Errors with AI masks on new MacBook

DxO PhotoLab macOS topics
1

Hi,
I’ve purchased a new MacBook Air with M5 CPU and 16 GB RAM. Unfortunately, AI masks act very strangely on this machine. Editing the same photos on a MacStudio with M1 Max works perfectly.
BTW I’m using PL 9.6.

Problems on the Air:

  • Sometimes, creating a mask like “background” results in an error message saying (roughly translated from German) “The AI mask couldn’t find anything matching your selection”.
  • Sometimes, the same mask could be created. But after restarting PL, the mask is displayed with an exclamation mark icon and a tooltip saying “The mask could not be chosen”.
  • Changing the AI engine from “Apple Neural Engine” to “Graphics Processor M5” and restarting PL makes the mask work again: No exclamation mark and the adjustments applied to the mask are visible
  • Not changing anything at all and restarting PL one more time: Exclamation mark is back, adjustments invisible again
  • Switching to CPU and restarting: Mask is back again

So this is pretty odd and as I said: No problems on the MacStudio.
Is this a known bug? Is there a known workaround?

Thanks!

2

I cannot comment on this specific issue but you will see from my separate post that I have had processing problems on a top end Macbook pro with the dxo9 trial which I did. not have with previous versions. It feels like dxo9 does not play particularly well in a Mac environment and I have decided to wait until this is ironed out before considering buying v9.

3

I’m about to buy a MacBook with 16GB M5 configuration. Appart from this issue, the rest is working smoothly or do you find DxO Photolab slow?

4

Apart from this issue, PL is working very well. I haven’t done any benchmarks, but compared to the M1 Max with 32 GB, it seems to be running noticeably smoother on the Air.

There are still actions where the performance becomes ridiculous slow, like rotating an image which contains an AI mask. But that’s the same on the M1 Max and these situations are rare.

1 Like
5

The problem occurs when the predefined AI mask doesn’t work for some reason (image content). Previously, the program would often crash – now simply skip that AI ​​mask.

6

I’d go for 32GB if you can. I went for 48 and it runs very nicely compared to my M2 with 16GB. Big step up.

1 Like
7

The problem occurs when the predefined AI mask doesn’t work for some reason (image content). Previously, the program would often crash – now simply skip that AI ​​mask.

Strange thing is: Same photo, different machine → mask is created.
Or even on the same machine: Switching to a different AI accelerator (one of neural engine, GPU, CPU), restarting PL → mask is created. But the same mask on the same photo isn’t working after a restart without switching the AI accelerator first.