I do hope that DxO does more thinking about backwards compatibility though, since having to review/adapt previously edited images after an upgrade is seriously unsatisfying. I’ve been looking through the images I’ve edited in PL2 and am finding differences between the PL2 and PL3 rendering when I’ve used the repair tool. Nothing as bad as a third eye in my case (yet), but enough to bother me. Even if changes are small between two versions, over time (ie. multiple upgrades) they might not be, so best effort is really not good enough.
Either that or document that all editing is version-specific and that export may be required to retain an existing rendering. That’s a valid choice, but it would be a significant drawback for me at least, and probably more than me I suspect.
I really don’t understand when other programs build backwards compatibility in but PL doesn’t. I rember Affinity users complaning a beta hadn’t done so (they are cleare that beta programs may make ajustments that may/will not be compatable with earlier versions which is understandable) and the support staff accepted it hadn’t and rapidly produced an amended version to protect compatablity. This has been raised a number of times and as with the current problem with healing points been ignored. My work flow is to depend on the dop for adjustments, I don’t keep copies other than on flicker and some prints so I can live with miner veration but more than that to have to recreat a imige because of lack of backward compatablity is just not on.
I don’t think even minor variations are acceptable, just because four or five of them over a few upgrades might not be so minor.
Either they add the necessary machinery to be able to respect the rendering of editing done in a previous version of the software, or dop files are version-specific in practice. If I need to export everything as tiff or rework previous edits then there are better alternatives than PhotoLab unfortunately, however much I like the editor.
The worst I’ve seen in Skylum Luminar. I thought Luminar 2018 was a nice editor, but I dropped it pretty quickly once I discovered that rendering would change even after minor updates. Completely unusable for anything but export and forget.
I’ve only been around since PL1, so I don’t know how good they’ve been at backwards compatibility in the past, but I’m after something better than possibly minor variations between upgrades, which is the answer I’ve received from support. Others can do it, so why not DxO? If they can’t/won’t then fine, that’s a choice, and then PhotoLab is probably not for me. I don’t find any differences that require me to re-edit a potentially large number of images to be minor, but we all have our own red lines.
I think it’s much more than a minor problem if i’m honest. There were 15 out of 300 photos that had significant unacceptable healing discrepancies. Then there were the “minor” ones too. I actually decided just to reset all the healing on all 300 images and re-do with the new healing algorithm for peace of mind.
It would appear that DxO have accepted it’s a problem “I see the problem now. I’ll send your data to the FW team for the investigation.” and “It will require the change of the significant part of the behavior so you’ll have to wait”. For me it’s the fact they didn’t really take into account that this could be a problem. Until a fix is made I am unclear as to what happens to and correction that has to be redone, will the fix create yet new problems?
The thing is, I sometimes go back to “finished” photos in PL and produce new exports using different crop ratios for different purposes. i.e all my photos are cropped to 7x5 for printing into the family album, but I might go back to one and temporary re-crop to a different ratio for a canvas print etc. The photos are never truly finished.
Personally, I can’t see it being a hard thing for PL3 to simply respect the chosen healing source point that PL2 decided to use and save in the DOP/database.
That fits PL in the current state, but not everyone wants to maintain an archive of processed RGB images alongside their raw archive in order to retain existing rendering, or lose the ability to revisit previous editing after an upgrade. It’s not necessary with other editors (can’t speak for all, but C1, Exposure, and LR all seem to take backwards compatibility more seriously), so not an advantage for DxO.
I did have a problem with upgrading lightroom once to be honest. If memory serves I think it was upgrading from v4 to v6 where v6 used a different algorithm for red eye removal. All the red eye removals from v4 came out wrong (my old cameras produced some red eyes, lol)
Ok, was that without changing the process version? I never use flash myself so that’s not something I would have run into.
I went from Photoshop CS2 and LR3 to CS5 and LR6 without any problems, then jumped off the Adobe train with subscriptions and have bought/used all three of Exposure, Capture One, and PhotoLab since then. I tried Luminar and ON1 along the way, out of curiosity, but they’re not really in the same league in terms of reliability IMO. (Like I said earlier, Luminar was a disaster in terms of backwards compatibility. ON1 has just been buggy when I’ve trialed it.)
I’ve had a hard time choosing between the three since they all have their irritants (eg. I recently discovered how easy it is to overwrite existing edits with C1 when using a shared filesystem), but I have an especially hard time with PL’s relaxed approach to backwards compatibility, I’m not a fan of the licensing (lose your old license after upgrade, which is an especially bad fit when backwards compatibility isn’t solid), and the rather aggressive dropping of macOS versions is another minus (drop another version with each PL release; C1 is no better, Exposure is much better). The support has also been worst-in-class in my experience: little effort to understand/investigate problems, never any feedback on if anything reported has been addressed (which has rarely been the case), and then poor release notes: “minor bug fixes” says absolutely nothing.
I like the results though, so here I still am two upgrades on from PL1, but I can’t complain about with C1 or Exposure either, so I may well drop PL if I continue to see upgrades that force me to review/revisit previous editing.
Affinity Photo is another one I use, but mostly for stacking so far. The lack of sidecars and big afphoto files makes it less appealing for raw development, but if they were to do something about that and the destructive step out of the develop persona then that’s some seriously stiff competition. People complain about the quality of the raw processor, and I admit that I haven’t looked very closely, but it seems to me that you do have to look quite closely to find deficiencies: I haven’t really seen anything to complain about.
