I’ve been working with PL 10 for a good three weeks now and have processed just under 1,000 images with it during that time.
My verdict is damning.
What DxO has delivered here is a joke. PL 10 is half-baked software, and we are the beta testers.
The interaction between the software and the graphics card is completely off. The AI masks—of which you only see a small initial selection anyway, depending on the image—are highly unreliable, and the workflow slows down dramatically.
This just won’t do. It is an absolute cheek to launch a half-finished product like this.
I have to chuckle, as PL10 is already in a far far better place than PL9 was when it first launched. I remember months of issues before that release ran stably for me, but I’d feel confident working from PL10 as of right now. It’s stable on my (admittedly aged) machine for the most part.
I do have two issues, one periodic, one constant:
- When switching between folders of shots, there’s sometimes a long delay or even a crash, before the images can be loaded/previews generated/work can begin.
- Exporting in DeepPrime XD3 takes a long time (something like 3x longer) to export each image vs. export times in PL9.
I’m finding AI masks generally OK, but I tend to stick to dragging a selection rectangle over the objects I want, not relying on auto-detection, which can take a while to (presumably ‘scan’) my image and allow me to select anything.
Workflow actually seems semi smooth (again considering my older machine). I do find it slows dramatically if I locally adjust Lens Sharpness Optimisation or “force details” in the denoising sections. If I leave those as global values, performance stays predictably steady.
PhotoLab seems to like to generate an entire new preview the moment any change is made, and it seems some things take longer to calculate than others, so changing them means every other change also takes longer to make as all these individually minor delays can easily become big delays overall.
Learning how this works and how to edit with it in mind has kept things smooth as is possible for me.
I didn’t realize it is maybe a little bit challenging to do a such feature since masks should be generated at pixel level (after demosaicing), and PL should need mask before demosaicing for sharpness optimisation since it seems demosaicing/denoising/sharpness optimisation is a one step Ai process.
Unless they do apply the Ai proceess several time depending on masks.
Maybe my logic/pl knowledge is wrong here.
EDIT : I think the second way is the used one. Seems the easiest to implement seeing how PL works.
I agree : Photolab is absolutely unsuitable for large workflows.
Sadly it seems either this is not DxO target, or they are not able to do that.
Try pureraw for that if you like DxO demosaicing/denoising/blablabla.
IMO PhotoLab shouldn’t apply any such sharpening or de-noising until the point of export. Doing it while I’m editing doesn’t do anything but slow my experience with the software. (Especially since I seldom need to fine-tune these settings, so I don’t need them shown to me in every other edit I make)!
When dealing with very high iso shots (where PL is the best), it can be hard or even impossible to work without denoising in preview.
I couldn’t say how often I run in to that!
But then I guess the obvious solution is to let us control what tools are included in the preview / any renders that take place before export.
Then we can pick our own poison, so to speak! 
Being able to choose the order tools are applied (where it make sense) would be great.
You can already do that : don’t check denoise/sharpen until you export.
Just make a partiel preset with denoise/sharpen and apply it to all your images just before export if you want a fast solution, or tweak this for each image before export if you need a more precise control.
Hmm interesting idea! It feels a bit more cumbersome than I’d like (I’d worry I’d forget to apply it, or it’d overwrite something, which might be irrational).
I’d rather it was something that quietly worked in the background, but what you’re saying does make sense! I’ll have a play with it.
just check this with virtual copies.
Whether PL suits your needs or not depends on your needs. Yours seem to be somewhere outside of PhotoLab’s capabilities and then, it’s sensible to NOT use PL.
And that is exactly why I’m still with Lightroom Classic and use PL for the features it excels in. PL is not the app that covers my most important requirements. So what!
Well, it’d be nice if it did, wouldn’t it?
I’d rather only need one bit of software than two.
Look at the good part: approx 1000 photo processed. Great!
The interaction between the software and the graphics card is completely off
May the community helps. What its means, the ‘completely off’?
