The future of selections and masks in Krita

[Bart Coppens]

On Tuesday 01 August 2006 15:24, Boudewijn Rempt wrote:
> True. Insistence on use-cases is something that seems to have happened in
> the last week or so; before that, we did without.
Yeah I also don't know why they suddenly came up. It's not a bad thing per se, 
because it can add a better focus as to what kind of features we can add, but 
as I said on irc as well, insistence that features only get added when they 
are in line with the personas (which I guess means use case here?) (like 
currently suggested on the personas page) is too harsh for me.

> I must admit self-interest here. I may be stupid (but rule one of
> interaction design is 'never make the user feel stupid'), but I have never,
> even with the manual in hand, been able to use Photoshop's channels and
> masks. I still cannot figure out what masks can do. Although my Photoshop
> manual seems to suggest it's more than just making parts of a layer
> invisible.
I must admit self-interest here too. Before last week, I didn't know anything 
about masks either. I didn't know a good use for them either. The reason I 
actually became interested in them in the first place, was because there was 
a user in the irc channel asking if it would be possible to get them. And 
then he showed me a nice pic of actual nice usage:
http://bluesceada.funpic.de/examples/DSC_64xx_landscape-keltern-1_small.xcf.bz2

It's a very nice, non-destructive way to merge images. We can already do this 
with quite convoluted layer arrangements (like I once explained on this list 
a while ago). But it's difficult, and you actually need to know the trick, 
but possible.
Now, when I see very cool pics made with the Gimp, that can't be made just as 
easy with Krita, I get a bit jealous, especially if the feature looks quite 
easy to implement (see the self-interest kicking in? I want to be able to 
make just as cool looking pics with Krita, and show it to other people that 
Krita can do that easily too for their pics!).

Now, as far as I understand it, it is indeed just a local-working adjustment 
layer, mapping the mask to alpha values of the adjusted layer. But the 
problem is that that requires changes in the implementation of the adjustment 
layer, but rendering-wise, and conceptual (be able to put them as sub-layer 
to other layers).
Personally, if I would be able to just chose, I think I'd chose this approach 
over having a mask, because I think it's conceptually a nicer usage that we 
can offer with the adjustment layers overall, then. The problem is, that this 
would require much more changes than a simple mask (the masking code is quasi 
trivial, both editing-wise as rendering-wise).
My personal choice of action for _masks_ (not the rest of this overheated 
discussion), would be to let the current mask thingie stay. Then, if we 
decide to do the adjustment sublayer thing in Krita x.y (and actually get it 
implemented), we can just backwards-compatibly map the old mask to the new 
structure of alpha-changing, local sublayers. But some people disagree on 
that, I fear.

> I'm feeling better already
Great to hear!

Bart