AI masking and RAM

I have been testing the Photolab 9 with somewhat discouraging results. In my case I am quite sure that the main problem is not in the software, but in my computer specs (1,80 gigahertz Intel Core i5-8265U and Intel(R) UHD Graphics 620, win 11, 16GB RAM). However, the annoying thing is that some things actually ALMOST work, as applying AI masks by selcting an area. I feel that it wouldn’t have to be very much faster in order to be “bearable”. So I am looking at some solution that would save me from upgrading to a better computer for the moment. The only possible thing would seem to be to double the RAM to 32GB. Does anyone have any thoughts about whether this would make any noticeable difference?

I realize that I anyway would have problems with the advanced DeepPrimes, but that I can live with. I should perhaps add that so far, I have been using Photolab 7, which works well enough on my system.

I cannot say for certain until I make the move to PL9.

However, why not check the memory usage in task manager (right click on taskbar). That would give some idea of whether or not PL is constrained by your available memory.

Best wishes
Malcolm

1 Like

I already did that, but wasn’t sure how to interpret it. 75% when doing AI masking (when writing this RAM use is around 30%).

Please check the requirements. While RAM can help, your GPU might be the limiting factor.

1 Like

You only have Intel’s integrated GPU which is not good enough for PL. I would suggest you get a graphics card as a start as it will really help.

1 Like

If you have been happy so far, why not stick with it.

As much as that’s true, and good advice, I would say I have a graphics card (1080Ti) with 11gb ram and preset AI masks like Subject or Sky still don’t work.

Object selection (by clicking on what I want to select) , however, does work.

I’d be fine with e.g. Subject selection taking longer on my not-cutting-edge system, but I it shouldn’t be failing completely, almost straight away.

My point is, exactly, that you seem to get some things to work even if your system falls short of the requirements (which , of course, I have checked). So my original question does not concern how you get everything to work, but whether it is possibility to increase the pace of some aspect of it you happen to like (by adding RAM, since this seems to be the only thing you can do on a laptop nowadays). But perhaps no one has tried.

@KeithRJ I believe the processor is for a laptop so adding a GPU is not going to be possible, i.e. a replacement laptop (new or second-hand) with a GPU with suitable power and associated VRAM is the only option.

@Bhard Firstly are you using what GPU power you have by configuring PhotoLab to use the IGPU instead of CPU. In tests I did some time ago it was better to use IGPU than CPU but way worse than a dedicated GPU.

I have a problem simulating your configuration although my i7-4790K might give me an insight. It is a long time since I ever wondered about how much memory I should configure for the IGPU and that is probably the only reason I would consider adding additional RAM to you laptop(?), i.e. if some of that RAM can be allocated to the IGPU (if you are even using it).

I understand why you want to extend the life of your machine further, adding more memory should be reasonably cheap, providing your laptop has a nice convenient flap to allow access to the memory cards and you don’t need to throw the 16GB away in order to upgrade to 32GB!?

I don’t think that is going to happen just by increasing RAM, it needs a bit of GPU grunt and that GPU needs enough VRAM as well.

I tried using the i7-4790K with the PL9 GPU selection to be the the 4600 IGPU but I got this when I applied an AI selection

Switching to CPU only and just moving from an image with an AI setting to one which has none took 35 seconds and the same thing happens on an image with AI set every time I move the cursor!

Using the autobrush feature of PL9 with just CPU selected works without undue delays and appears to be way better than automask in PL7 so there is something to be said for PL9 but adding the price of a new machine makes that an expensive upgrade.

The tulips looked way better the way they were but I was just testing and needed to show the difference.

@Bhard Sorry but I don’t see how any more memory is going to help, even when the AI issues are corrected it appears to be too much for the processor to handle but might (or might not) allow the IGPU to help.

@Fineus the temporary “fix” is to go back to old NVidia drivers, unless that is going to interfere with other software. That this is necessary is reprehensible but …

My “problem” image with Sky AI doesn’t work with even very old drivers but it is a 2060(6GB) and I believe memory is the problem there. Plus the machine I used here was never intended to have PL9 I just got an additional trial copy to see if 6GB VRAM would work with the current workaround and it didn’t in my case.

1 Like

This is what most users of NVIDIA GPUs are experiencing - me included!

1 Like