Things may related with GPU VRAM memory amount, what is yours (system)?
The AI masks—of which you only see a small initial selection anyway, depending on the image—are highly unreliable, and the workflow slows down dramatically.
May attach (link to some drive/cloud) a few sample photo - may its helps for the forum community to better understand what the ‘highly unreliable’ means.
The ‘Sky’ pre-defined AI mask definitely far from good (you can read about that quite a few times in the forum), other AI masks seems works just fine. Tractors (no kidding!): seems may a bit more training (AI) needed in the case of tractors, usually works fine, but some case its miss a few tractor out.
Not that it excuses how far behind PL10 is from the competition, but you should have seen PL9’s state when it launched… it would put things in perspective.
Nothing has been said that surprises me but I still find the results from PL better than Lr. Maybe that is down to the “Pure Raw” element but I am not convinced that is the only reason.
A lot of people trying PL for the first time expect it to work the same as the software they are used to. Take the time to learn how to use PL properly and then you will realise how good it actually is.
Control Lines and Depth Masks are some of the most useful features that have been introduced in the last few versions. Control Points are still some of the best tools for subtle adjustments. Control Points can now be adjusted into ellipses which has been a long time requested feature.
The point I want to make is: don’t expect software to work the same as the software you are used to. Find the software that works for you and don’t complain about software you are testing before fully testing. Also make sure your hardware is up to spec to run the latest software now that AI is so prevalent these days.
I have tested masks in PL10 trial. Not as extensively as I would have like (lack of time and I only have 3 days trial left and am working on other things now), but I can say with this flexible system combining AI and specific PL tools that no other software has, I have been able to mask ANYTHING I wanted VERY easily.
I din’t test some difficult masks like hairs in different conditions.
Those who aren’t used to graphics software and mixing layers and those who are new to PhotoLab and aren’t familiar with its specific tools for this purpose will need to get used to using it, but once you’re familiar with the various options and combinations, it becomes extremely simple and fast to create nearly any kind of mask.
I have suggestions on certain points, particularly regarding accuracy with some images with a shallow depth of field, and on the diffusion setting for example. I won’t go into that here and now – I don’t have time to flesh that out.
And I am confident DxO will make it to evolve and is putting it in its priorities.
So, for me, this masking sytem is already powerful and very promising.
Many thanks for the numerous replies.
A few words about my background:
I’ve been using DxO since PL6 and have always been satisfied. I regularly process large batches of high-resolution image files. Editorial teams hire me because I reliably and quickly deliver quality images from events of all kinds. My “slots”—the deadlines by which print-ready images must reach the editors—are usually very tight and ambitious.
For me, speed is essential, and I need to be able to rely on my software. I don’t have time to denoise images in one program and then finish editing them in another. PL is my go-to RAW converter for everything. Whether it’s sports, wildlife, weddings, concerts, or reportage of any kind, I process everything using PhotoLab. And that’s how I’d like it to stay.
PL9 ran stably for me right from the start. As far as I know, DxO wasn’t responsible for the issues PL9 experienced in the first quarter of 2026; rather, it was an interaction issue with Nvidia drivers. After reverting to PL 9.2, everything ran perfectly on all my machines. I work on a laptop with 32GB of RAM and an RTX 4070, as well as a PC with 64GB of RAM and an RTX 5070 Ti. PL 9.xx ran stably and quickly on both systems. From the end of March onwards, the latest PL 9 updates also ran perfectly again.
I process an average of over 300 images a week for business and private clients, often under significant time pressure. I need to be able to rely on my tools—and I can’t rely on PL 10.
That’s the problem. Sure, I can make do with U Point, brushes, and other tools when AI masks are unreliable or fail to work altogether. I can do that when I have time and only need to finish a handful of images. However, for me, photography isn’t a hobby for creative self-expression or relaxation; it’s the job I use to earn a living.
Well… then just keep working with PL 9.