In V5 processed images look different in V6

@Hugo

The same way as I did in → In V5 processed images look different in V6 - #24 by Wolfgang,
in PL5 I applied a few different renderings to a virtual copy of your pic.

  • with “5 - Landscape - Standard”, then shown as
    grafik
    no problem in PL6

  • with “3 - Neutral colors”, then shown as
    grafik
    no problem in PL6

  • with “3 - Portrait - Candy colors”, then shown as
    grafik
    different colours in PL6 !

( As in PL it is not possible to compare the generated TIFFs instantly, I did so in PS. )

1 Like

@Hugo Please raise this fault via the DxO support channels because raising it here gets you opinion, advice and “advice” and confirmation but I am simply not sure anyone in DxO is monitoring the forum anymore, let alone actually preparing to react!?

2 Likes

I raised the issue through the support channel. I hope the are able to fix it soon. The number is 387371.

@Hugo good here’s hoping there is an “easy” fix.

During Beta testing, the issue that making any change to existing customer presets was to be avoided as much as possible seemed to be accepted by DxO and the ‘Wide Gamut’ feature was made optional but something seems to have slipped through.

Whenever new features are provided, particularly when they involve image editing, it must be possible for the user to continue with the “legacy” features wherever possible until they decide to adopt the “new”, or not, as they choose.

Not that all developers of image editing software even keep the interface the same going from one release to the next!!

I tested @Hugo’s image in DPL5 and DPL6 and tried to find differences. This is what I did:

  • In DPL5

    • Applied No Correction
    • Set Generic Renderings / DxO Camera Profile (Oly…)
    • Manually exported the DOP/settings sidecar file
    • Exported the image as 16 bit TIFF
  • In DPL6

    • Manually imported the DOP/settings sidecar file
    • Exported the image as 16 bit TIFF
  • In Photoshop

    • Opened both images as layers with 32 bit/channel
    • Subtracted the two layers
    • Set level 255/255 to 3/255 (amplification of 64, or 6 stops)
    • Set curves 255/255 to 63/255 (amplification of 4 stops) for a total of 10 stops
    • Raised exposure to150 in order to lighten the resulting image
  • ** Result**

    • A perfectly black image, which means that both exports contain identical images as measurable with the procedure outlined above!

Doing the same steps as above with a different image, I get two seemingly identical images:


Note: The histogram did not even blink when I switched images…

BUT…

Going through the procedure in Photoshop, I got this:


This is not a perfectly black image, which means that differences exist… I still needed the >10 stop (roughly 1000x) enhancement to make the differences visible. A difference of 1/1000 cannot be distinguished by an average human eye/brain, that combo is good for about 1/100 only.

Lessons learned

  • Images that have been migrated from DPL5 to DPL 6 and exported without further adjustments can be identical, except images containing fairly saturated colours that fall between Adobe and DxO Wide gamuts. Hugo’s image has no such colours.
  • I tried different target profiles in soft proofing and found that AdobeRGB and SP (off) produce identical results, no matter if I set perceptive or relative rendering intent (which might be meant to act on paper simulation only)
  • Export was done with ICC = “As shot” in all cases

Tested on macOS Monterey 12.6.1 in iMac 2019.
I could also test on a M1 MacBook Air running Ventura, but I don’t feel like it right now.

3 Likes

Just a couple of “FYI” comments;

Having Soft Proofing activated or deactivated makes no difference to the exported result - - the SP setting impacts only what you see on-screen/within-PL … because;

  • within PL, it determines whether or not the Protect Saturated Colors Algorithm is applied (See more about the PSCA here)
  • whereas, when Exporting-to-Disk, the PSCA is always applied.

Similarly, the “Intent” setting;

  1. I agree with your suggestion that it relates only to the “Simulate paper & ink” setting
    – That is, presumably - - we’ll know for sure “soon” !
    – The current UI layout for this setting is confusing (I’m guessing that will change … hopefully !)

  2. Does absolutely nothing in the current version of PLv6 … instead, the concept of “intent” is handled by the Protect Saturated Colors Algorithm (See reference above).

So, the result may have been exported with an AdobeRGB profile (if that was the camera setting).

… … … … … … … … … … … … … … … … … … … … … … … … :laughing:

John M

Thanks for all the effort you put in. It seems that the issue is (only) related to generic settings - Dxo portrait.

