I use PL3 to edit the NEF-Files, sometimes also JPG, from my Nikon Z7. I work on an 2017 5K iMac 27" (24 GB RAM and Radeon Pro 580 8GB) with all the latest updates for OS X and for PL3. In general the performance of PL3 is mediocre, compared to other tools like Affinity Photo for instance.
When editing a lot it gets really, I mean really, bad. When working with local corrections (repair) PL3 “thinks” several seconds, before it is showing up with something.
I think PL3 is not optimized to handle 40 to 60 or more MPixel files very well. This performance flaw is a pity, because despite this I really like PL3. Are there plans to improve performance substantially?
Hmmm, what is? My Sony files are fine, but I prefer my ORF’s. It is about the Size of the file and the computational power of the PC. Since I discovered i have no use for 45 Mpx files, I am much happier.
The computational power is by far enough. Affinity Photo handles these files much better, concerning speed.
The solution is not to state: who needs 45 MP?
It would be a fatal decision by DxO to go that way. 45 MP are real, 60 MP and more are coming. People will choose the tools which will handle this appropriately.
To make it clear: I’m not talking about the final processing of the image. I‘m talking about the time PL3 needs to display changes and adjustments while working on an image. It’s the delay between action an reaction that hurts.
One thing that can speed up PhotoLab is to work on folders that contain tens instead of hundreds or thousands of image files. Although this is a workaround, it makes life easier.
I also find performance to be PL3’s Achilles’ Heel. I have a more lowly specced Mac, but I also only have 24 megapixel images and it struggles at times, with the fans often spinning up audibly and I have to wait for adjustments to be rendered. This is in contrast to any other graphics programs including Affinity Photo and Lightroom.
Which parts of the program seem slow to you? I find PL to be much more responsive than Lightroom. I have an 8 core AMD Ryzen 7 3700X CPU and a 16Gb of RAM and I have never felt that PL was slow. With Prime noise reduction enabled my PC processes a JPG image from my D750 RAW files in about 10 seconds per file.
To make it clear, I didn’t propose a solution to you nor did I say who needs 45 MPx. It was you who spoke about PL being slow for large MPx files. I just made a general comment that editing and processing my SONY 42 Mpx files are handled without much effort, but it’s all personal. I have only been on the forum for a few months but have not read many other complaints about speed of editing.
To confirm that I am not crazy, i just reprocessed a 42 Mpx Sony ARW file
I performed a series of edits including several local adjustments. I didn’t do any pixel level edits like removing something, as I would prefer to do that in Affinity Photo. I think PL out performed LR for a similar set of edits. Even with final rendering.
I’m getting the same as Georg.
A small week to a slider has the ‘correction preview’ pinwheel spinning for far too long (Up to 10s)
Mac pro Catalina 12G RAM DXO 3.2.1 build 51 - on ‘small’ 12MP files
i agree withe the original poster: repair can take many seconds before showing the result. my files are 10-20mb, not huge by any means, and i am on a good graphics card, with 32GB, M2 drive and AMD 3600. Top of the line one might say. Just eft LR and PS and finding this very sluggish to the point of unusable.
I hope there are plans to improve performance. It needs serious improvement in this aspect of PL3 more than any other.
I live with the frustrations (repair tool, for example, is useful, but a performance dead-weight)
PL3 should be snappy, responsive, speedily supportive of experimentation and it’s none of these.
It produces brilliant output and the tools are fantastic to work with - albeit slowly.
So, maybe in your next test you try the repair tool with your 45 MP Sony files? That is what I explicitly mentioned in my initial post. And please try at least 15 or 20 repair spots as you would need it heavily for portraits to clean up the skin or hair. To state that you do it in Affinity does not solve the performance problem of PL3. I can do the whole RAW processing in Affinity, but this is not what I want to do. I prefer to work on my images with one tool.
And again: I do not discuss the speed of the final rendering, because it doesn’t really matter. I can do other things while it is rendering.
BTW: there are complains about PL’s performance on the net, especially about large files with plenty of MP. That is what brought me to my assumption.
In my experience, the preview perfomance issue is more due to the display resolution than the input file resolution.
As for me, going from a 25601440px to a 38402160px monitor impacted performances way more than going from a Sony a7 III (24 Mpx) to a sony a7R IV (61Mpx)!
That said, I also have a performance issue due to my huge library (~50k photos (15years of photos!) with no subfolders). It works like a charm with other software, but DxO Photo Lab struggles…
Many thanks for your contribution, you made a very valuable point! I already had the same assumption, but have no comparison since I started PL3 only right on my 5K iMac. My display resolution is 5120 x 2880.
The performance of the library is another thing…
You can try it by varying your windows size, you’ll see how preview time seems to increase with the number of pixels the preview occupies.
And this can be expoential!
For example, if you go from Full HD (1920 x 1080px) to Ultra HD (3840 x 2160px), the number of pixels doesn’t double, it quadruples. And in adition to that, unless you use magnfication for the interface, it proportionnaly occupy a smaller portion of the screen. So the preview area may count 6 or 7 times more pixels!
That said, I read that DxO Photo Lab was using GPU accelerated preview above a certain level of zooming. 75% if I remember correctly.
If that is correct (to be checked with DxO staff), you may also have some threshold effects. For example a “good” one if you go to a higher resolution display and that it can fit the whole image with 75% or more zoom. Or a “bad” one if you buy a higher resolution camera whose files will required less zoom to fit.
But this has to be confirmed…
this is true.
But image size has also an impact on preview performance as DPL has much more pixels to calculate.
At some point, it has to demosaic the full resolution file, so yes.
But when it’s done, my guess is that it generates a preview image at a lower resolution that depends on your zoom level. (Note : I’m not altking about that uggly very low and blurry preview that they use sometimes, just something matching the display windows resolution.) Thenafter, when you move the silders, only alter this lower resolution file.
In such case, the preview speed would only be affected by the input file resolution if you change the zoom level.
That could be also a more complex approach that deal with different versions (resolutions) of the images depending on what you play with.
But this is pure assumption. Maybe the DxO staff can confirm/refute?
That’s a very interesting point about the window size, but there’s no way I want to make it smaller. I want to see as much of the detail as I can when making adjustments. That’s why I have PhotoLab, because of the detail it can pull from my RAW files.
But that may explain why on a 5K display performance is hurt so badly.
As for me, DxO Photo Lab preview performances is an annoying issue. I have a very powerfull rig (3700X, 32GB, RTX 2080, NVMe SSD…with CPU, Inifnity Fabric and RAM pushed to the limit)…and it’s desperately slow on my 4K display.
I don’t really care about export speed, for export I want max. quality, even if it’s long.
So, that’s exactly my point. Export doesn’t matter in terms of speed. It’s the workflow that hurts with PL3 if you use it for what it is meant for and not for just some color adjustments. And that on a system like yours! Nobody can convince me that it is not adequately powered for PL3. That’s ridiculous. Unfortunately until now, no PL3 staff has responded to my initial question:
Are there any plans by DxO to further and substantially improve the performance, speed, responsiveness of PL or is it just “it”?