PL8.1.0 Still adds Chromatic Aberration to images between 25% and 74% & to the Loupe

[BHAYT (Bryan [WIN10, PL8, VP4, FP7, G9 & EM1 Mkii, RTX 3060])]

@bildbaendiger

@bildbaendiger

I was not aware that these are artifacts in the calculation of the image for every specific program.

Neither was I and thought that I had a particularly bad image/camera/lens and tried my utmost to clean up the image but that was never going to be successful if/when the image is displayed at 74% or less down to 25% because I was trying to correct a “figment” of DxPLs rendering algorithm.

So at 75% and above you should be seeing what you are going to get when the image is exported but sadly that isn’t the whole story either as my little experiments have shown.

DP XD throws a little CA into the mix when exporting and DP XD2s throws a little bit more into the mix!

How can a user remove something from the image when it is actually being added by the export code!?

It is interesting that you have discovered that CaptureOne seems to have a similar “habit”, I own a copy of Affinity but as a Panasonic owner NX Studio won’t do me much good!

Yesterday was an opportunity to grab a little sunshine, which promptly vanished but we visited Leonardslee Gardens anyway before the last of the Autumn colour was gone but nature had mostly beaten us to that as well.

So plenty of “twigs” to photograph to see what DxPL manages to do with them, nothing good, I am sure of that.

@Lucabeer Nice image which I thought I couldn’t beat until I found a real horror. It retains some CA throughout processing and some is present in the export., even without any Noise reduction to add even more CA to the export.

Sorry, I started to do the whole set of image comparisons all over again and then decided what is the point!?

With DxO it is like “banging your head against a brick wall” only “lovely” when you stop. So, stop I must, while I have a little energy and patience left.

@Wlodek

I would like to have a button to force full quality image preview for that purpose (ok, topic beaten to death…).

@Wlodek As would I but if we don’t continue then DxO will simply continue as they have for too long and the product will fail to be what it should be.

However, I had substituted programming for being involved with the Forum and almost “weened” myself off any involvement with the forum and then PL8 was released!

With respect to your Moiré comment I had written a response to the Moiré topic adding my CA concerns to the topic but decided not to post it, how weird is that.

Take care

Bryan

[Wlodek]

@BHAYT

DP XD throws a little CA into the mix when exporting and DP XD2s throws a little bit more into the mix!

Never seen that. Which camera/lens you have in mind?

[BHAYT (Bryan [WIN10, PL8, VP4, FP7, G9 & EM1 Mkii, RTX 3060])]

@Wlodek

@Wlodek

Never seen that. Which camera/lens you have in mind?

The following is a comparison of PL7 versus PL8 and within those categories we have the same edits with NO NR [M] and with DP XD(2s)[1].

This image seems to have more residual CA than other images I have tested, i.e. in the NO NR case.

This is what FastRawViewer makes of the sharpness of the exported images

What differences do you see?

I have other images where I believed the differences were more obvious, I need to dig them out.

PS:- The original export images zipped

P1138044_PL7-10R.zip (40.5 MB)
P1138044_PL7-10R_1.zip (33.2 MB)
P1138044_PL800R.zip (39.9 MB)
P1138044_PL800R_1.zip (31.4 MB)

[Wlodek]

What’s the point?

[Lucabeer (Luca Chichizola)]

@Wlodek

it was impossible for me to find the examples like yours

Crank saturation and vibrancy up, and you’ll see plenty!

I tend to shoot in overcast days to get low contrast images, and then raise saturation in PL: and that’s where the issue becomes very evident.

[BHAYT (Bryan [WIN10, PL8, VP4, FP7, G9 & EM1 Mkii, RTX 3060])]

@Wlodek If there is any difference then if we have any control over what the editing process does we have absolutely none over what Dxo builds into the export process!?

[Wlodek]

Tried with CVP=50 (25 actually used) and some sat and vibrancy added – sharp difference in sharpness but not so much in color. Z24-120/4 @24mm/f5.6, Z8. This must depend on monitor too, I guess.
At 70%:

At 75%:

@BHAYT

If there is any difference then if we have any control over what the editing process does we have absolutely none over what Dxo builds into the export process!?

Not sure how to understand that. Communication breakdown, so I give up.

[Lucabeer (Luca Chichizola)]

@Wlodek

Tried with CVP=50 (25 actually used) and some sat and vibrancy added – sharp difference in sharpness but not so much in color. Z24-120/4 @24mm/f5.6, Z8. This must depend on monitor too, I guess.

I use a 4K monitor, and yes… that blurry preview at 70% would be enough to cringe and to think of a faulty lens (when the lens is actually fine), even without the colour artifacts of my examples.