Personally I’ve tried all the Raw software out there, and in my opinion nothing comes close to PL (especially for my noisy crop sensor cameras). I switched from Lightroom. The noise reduction & lens sharpness in PL are on a different planet, the lens distortion corrections perfect, the automation etc, but most importantly I actually enjoy using it. It’s simple to use too. This backward compatibility issue of the healing tool has been the only issue for me, and I’m sure they’re looking into it for a solution, rather than it being deliberate.
I don’t disagree with your assessment of PL - there’s no question that it’s very good - but I get good images out of the others as well, so for me backwards compatibility is a big deal. You’re much more generous than I am in being willing to redo repairs on 300 images.
I’ve scanned through all of my images over the past week and it turns out that I don’t really have that many that use the repair tool, but in those that I do I certainly see differences between the PL2 and PL3 rendering. I’m sure it isn’t intentional, and I do hope DxO sorts it out (they now have both PL2 and new work done in the existing PL3 to take into account), but they just don’t seem to take backwards compatibility as seriously as I think they should, the case that John7 pointed at being another example. (In which there was, somewhat surprisingly, no response from DxO.) I’m just not happy with having to go through my images after each upgrade to find what might be affected this time around.
The bad news is that I’ve looked at the HSL tool, that is also changed in PL3, and see differences there as well. These are admittedly very minor, but they’re there. For example, in one case I’ve reduced blue saturation, and PL3 renders it a little darker than PL2. This is certainly nitpicking - I would have been happy with either rendering when I edited the image originally, and it’s surely smaller than would result from editing the image again from scratch to arrive at a pleasing result - but this is the kind of thing that can lead to an unacceptable deterioration of your images over time, upgrade after upgrade. This time it’s a little bit more blue and different repair, next time it’s shadow recovery, then white balance, etc, etc. Software-induced bitrot.
To be fair, I haven’t given Exposure of C1 the same grilling after upgrades, but mostly because I haven’t seen any evidence of problems. I’m going to take a close look to see if I can find any differences with the new C1 20 vs 12. If their backwards compatibility has holes as well then this RGB archive I’d prefer to avoid may be inevitable.
I think DxO just need a bit of time to remedy the healing problem. They didn’t know about it until it was pointed out & so I don’t think it’s deliberate coding, just poor testing. Bugs happen all the time in all software.
Nope, that’s not true at all. The algorithm was changed and the migration of the previous settings is not 100% equal. We know about that and in most cases it is still acceptable and effective. But for example, for your case I agree it’s not but it’s still can be fine-tuned manually. To introduce the behavior you want to see needs time and efforts. I have already told you, that we agreed we should change it. But you have to wait.
That is, you change a tool or algorithm, are aware that this may change the rendering compared to the previous version of PL, but think that in most cases this is acceptable. (Which is where I disagree, obviously.)
But for example, for your case I agree it’s not but it’s still can be fine-tuned manually. To introduce the behavior you want to see needs time and efforts. I have already told you, that we agreed we should change it. But you have to wait.
This I interpret as saying that the changed rendering with the repair tool in this case is too much, so you will work to improve it. Still no guarantee that the rendering of work done in the previous version of PL (not just repair, but anything) should be unchanged, just best-effort in trying to minimize the differences to what DxO deems acceptable.
Correct so far?
So to be clear the policy isn’t to protect past adjustments but hope they will be sort of OK with a change?
Not really a hope. The tests were done on thousands of images. But anyway, I want to complete this discussion as we are going to change the behavior for this case.
Not just a hope, but certainly no guarantee either. You change tools and algorithms and then try to tweak/test your way to rendering that doesn’t change things too much. How is that not what John7 is saying? You’re just trying to minimize changes in rendering, not eliminate them by design.
Like I said before, the problem with this is that the change from PL2 to PL3 may be minor, acceptable even, but all these minor changes might not be so acceptable by the time we get to PL7 or PL8. In PL2 it was ClearView Plus (apparently; I didn’t experience it myself since I didn’t use ClearView), in PL3 it’s repair and HSL, and future upgrades will bring new changes to who knows what.
This is completely different from Capture One and Adobe, with their process version. It may not be as bad as Luminar’s rate of change, but you run the same risk of eventually ending up with a different rendering than the one you originally arrived at.
If best-effort is all DxO intends to provide, then exporting before an upgrade really is the only way to retain an existing rendering, and DxO would do well to make this clear in the documentation. The benefit of saving small sidecars is only a benefit until the next upgrade: after that I could just as well have worked/saved in Affinity Photo or any other pixel editor if I’m not willing to review all of my existing work after the upgrade to see what might have changed this time around, and then re-edit if/when I find something disagreeable.
Edit: Case closed, so that makes it pretty clear what the answers are. As much as I like PhotoLab as an editor, you can’t compare it to the applications that take backwards compatibility more seriously to retain the rendering of images upgrade after upgrade. It’s DxO’s choice, unfortunate as I might find it, but the way the PhotoLibrary functionality is marketed (“everything you need for perfect asset management”) is fairly dishonest unless you only manage exported images: I don’t think anybody will be happy discovering that their raw edits are slowly mutating over time. Then again, they claimed raw support for Nik by selling it with a copy of PL, so I’m not that surprised.