On Mon, Mar 17, 2008 at 10:45 AM, Bill Skaggs <[EMAIL PROTECTED]> wrote:
> David Gowers wrote:
>  >  Myself, I would use it for nearly every non-symmetrical brush (and
>  >  wish for the ability to v- or h-flip it, too), particularly when I'd
>  >  just copied something and was using the clipboard brush.
>  > Every brush that is remotely 'directional' needs this
>  >  (or, possibly the option to rotate the brush to match the movement 
> direction -- that would work fairly well much of the time.)
>  If the major usage case is to rotate to match the direction of
>  movement, it would make more sense to support that directly,
>  since it would be a p.i.t.a. for users to have to go into the options
>  to set a direction for each stroke.

I was envisaging it being settable via the keyboard. You can currently
set VBR angle by keyboard shortcuts.

In any case, you seem to have missed that the directional rotation
requires a 'normative' rotation value -- since many brushes, while
asymmetrical, have no obvious 'direction', we might want to allow a
default rotation to be set per brush..
It *would* be necessary to allow normative rotation to be modified, in
order for directional rotation to consistently remain sensible to the

>  This shouldn't be too hard in principle:  it is already supported
>  as a mode for "image pipe" brushes.  In practice it hasn't worked
>  all that well, because it's not so easy to cleanly estimate the
>  direction of motion -- especially at turning points or the start
>  of a stroke -- so it has been hard to avoid getting anomalies
>  every so often.  The new motion-smoothing code that Alexia
I think that we need some trickery to happen -- for example, the right
thing for the start of a stroke is probably to match the angle for the
first event to the angle of the following event, so we would need to
draw the first dab, and then redraw it if the user continues the

>  recently contributed might actually make this work better --
>  this should be investigated.
Because it's inertia-based, it does..  as long as there is *some*
smoothing happening, angle is much more coherent.
(note: I'm currently using 2.4.2, rather than SVN trunk, sadly.
Gimp-developer mailing list

Reply via email to