PL4 HQ noise reduction does not reduce chroma noise enough

Hi all,

Well done for PL4 - Looks like a solid upgrade which I did the day the software went out. That being said, it does not look like it’s bug-free. One annoying thing that I noticed is that although DeepPrime is superb, Prime is as good as it was, the standard HQ setting seems somewhat broken.

Even at the Chroma noise reduction 100 setting, there is still visible chroma noise in the image, and in any case much more than there used to be for the exact same image in PL3. This cannot be corrected by further pushing the slider as, as I’ve said above, it’s already at the max 100 setting.

It would be great if this could be corrected in a subsequent built - it certainly doesn’t look like something that is too difficult to fix.


1 Like

Have you tried Moire removal and Chromatic aberration removal in addition to Chroma noise removal? Sometimes what looks like chroma noise is something else. Is it possible that you are using different settings for these between PL3 and PL4?


no, the settings are completely carried over from PL3, which is actually a nice thing. It’s just the chroma NR which is different, not dramatically but weaker and not at the strength I’ve become accustomed to.

Hi Julien,

We did some minor optimizations in the pipeline, but normally the difference should not be perceivable. That being said, we simply cannot validate everything we do for all camera bodies at all ISO levels—so it is possible that your observation is related to these optimizations. Could you please share your photo with us ( so that we can investigate? Or at least the camera model / ISO / exposure time of the images where you observe the differences?


This is the biggest weakness of PL IMO, that after every upgrade people discover changes in how work done in previous versions is rendered. I certainly found them in PL2 and PL3; haven’t yet looked for them in PL4.

Trying to minimize the changes and patching problems as they arise is a poor substitute for actual versioning.

I think you’ll find that such differences are fairly common, whatever the software. It’s not just PL.

Haven’t had problems with C1 or Exposure, or Lightroom in the past (although I was never a heavy user). Both of these have something similar to Adobe’s process version, and even nicely version their dop equivalents in separate directories across major revisions. Of course, there can be problems I haven’t seen (I recall mention of one LR blunder in an old thread here), but I have seen differences after PL upgrades.

The absolute worst I’ve ever experienced is Luminar, which could change the rendering even after minor updates. I thought Luminar 2018 was quite nice, but the only other good thing I can say about them is that they gave me my money back after I discovered how variable their rendering could be.

It’s just a poor property of any rendering application to not give you with the same result when you revisit work in a future version of the software. If there have to be changes then there should be flashing lights and sirens in release notes warning us of them.

DxO doesn’t bother to write decent release notes though. (Not alone, but compare with C1 for example.) Nothing about bug fixes in the PL4 release notes (or maybe there are none) and half of what’s listed as “New features in DxO PhotoLab 4” are things that were already in previous versions. Every upgrade/update is an exercise in discovery.

Hi wolf,

Sorry to reply this late - the model is an A7RII, and the chroma noise reduction difference is pretty much visible at any ISO/exposure time. It’s not much but there remains chroma noise in PL4 even with the slider pushed to the max, while PL3 cleaned the noise completely.

TBH this is not a big issue as I am now using deepprime for every shot, given how fast and efficient it is.

It’s odd, I just tested it with a Sony A7RII picture, 3200 iso, and I found exactly the same result on the chromatic noise between PL3 and PL4. In both cases with all the settings in auto (which gives me on this photo a chroma slider at 100).

Here are two screen grabs from PL3 and PL4 at iso 320. Same processing parameters, different results, with more chroma noise in PL4. You could argue that PL4 is better because it retains more color information, though. Not a big issue anyway, but still, it would be nice if older processings could be maintained from version to version.

An important point: I am in Windows version, so the results may be different.
After rechecking following the examples above, I confirm what I wrote: the result in HQ is strictly identical between PL3 and PL4. But … this is the result of jpeg or tiff output, not screen viewing in PL. Indeed, the displays in PL are a little different, at 400% maximum zoom in Windows version (which seems more than sufficient to me). So if something has changed - still in Windows version - it may be the display engine, but not the processing.

On my side the difference seen in the preview windows are also seen in the final renderings, as should be the case (if not the previews would not be previews :slight_smile:).