  1. When I open an image in PL6 it looks fine.(screen shot 1)
  2. When I switch to color space classic it becomes a bit brighter (screen shot 2).
  3. When I apply generic renderings - DxO portrait, the image is muted (colors, brightness, contrast) - Screenshot 3
  4. This differs from PL5 - Screenshot 4

    The differences are clearly visible
1 Like

Colour renderings are intended to change colours and tonality of images. Depending on selected rendering and WCS, images change more or less. This is to be expected.

The issue I have with DPL6 is, that renderings use names as in DPL5, even though they render differently, and I hope that DxO wil come up with respective fixes and communications.

At the time being, we are confronted with new half-baked features in a paid release. Bugs happen, proposing half baked features is a deliberate choice and bad manners imo.

6 Likes

@platypus Please explain the following that from exactly the same starting point (eventually) the same image produces different results when the same DxO supplied preset is applied.

This is not, repeat not, the same issue that @Hugo encountered (although it may be!?) because in his case he was taking a PL5 DOP over into PL6 when he encountered an issue!!

  1. Take a “boring” photo and write to two directories PL5 and PL6
  2. Open PL5 directory image with PL5 and apply ‘1 - Portrait - Standard’ and snapshot and export
  3. Open PL6 directory image with PL6 and do the same as in 2.
  4. Compare the images with whatever tools at your disposal, mostly with eyes!!
  5. Repeat tests because I had become confused after initial PL6 change but before export and opened PL5 directory and discovered the images were not starting from the same point!

PL5 Test 3:-

PL6 Test 3:-

The exported images:-

The comparison:-

The differences are “interesting”, tests with other homegrown presets do not typically cause an issue but this, and possibly other DxO presets are causing images to change dramatically between the two releases, given that they arguably shouldn’t have changed at all.

I don’t necessarily consider that the other changes in DxPL should have caused this but @DxO_Support-Team if you are still alive a response would be nice!!??

PS:- The reason for my multiple attempts was principally because PL5 contained lens fixes and PL6 didn’t!

I now realise that ‘1 - Portrait - Standard’ must be a ‘Partial preset’ and the PL5 starting preset and the PL6 starting preset were not the same, hence they were both set to 5 - No correction for the last round of tests!!

@platypus There are differences between images using home grown presets they are just a bit harder to see - IGNORE wrong preset used on PL6 test, Homegrown 56 - … instead of 55 - …!

Instead of running the test, I had a look at a few of the roughly 200 presets that come with DPL5 and DPL6 respectively. I found presets of the same name that have more or less differences, which might (or might not) explain, why looks will be different in DPL5 and DPL6.

While DPL6 is still incomplete, I’ll refrain from running yet more tests that will show (or not show) differences between the looks. DxO has a hell of a lot of things to iron out and I think we’ve made it clear that some of the unripe bananas are in the colour handling basket.

2 Likes

@platypus agreed but please ignore the above test I managed to apply the wrong preset in the PL6 case!! So clean slate and correct preset in both cases and we have

But for the previous tests we have

One of the biggest takeaways of all of this, in my opinion, is that it is approaching shameful that DXO doesn’t have an official presence on this forum to help with our understanding of everything PL6 vs PL5, as well as the dng issues/questions. Someone with some expert, inside knowledge could relatively quickly lessen a lot of the dissatisfaction and confusion…including my own…that is so very clearly expressed on this forum.

DXO is missing out on a very valuable opportunity to reinforce and expand their customer base. They will never be able to have a customer base with a significant number of working professionals without being more involved and present here. The advanced amatuers and hobbyist photographers know enough about color management and the need for full dng support to lose confidence in investing (or reinvesting) their hard earned money in the DXO universe.

Please join us in these discussions DXO.

Having said that, I really like PL6, as I did PL5 and PL4. I am not about to change yet as there really isn’t another raw converter to compare…yet.

8 Likes

It looks like a policy decision with V6 to effectively stop having a presence here. If so its one of the more stupid ones DxO has done as there are some valuable discussions going on that I would have thought of value to them if they had any intrust in how their products were being received and problems.

5 Likes

@John7 and @bconner if it is a deliberate policy then I would contend that it is a particularly self-destructive one!

I received information that @sgospodarenko was assigned to another assignment and that the @DxO_Support-Team was now the point of contact. When using this @DxO… you get a warning that it is assigned to 4 people and there are no responses in the forums from this as a group but, presumably individuals who are part of that team can respond individually, as can any other DxO employees who are assigned the task of investigating or assisting.

