Noise and Sharpness: Lighroom CC, Nik Dfine 2.0, Topaz Labs DeNoise AI and Sharpen AI

Recently, I mentioned a few programs in a comment. Program relating to reducing noise in photographs.

Those programs were Lightroom CC, DxO PhotoLab 3, DxO Nik Collection 2.5 (Dfine 2), and Topaz DeNoise AI, Topaz Sharpen AI.

It should go without saying that I have no financial, personal, or emotional investment or stake in any of these products. I own them all, but most have free trials for anyone interested.

This post aims for a quick comparison between the tools using this photo (as shot, no adjustments)

If you click on the photo, you’ll get a version that’s 1280-pixels-wide. If you click HERE, you get the full-size version (9MB, 5568 x 3712 pixels).

A 1280-pixels-wide crop would give you this . . .

It still looks pretty good, but if you clicked on it and then zoomed it and look at it at 1:1 magnification, this is what you would see

Pretty grainy, ain’t it? It’s because it was shot at ISO 4500 on a cloudy day.

Here I should mention that if you’re just going to show the uncropped version, noise is not going to be an issue. For instance, here’s the processing of that photo using my standard editing for showing it on the blog.

Since hardly anyone clicks on images or goes to SmugMug, That would be good enough because no one would see this:

Obviously, if I wanted to show the cropped version, I’d have to do a better job.

The first line of noise reduction (if you have what I have) is Lightroom CC. It has a pretty good noise reduction function.

Oh, one more thing. When you do noise reduction, you can and often have an unwanted loss of resolution and detail. If I just removed the noise, this is what it would look like; as a reminder, the original:

. . . and — if I wanted to remove most of the noise — with the elimination of the noise you also eliminate details:

So, normally, I play a game where I increase the sharpness — which accentuates noise — and apply noise reduction — which reduces sharpness.

At some point, I decide on a workable compromise. For instance:

I reduced some of the noise while retaining some of the details. It’s not bad, but if I want more, I try other programs.

My next step might be DxO PhotoLab . . .

. . . which, with its Prime setting, gets rid of a bit more noise. The sharpening function of PhotoLab is a subtle thing, so while there is an improvement, you really have to examine the photo, which most people don’t do.

My first choice is usually the Dfine 2 module in the Nik Collection. Here, I might get something like this:

Dfine doesn’t recover sharpness, so I either use their sharpen module or just use Lightroom’s tools to regain some detail. For this next photo, I played with Lightroom to recover some of the detail.

Again, let me put the original back up . . .

It’s worth mentioning — again — you only go through this process if intending to either crop really close or print really large and if the noise bothers you.

Note: I go through some version of these processes because I like to pixel-peek, although I’m less anal these days and seldom try the impossible.

I have another tool . . . Topaz DeNoise AI. That program actually incorporates an earlier AI module called AI Clear in addition to the new DeNoise module.

DeNoise AI uses a different algorithm than AI Clear, but both are included and you can choose whichever you think does a better job. 

You also have the option to use the “Auto” settings for each, or manually tweak the process. In the Auto mode, the program decides on the amount of noise reduction and sharpness to apply.

DeNoise AI (Auto Mode)

Notice there’s some funky patterning that’s happening on the rock. The program might have decided the gray rocks are the same as the frog and “recovered” texture that was never there.

Here’s what it looks like on a 1280-pixels-wide crop.

It looks OK here, but if you click on it and then click on different parts of the photo, you can see uneven applications of adjustments.

AI Clear (Auto Mode)

Here, you see less of the funky patterning. Again, it’s a different option in the program and which one you use depends on the photo you are processing and what details it contains. For this photo, Clean AI does a better job (in my opinion). Here’s the 1280-pixels-wide crop.

Again, if you click on it and then click on different parts of the photo, you can see a better implementation of noise reduction.

The manual options for both modules give fairly similar results, but I still liked the AI Clear version better.

AI Clear (Manual Adjustments)

. . . and the 1280-pixels-wide crop:

So, that’s what can be done with each of those programs . . . but, let me mention one other program . . . Topaz Lab Sharpen AI.

Mind you, I could take the above and sharpen it a tad in Lightroom, but I start introducing noise again. Sharpen AI occasionally performs magic . . .

Sharpen AI after DeNoise AI

Let me show the original for comparison . . .

We have come a long way . . . here’s the 1280-pixels-wide crop.

. . . and here’s the original version of the 1280-pixels-wide crop. . .

One more time, cropped close:

Topaz Labs Denoise AI and Topaz Labs Sharpen AI

A few more details, if you’ll pardon the pun . . . DxO PhotoLab and Nik Collection make use of “Control Points” to modify which part of the photo is affected by any adjustments you make. A guy Anthony Morganti has an excellent series of videos about the Nik Collection and every module in it.

Topaz and Lightroom both make use of “brushes” to paint (add or delete) from the mask used to make adjustments. You can find Topaz instruction videos both from official and aftermarket sources and Lightroom is no slouch in that department.

Finally, the reason for multiple tools (my reasons for multiple tools) is that any given tool works for some things and not for others. If you had to choose just one, I suggest taking advantage of the free trials and evaluating them on your own photos.

I hope this was useful to someone, but if not . . .

. . . did you know there are three new stories you can read on this blog?

Not only that, but you can also vote for the one you liked the best. Just go HERE to find links to the stories and the poll where you can vote.

That’s it. This post has ended . . . except for the stuff below.


Note: if you are not reading this blog post at, know that it’s copied without permission, and likely is being used by someone with nefarious intentions, like attracting you to a malware-infested website.  Could be they also torture small mammals.


If you’re new to this blog, it might be a good idea to read the FAQ page. If you’re considering subscribing to this blog, it’s definitively a good idea to read both the About page and the FAQ page.