feature request - in other words, a request for something the software cannot do yet. I don’t see what you’re objecting to, Mark.

About “deconvolution sharpening” - I’ve seen this used in software and in certain camera JPEG engines to mitigate the effects of diffraction at small apertures. Topaz AI sharpening doesn’t seem to rely upon this kind of sharpening:

…though perhaps in reality Sharpen AI and similar Topaz products use a combination of deconvolution and machine learning. What DxO software does is more mysterious. If you want to hear from DxO about what Lens Sharpness does, I’ve seen no better discussion than this one:

Several users on the forum have speculated that DxO uses some sort of deconvolution algorithm to correct image softness from RAW data. I believe DxO didn’t deny that - they only said that they had no short-term plans to provide a means of diffraction correction in PhotoLab without the use of optics modules. Long-term, who knows. As far as I know, nothing has been said about what is planned for the Nik Collection, near-term or long-term.


Ahhh - The good old days …

We’d ask a question - Svetlana would answer it, or refer it to someone else … and we’d all learn a bit more about how PL works, and how best to leverage it.


I am not objecting. I am just suggesting that its unlikely to happen. Both in private and in public communications DxO staff have indicated that DxO products are intended to maximize image quality not repair images as a result of human error.


you are again stuck in the wording … deconvolution is just another method “to maximize image quality”

1 Like

I never used the word deconvolution, nor was I responding to the use of the word. I was responding only to the request for a tool similar to Topaz Sharpen AI which would fix blurry or out of focus images as a result of user error and the likelihood, based on feedback from DxO, of features to do that being implemented.

I initially intended it as a simple response to a specific request and amended it to make it a bit clearer. This is getting unnecessarily overcomplicated and really does not require a continued dialogue.


1 Like

Post deleted …