Masking doesn’t have to be perfect as long as you are not cutting out a selection for compositing or raising Exposure by 3 stops etc. For the kind of adjustments that you tend to make, small adjustments in exposure, contrast, clarity etc missing a few feathers or part of a claw are usually not significant, I also feel that cleaning up an AI mask is far quicker than a manual masking which still has to include clean up of your manual mask.
Have you tried sessions in C1?
I have a work flow, FastRawViewer, Iridient S Transformer, Affinity Photo and Mirage Print made up of non subscription products that don’t have new versions coming out every year. Obviously I’m not keen on the subscription or the pseudo subscription business models. (I do pay for Topaz Photo AI every year but they are innovative and I get weekly upgrades.) I have PL as its easier if I have a lot of “good” quality images to process. I really see PL at the level of a non subscription product, like Affinity Photo. Can’t see them competing with Capture One. And as has been pointed out by others, Lightroom/Photoshop is cheaper.
Actually, there are a lot of work where a layered approach is far from optimal. The way you can select areas in PL with Control Lines and Control Points is faster and superior in many cases. Think of sky behind a tree as one such case… So, I for one don’t want a total rework of PL, but could do with a way to do finer selections of objects, especially those with hair or fur… 
… and also ‘drawing’ straight lines to mask & erase.
e.g. click > move pointer > Shift click = straight line between. Definitely.
I have updated the list of expected AI-based features.
Perhaps one of them will be implemented in PhotoLab 8?
Even if C1 only matches LR that would be excellent performance. I wish DXO would look and learn from C1 on how to design a layers UI. They have most of the elements in place but can’t seem to integrate it as well, mask feathering, refinement, a filled mask, mask editing, showing mask in real time etc.
A sky behind a tree is no match for AI either in Lightroom or CO. It’s a poor exampel. CO is capabel of solving much more complicated case than that.
There are examples out there where CO even seem to be better at refining and cleaning up complicated masks. Tools LR seem to lack or beeing a bit inferior.
I’m not quite sure what you mean, but here is an image, with the B&W mask showing, of a Control Line over the sky with trees in front…
PL6
Before…
During…
After…
Sure, this is a very good example what easily could be fixed with a lumunosity mask but say you wanted to mask or demask one particular of the trees that have the same luminosity as the surrounding trees. In CO you would only need to press Alt and click to deselect that tree and to select a certain tree could be made by a click or drawing a rectangle over that tree. When you do the last you will see how the mask adapts to the subject just by hoovering over it. It is rather flexible and intuitive
I compared with traditional layered approach without AI, where this kind of selection takes a lot of time, if possible, at all.
If anyone has time and want to test, in the sample I have included, how well do AI select sky, person, house and tree here?
Here you have a video that targets AI masking of complicated areas in both Lightroom and Capture One. It about shows where we stand now and the small differences between Lightroom and CO that obviously are there. The most obvious is that it is dedicated function to target a person. In CO you can either choose between “Subject” and “Background” to begin with.
CO seems to have an ace LR seems to lack and that is “Refine Mask” That tool solves for example to sort the problem refining hair on people that might be a problem with the LR Tools. Besides that CO also have Magic Erase and AI Erase that helps cleaning a messy precission jobs.
See for yourselves. In Thorsteins image above I should start with Subject when masking the man for example and then use the AI Erase if it will be necessary or just hoover and click when the mask looks best and take it from there. I doubt luminousity tools would fix this more easily sins there are light parties both on the man’s body and the tree.
Here is a video that better shows the different options there are now in AI masking in CO.
Nice videos!
to complete the overview, take a look at the last video on the same channel:
@noir.fonce
There is a lot to like about DxO:s Control Points too and all that have used them for long can advocate that. They can be very effective and useful generally. One problem though i that the precision sometimes just isn´t good enough. It has advantage in the cases where might WANT the edges to bleed over the “edges” into the background smoothly but that is not always the case and I have long before AI looked to CO in situations like that.
Thanks for the video link it was a good complement to the other videos.
Hej Torstein!
Here a few examples of masking in Capture One 16.3
I have started with AI mask of “Subject” and as you can see it both missed the lover part of his legs
After a couple of more clicks with the AI Brusch and the AI Eraser
It´s not an easy task to get a “perfect mask” of the sky here when it it so cluttered with “needle leaf balls”, maybe this would have been better of with some luma masking and work with the collateral mask of the man that probably had been the case then.
It is kind of interesting to see that it was not the new AI Brusch tool that gave the best mask of the sky. Instead it was the older tools called Magic Brush. The way that tool is working is that you click the few various areas with special color or tone and then the mask is growing as you click. I think it took four or five diffferent clicks to get this result, which I think is pretty good. It solved the problem pretty well. I think Magic Brush came in a feature release with version 22 a couple of years ago.
So, it is absolutely possible to make a mask of the sky even if the background is very cluttered with Capture One. The Magic Brush seems to use both Luma and Chroma to manage to isolate the sky from the person and the tree and is a good complement to AI masking.
Absolutely but version 7 is pretty different than version 6 since they now have changed the interface substantially. That was necessary but unfortunatelly not even the new UI is all that good because the devision of the UI into a global and a local part is inproductive. It is far better to make layers of it all like in Capture One and make all adjustment tools work even in Local Adjustment. Then you get a consistent UI, The one in Photolab is not. Not everything in Photolab is so confusing, poorly implemented and ineffective as the Color Wheel implementation but it is far from the only exampel.
Have you tried to use Capture One with “Sessions” instead?
I only use CO with Sessions. Originally there were just sessions often used by studio photographers.
You have to rig sessions - session by session. It is not as straight forward as Photolab but it’s possible.
CO AI masking is brilliant for a first interation. PL is definitely lagging behind the competition in many ways. ON1 and Neo have workable AI masking that save significant time. Rather than give us new masking capabilities, DXO releases Luminosity masking and diffusion control in other products but not PL.
I guess DXO:s focus on color management issus like LUT and calibration, functions that might get used by a very limited user group has drawn resources from AI masking and layers. The year before it was the PictureLibrary.