Unfortunately the more I am ignored the more I want to elicit some sort of reaction, any sort of reaction would be good at this point, but …

In the meantime we will just have to keep ourselves occupied/amused without any DxO expert assistance/guidance. The danger is that all or some of these posts find there way onto the greater web, that will include real problems, “imagined” problems and gripes like this one!!

Regards

Bryan

3 Likes

Its a pity, whatever the downsides of the new Affinity there forums get fast feedback from staff even if you may not like their response. I know this it is there support but there is input over a wide range of topics. I have even seen postings saying they have got something wrong! They also have a lot of new videos on the new version something lacking here. As I understand it the size of the firms is about the same (if you believe the internet information DxO actually has more!) so I wonder how they can do it but others fail?

1 Like

@Hugo I did what I should have done way back which was to compare PL5 preset with PL6 presets and when I did I got this

which clearly shows some major “upheaval” with the PL6 DxO presets!?

I attempted to create a new preset by applying the ‘1 - Portrait - Standard’ to an image in PL5 and then creating a new preset and then importing that preset into PL6 (looking for a workaround)!

The result was essentially the same as applying the original DxO preset in PL6!

So new test(s)

  1. Single image in PL5 directory and PL6 directory both start with ‘5 - No correction’
  2. For each create 6 VCs
  3. For each VC assign from the following table to VC

image

Please note the difference between PL5 and PL6 menus -

  1. Export the resultant images.
  2. Compare and contrast!?





None are as extreme as VC[2], the ‘DxO Portrait’ rendering, but I was hoping for better consistency between releases, i.e. we want no colour to be present in the comparison window and that was not the case!

Repeating the tests with my own ‘055 - SL(7St) CV(20) C(0) Mc(5) LS(0)’ preset produced the hoped for grey comparison screen on most ‘Rendering’ options, except for VC[2], which gave the following yet again!

and again when I created yet another test directory and tested a [M]aster image with my preset and the ‘DxO portrait’ ‘Rendering’ which does not produce the same result on PL5 and PL6.

and except for whatever is happening here!! unless that is simply jpeg artefacts at work!?

Which appears to be happening on all the otherwise equal comparisons!?

So can I please quit now 'cause it is all getting too much for my old brain and I am not being paid nearly enough!

The one area of DxPL that we emphasised, during Beta testing, that must be treated as sacrosanct was that it should be possible to put images into PL6 and be able to get the same results out as was happening with PL5!

Hence it should be possible to use the same workflow and get the same results!

Changing the workflow to take advantage of new features and get enhanced results is acceptable but it should always be possible to ‘retreat’ to the old workflow with the new release not have to revert to an old release!!

As always I may have done something wrong in the execution of these tests but I have tried to rectify errors wherever possible!

1 Like

Wow, you put in a lot of work. Thanks! Let’s see how DxO treats the issue: request #387371 (https://support.dxo.com/hc/requests/387371). I hope you can follow it if not issued the request. Otherwise I’ll keep you informed.

@Hugo you can keep me up to date. The above rambled because I wrote it as I tested and retested and …

So the summary is

  1. If using the ‘1 - Portrait - Standard’ Preset from DxO there appears to be no direct correlation between the two releases, all outputs will differ between PL5 and PL6 with the ‘Render’ of ‘DxO Portrait’ causing the most (major) difference!!

  2. If using your own presets then all but the ‘DxO Portrait’ render appear to be compatible between the releases but ‘DxO Portrait’ appears to be way off again!! Notwithstanding the strange random dots that appeared to crop up!?

That should have appeared at the front of the post but hey ho!!

This lack of DxO’s presence on the forums is a major “downgrade” to the helpful, friendly relationship we used to have until the PL6 cycle. And it’s not just in the public forums, the beta forums were deathly quiet as well.

As a result, I will be installing my beta reward licence instead of paying the upgrade price as I used to for previous versions. I will use this only for helping others in these forums, staying with using PL5 for my production work until DxO get their act together over explaining this colour pipeline stuff. Dxo staff don’t have the time to respond here? I don’t have the time to continue finding nasty surprises part way through working on a project.

I also find myself liking what I see in Affinity Photo 2. As some have said, their support forums are far more responsive.

3 Likes