[BHAYT (Bryan [WIN10, PL8, VP4, FP7, G9 & EM1 Mkii, RTX 3060])]

@Wlodek I am glad you are not in a situation where you are getting any CA, original or added by DxPL between 25% - 74% zoom, or present in the Loupe when DP XD2s has been applied to the image (the Loupe showing what will be in the final export) or in the final export itself.

With the original CA, my camera is a G9 and the lens an Olympus 12-200 (24-400 equivalent), the hope is that the photo editing software will reduce or remove that from the image.

But

1. DxPL adds a lot of CA to many of my images between 25%-74% zoom, i.e. not only does it not apply the CA correction below 75% as it states but it is adding to any CA present to a considerable extent.
2. At 75% zoom the CA, added by DxPL vanishes and it applies the CA settings which should help reduce any real CA that was present in my image.
3. But if I apply DP XD2s to the image then the Loupe will show an increase in the CA compared with a situation where DP XD2s has not been applied, the best example is from an earlier post where we have this with the Loupe at 100% over an image at 75%

First without NR being active:

and the second with NR active

where has the increased CA come from?

The answer is (must be) that it is being added by the now active DP XD2s.

I originally blamed the Loupe mechanism for this, hence it is included in the title of this topic, but eventually realised that the Loupe was showing what would be exported, i.e. the Loupe was showing the reality of what the export process would create with DP XD2s applied.

If you encounter none of this then you are lucky or a very skilled photographer or simply not taking photos at this time of the year. Your images are of a trees in leaf, i.e. this is an image of mine taken earlier this year (May 17th. 2024).

Tiny elements of what I am talking about are present but not to anything like the same extent as the other images above

but with a photo taken a little earlier (25th April 2024) we get

and taken this month

Sorry, I haven’t been there lately but Jan 18th. 2023 we have

[Joanna (Joanna Carter)]

What am I missing about all this fuss?

There are two points in the digital image production process

1. Where there are pixel pitch conflicts between the subject and the sensor
2. Where there are pixel pitch conflicts between the digital image and the viewing mechanism

The first needs manipulation, either in camera or in post processing. This may sometimes be possible in PhotoLab, but not always, depending on the pitch of the pattern and the pitch of the sensor.

The second is transient and is a natural result of image resolution vs screen resolution, where changing the viewing scale is sufficient to eliminate it.

Providing it is possible to eliminate such apparent CA at certain magnifications, there is really nothing any software can do to affect the final image - just change the pixel pitch of the screen if you find it annoying. If the image is destined for printing, it is doubtful such “aberrations” will be visible on the print.

[Lucabeer (Luca Chichizola)]

@Joanna

there is really nothing any software can do to affect the final image - just change the pixel pitch of the screen if you find it annoying.

I don’t accept this workaround for two reasons:

1. first, one would like to view and evaluate the image as a whole (with correct sharpness and colorus) while editing instead of being forced to view only a part of it (75% zoom or higher)

2. other RAW editors show a correct preview regardless of zoom level, so it definitely IS possible to do so if only DxO wanted to. Lack of accuracy and artefacts are not acceptable and not justifiable as “unavoidable technical compromise”, when the competitors manage to do much better.

The examples posted above show a COMPLETELY different preview compared to what the final output is. Completely different in terms of not only sharpness but also colour. And that is unacceptable. All the rest (ranging from “I don’t see/I am unable to see it in my images” to “just change your zoom level”) is moot. The objections to fixing this behaviour are of the kind of “my car starts shaking when I exceed 120 Km/h” and someone replies “well, don’t go beyond 120 Km/h!”. If you don’t go above 120 Km/h, how does it hurt you if the car maker also fixes the behaviour at above 120 for those who want to go at those speeds?

[JoPoV]

@BHAYT

The answer is (must be) that it is being added by the now active DP XD2s.

Did you try with deepprime ? Does it adds CA as Deepprime XD2s does ?

[BHAYT (Bryan [WIN10, PL8, VP4, FP7, G9 & EM1 Mkii, RTX 3060])]

@Joanna

@Joanna

Providing it is possible to eliminate such apparent CA at certain magnifications, there is really nothing any software can do to affect the final image - just change the pixel pitch of the screen if you find it annoying. If the image is destined for printing, it is doubtful such “aberrations” will be visible on the print.

I apologise if I am either mistaken or failing to make the point correctly but most of my comments thus far (please see below as well) have been that DxPL makes matters worse, at least with respect to what the users sees below 75%.

If it showed what I was actually going to get then I would have no legitimate complaint to make but is doesn’t, its shows any CA it finds greatly exaggerated in some cases and just exaggerated in others rarely even vaguely accurately!

So we have the “hobbled” sub 75% view not just “hobbled” but also corrupted!

If it stopped there then can I accept an editor that shows my landscape full of something which is not really there.

