Since the evolution rate of PL and its upgrade policy adapted to this makes upgrading every two years a reasonable choice, I think DxO is absolutely right to use standard computers currently available as their target platform and no less.
32 GB ram can be consider standard for small actual graphic stations (and frankly, I think that’s the bare minimum for low demand graphic station - not standard.). 5xxx nvidia serie, 4xxx nvidia serie and 3xxx Nvidia serie works like a charm. (3xxx is the one I tested PL10 with and it was running smoothly with ALL preview options enabled).
So I don’t think DxO is wrong here.
You can’t choose the “best” AI demosaicer/blablabla on the market and ask it to be designed for ten years old stations. That seems SO obvious.
Absolutely understandable, unfortunately I can’t endorse or cope with them as a company. They’re just awful.
This is further compounded by the recent launch of their “Prompt to edit” AI engine in Lightroom. Using this function, users can boot up a RAW photo and just type in what kind of edits they want and LR will do it for them. It completely mocks the photographic profession, the skill and knowledge we all worked to attain.
That’s not to judge you using it, but I just won’t go near it. I can’t.
I disagree.
It’s not the end export times I have an issue with - in fact for me PL10 has been spitting out some exports in as little as 15 seconds (for me this is great, PL9 often takes 20+ seconds per shot).
It’s the usability of PhotoLab that can be painful sometimes (often predictably more painful as more masks are introduced or over a longer editing session). PhotoLab is doing things that I’d turn off in a heartbeat (applying sharpening etc. to image previews - I can work on colour editing without needing absolute sharpness in my previews).
Meanwhile Capture One remains fast and smooth throughout an editing session. It’s a breeze to use.
I think it’s fair to say that PhotoLab could use some in-editing optimisation to put it on a par with the competition. If it then wants to take a little longer to actually export because of all that brilliant work it does, that’s fine by me, but it doesn’t need to be happening mid-edit.
Well that’s the problem, is it not? £500 on RAM alone as a minimum is a big ask for users, when the competition has software that runs better on less.
If I were to build a high end system from scratch today and even went vaguely frugal and didn’t grab a 5xxx card out of financial masochism, I’d still be looking at thousands and thousands. Systems are very expensive now, so it doesn’t do to be glib and expect your customers to simply have that kind of money (and build for it).
For example, the Chinese have just set up a radio telescope station operated by AI that is responsible for conducting observations and selecting all of its observation and research topics completely on its own.
No more scientists needed to conduct the research.
Then what’s next step ? Ai will share knowlegde with what ? AI ?
Make all your presets (if you use some) without settings you don’t need to preview (no deeprime XD3, no sharpening ans so on) and work from them.
Then create a partial preset with those disabled features for apply them with multiselection to all images before export in one step before export (or tweak them at this stage for each individual image as you would have done if this is your need - this shouldn’t break your workflow since you don’t want to see those function effects in preview).
Indeed a one button function would be a little more appropriate, custom shortcuts to manage all that would be even more appropriate and would give more opportunities, possibilities to create macros would expend this capacity too, a drag and drop function to create custom buttons the same, but anyway the possibility is already there and not that much complicated.
I can’t argue with that but there are not a lot of alternatives. Capture One does tick most boxes but its pricing is astronomical. That is why I tried so hard with PL but the lagging just drives me mad. I fear the “prompt to edit” functionality will drift into most apps in the end - I think Luminar has it now too. It can be ignored.
So I’ve just tried by making a couple of presets (one with my regular changes in it “Preset 1”, and another specifically containing denoising etc. “Preset 2”)
I edited a photo using Preset 1, then when I was finished I applied Preset 2 over the top of it.
Unfortunately it removed all my previous edits (I could recover easily enough with Ctrl+Z, but that just put me back to where I was before).
Any idea how to apply a partial preset, as you put it, that only applies the tools that are active in it, without overwriting everything else?
I’m so tired of this less than useful comments about PC / MAC.
But when talking about the subject, it seems MAC users has more problems with PL. If your MAC hardware is not up to the task you must stand ready for a mayor investment if you have mac… A PC owner can often just upgrade to a better GPU or whatever.
That’s it !!! More evolutive means a little (very little) more involvement from the user but much more flexibility.
These are debates that have been going on for decades because of Mac advertising campaigns designed to suggest to their users that their system is unique (I remember the “Think Different” campaign, for example, when Macs ditched Motorola to become … PC hardware with intel cpu !!!)
I found that by going in to my desired preset and removing every tool except the ones I wanted to change (DeNoise and Lens Sharpness Optimization) I could then save that preset and it functions as a Partial Preset.
However…
Even with the tool deselected in the preset, activating the preset causes the Advanced settings of the Contrast slider to reset to zero.
Everything else seems to work as expected (I edit a photo, I apply my Partial Preset, and the other changes persist).
I just lose those contrast settings. Most annoying.
Yes the fields are definitely deactivated in the preset, but a change is still being made when the preset is applied.
I’ll have a play and see what I can figure out.
Edit: Well rather than muck around with PhotoLab itself, I dived in to Notepad and removed the references to the contrast settings from the Partial Preset.
“AI safety researcher Roman Yampolskiy says he believes that once artificial intelligence starts building the next generation of AI itself, humanity will no longer be in charge, and says the threat is not that it will hate us, but that it simply will not care.”
The headline “PL 10 is useless for professionals” is a silly over-generalisation. Not all “professionals” have the same requirements or the same workflows. Or, indeed, use the same systems. We do not all need to export large batches, for example, or work on multiple images at the same time. I am running it on perfectly standard M4 Mac mini with 32GB of RAM. Image files are maintained on an external Thunderbolt SSD. It even runs, though certainly noticeably slower, on my old M2 MacBook that I use when travelling. On the M4 the delay while it scans for AI mask subjects is typically under 4 seconds. Before AI masking was possible, it would have taken me many times that to get a similar result using the old tools, I do not feel inclined to complain. I tend to edit one image at a time, export one image at a time, and I do not have loads of other stuff running in the background. Perfectly happy with it. I am sure some things will be refined in future releases.
Agree, same here besides images on an external HDD. Still fast enough. AI scan also just around 4 seconds. And I have preview high quality enabled. DeepPrime rendering I only activate, when I have high ISO and want to take a sneak peak. Of course, that really slows down performance otherwise
Still no PL 10 on my new Mac Studio M5. I’ll have to check later to see if PL10 runs even “faster” on it.
I actually got this computer to speed up video editing with FCPX and Compressor. But I still need to do a final test. Apple isn’t exactly a role model when it comes to optimizing its own software for its hardware.
P.S. Still Tahoe on this machine. To many stuff which will not work with Golden Gate. No idea if MacOS 27 maybe speeds things up with PL 10
