The future of selections and masks in Krita

[Boudewijn Rempt]

On Tue, 1 Aug 2006, Thomas Zander wrote:

> On Tuesday 1 August 2006 00:31, Boudewijn Rempt wrote:
> > The thing is: a global
> > selection is a weird thing. It is created (when using fancy tools such
> > as select similar or foreground extraction) based on the pixels of the
> > current layer; it acts on the current layer; which associates it to the
> > user (my dad in this case, not one of the persona's, he's reasonably
> > proficient, but makes this mistake every time) with a particular set of
> > pixels. And then they change layers, but the selection now suddenly
> > applies to this other set of pixels.
> 
> I stated this on IRC, but I'll state it here again for your benefit.
> If a persona is not included in the application personas, he is not a 
> target user.  Your father is probably not someone who will use a painting 
> app professionally and quite a lot of the time, right?

No, he just a hobbyist photographer. The closest persona to him is Arthur. 
But he is a real user: which means that, anecdotal as all user observation is,
it's a real, and not an imaginary data point.

> Then he is not a target user and him not understanding it is irrelevant to 
> the design of the application.

I disagree here. I may have misunderstood "Arthur the amateur landscape
photographer", but I did not get the impression he was a profession Photoshop
user.

> Maybe krita devs want to add a less proficient user to the list of 
> personas?

Yes, I think we need to have a persona who isn't pickled in Photoshop. It
now appears that our persona's are just three kinds of photoshop users; understandable
since all graphics artists are photoshop users nowadays, but not what I am
interested in. Because if that's where Krita is going, we may as well decide on
a particular version of Photoshop and clone it literally.

> Anyway; The example you bring up where things go wrong for your dad seems 
> to be due to the way he creates a selection and he remembers where the 
> selection came from and still uses it like that.
> Moreover; your example shows that the real problem he has has nothing to 
> do with selections (since he is not manipulating the selection, afterall) 
> it has to do with the fact that changing layers will only alter that 
> layer. And him not getting enough feedback to state that fact.

The issue is: something created based on, in association with one set 
of data suddenly applies to another set of data. That's confusing. That's
harmful, because if you're the kind of artist who doesn't plan in advance,
but works on all levels almost simultaneously, you have to remember to turn
the current selection into a mask before leaving the layer, and turning the
mask into a selection after selecting the layer.

I think that the original use-cases where completely centered on Arthur and
forgot all about Claire.

Boudewijn