I am having some trouble denoising jpeg images using the High Quality tool and PhotoLab 6.2.0. I am finding that it introduces artefacts on most images I test. The image I’ve shown in this post (straight out of a slide scanner) was exported using the default settings but with the Denoise Luminance value set to 100 (max), just to highlight the issue. The blacks splotches become increasing obvious as the Luminance value is increased. Adjusting the Chrominance slider introduces the same splotches, but they are coloured a burnt orange.
I am using a Mac running Ventura, but this was also present using Monterey. As these artefacts do not occur with the same images using PhotoLab 5, I am wondering if this might be an issue with the latest PhotoLab. Can anyone reproduce this or make any suggestions?
In my experience of scanning slides, a lot of the ‘noise’ in the scan is actually the grain of the film and yes, it is tricky to handle. I find the best way to deal with it (and any colour cast as in your image) is during the scan itself. I can do this because I have a Nikon CoolScan V ED and it’s associated Nikon Scan 4.0 software has dedicated features (Digital ROC & Digital GEM) that are closely tied to the scanner hardware that specifically address these problems.
With DXO you will not have good results from slides or film scans. What stuck said it’s correct. The best results you will get with a software Nikon Scan, Vuescan or Quickscan, and the you can work wit Affinity, Topaz or any other software. DXO is optimized for RAW development and denoising is very good for High Iso Noise.
Maybe you will give us more informations about hardare you use and so on, and/or a scanned TIFF or Jpeg
I am also using a Nikon CoolScan V ED scanner, but I have been using the Vuescan software. Possibly PhotoLab is not the best choice for images like this, but there does appear to be some issue here. I am working on an M1 iMac.
I’ve attached a comparison of the original image straight out of the scanner (cropped) and the same crop of the output from PhotoLab5 with the HQ tool set to 100%, and the corresponding output from PhotoLab6 at the same setting.
The degree of denoising appears to be the same for both versions of the software, but with PL6 there are clearly artefacts introduced as well. I am seeing this for a lot of images, not just scans from slides.
As I am new here, can someone indicate the best way to submit a support request.
Dear Sparky
like jok wrote “but I have been using the Vuescan software.”
Dear Jok,
the original one looks ok, but how you got the desaturation in the second and third one, and the black failures in the third one…no idea.
What happens if you reset all settings in PL6, then export as new jpeg?
Is it similar to the original or?
I use PL5 and tested some scanned tif (I would not scan to jpeg), and all is ok. The film grain I can reduce with Topaz.
If I export the original it looks fine; it’s only when de-noising that the issue occurs. Note also that I do not see the problem in PL5, only PL6.
I also see the problem in images that are not film scans, so me presenting a scanned image might have been misleading. Here’s the de-noised version of an image downloaded from a popular mobile phone manufacture’s website, for example.
Tested and compared this myself and PL5 works fine but PL6 do not.
Bugger. I just updated my PL5 to 5.7.0 b84 and that broke the HQ denoising.
Now both PL5&6 is broken.
So something DxO introduced in last revisions for both versions is definitely buggy.
That’s one of the reasons I’m still on 5.5 on Windows my main system.
There were and still are too many inconsistencies around the updates and implementations.
If DXO will not improve these methods and deliver Main and minor versions I will stay at 5.5 for the next years.
I’d rather have a stable, clean version with perhaps limited functionality compared to the new versions, but I can rely on it.
If necessary, I also encapsulate it in a virtual machine, or I change the complete software at some point.
Now went for the very same ad … again, no problem here.
@jok – somehow those ‘stains’ look like the result of a tonal compression … as if pixels get stacked, but honestly, I’ve never seen an outcome like these big fat dark blotches.
Indeed. Yours look great.
Hmm wonder what else it can be then.
I’m running a hardware calibrated Eizo if that might affect PL deep inside.
It’s very strange that PL5.4 works ok but not 5.7.
Anyway the devs would know what was changed in between those versions.