But then I discovered that applying DP XD (PL7) and DP XD2s(PL8) also appear to be either undoing the work of the CA edit or gratuitously adding CA into the ‘Loupe’ view and into the exported image.

and it gets worse!

In an earlier post PL8.1.0 Still adds Chromatic Aberration to images between 25% and 74% & to the Loupe - #15 by BHAYT I included this image

which shows a walkers coat having the colour removed at more extreme settings of CA in DxPL (V7 and V8).

I went looking for an image of Highdown Gardens from about this time of year to see how that would look and found this image of a similar view (looking back at the entrance gate) to the last images in my previous post.

P1109872.RW2 (23.1 MB)

I was concentrating on doing the same tests as I had done in the last post until I realised that at the default CA settings PL8 was corrupting the colours of the image.

Testing with ACDSee Gemstone and Zoner they manage purple fringing removal without (apparently) “damaging” the image or at least not to the same extent but that is not the case with DxPL!

but what is going on here

@JoPoV I have substituted the DP export it is the last image in this bundle and it looks to have no new issues!?

[Egregius ([Greg] (Micro 4/3 + Win11))]

@BHAYT

I was concentrating on doing the same tests as I had done in the last post until I realised that at the default CA settings PL8 was corrupting the colours of the image.

While you have provided extensive information, Bryan, with a lot of repetition, I think your presentation is becoming rather hard to follow. It might be a case of too much information. Take what you said here for example. You said that at the default CA correction settings PL8 corrupts the colors. But at this point I don’t believe you are talking about the image viewer zoomed in at 25-74% anymore, since DxO states that CA correction settings are only applied to the viewer at 75% zoom or greater. So what are you talking about? The loupe? The exported image? The image preview at full zoom? Or are CA correction settings actually impacting what you see in the image viewer at <75% zoom, contrary to DxO’s published claims and common experience?

You also bring up some other complaints that don’t correlate. The fact that “more extreme settings of CA” correction can desaturate or remove color from parts of the image (in your case, a distant person’s red coat and to a lesser extent a nearby person’s blue jacket) is an unwanted side effect, but to be expected with extreme corrections. That’s why the default isn’t the maximum setting. RAW editing is frequently a tradeoff between the correction you want and any unwanted artifacts. This problem is at best tangential to what you’re trying to argue in this topic.

I think all the work you’re putting into this is upsetting you a bit, because you don’t see DxO’s response as proportional to your own efforts. Am I right? I get the impression that you want DxO to take the problem more seriously and make fixing it more of a priority - but there’s only so much you can do to influence that. Making your case here in the forum might help, but it’s becoming exhausting IMO.

[JoPoV]

@Egregius

While you have provided extensive information, Bryan, with a lot of repetition,

I think @BHAYT had to do repetitions because he had to explain again and again it’s problem to people who did not read well it’s messages.

What he said is very simple :
On some images, DPXD2s adds CA like problems which does not exist without DPXD2s.

[Joanna (Joanna Carter)]

@BHAYT

DxPL makes matters worse, at least with respect to what the users sees below 75%

But my point is that, if zooming in PL eradicates the CA, it is purely an artefact of the relationship between the pixels in the image and those on screen. But the actual image remains CA free.

[Lucabeer (Luca Chichizola)]

@JoPoV

hat he said is very simple :
On some images, DPXD2s adds CA like problems which does not exist without DPXD2s.

Actually, and better:

1. On all images, PL shows in the viewer between 25 and 70% a lot of CA (and a marked blurriness) that isn’t actually neither in the original image nor in the exported file

2. On some images, DPXD2s adds CA like problems which does not exist without DPXD2s, and this translates to exported images too.

Me, I have noticed only 1 and never 2, but the examples provided by BHAYT are very interesting.

[bildbaendiger]

I download the picture and test it.

@BHAYT

P1109872.RW2 (23.1 MB)

I can confirm, that there is a big increase of display CA when i displayed the image with a zoom factor between 25% and 75%.
This is a issue of the display calculation and should be examined more closely.

But I must say:
The image is not free of CA, on the contrary, it has a lot of CA in exactly those places.
There ist absolutely no adding of CA with or without DPXD2.

[Egregius ([Greg] (Micro 4/3 + Win11))]

For what it’s worth: when an image contains moire, I have seen DeepPRIME XD2s and other variants of PRIME add false color in differing ways - but it’s actually an artifact of demosaiccing. (PhotoLab folds NR into the demosaic process.) DPXD2s has been proven to generate worse false color than DPXD, though it also can render fine detail in these areas better.

[bildbaendiger]

That was also my point:
The image contains a lot of CA, and of course these are amplified by the display and further processing.

By the way:
I see no point in processing an image that contains hardly any noise (ISO 200) with DPXD2s.
To emphasize again: I could not find out that DPXD2s adds any CA. You just see the existing CA of the input picture.