Patch review for optimization in paintop

[Cyrille Berger]

> > Can anyone have a look at this patch ? It optimized greatly paintop
>
> I didn't actually test it, but it looks ok. Just to make sure: if the
> profile is different, the m_dab will be recreated as well, right?
hum good question, doesn't the colorspace changes as well ?

> > Now, computeDab is still the expensive function for paintop (with 40% of
> > its cost coming from creating/destroying iterators :'( ).
>
> I think it'd be pretty easy to rewrite computeDab to use a rect iterator
> instead of maskHeight times a HLine one.
of course using KisHLineIterator::nextRow reduces the cost greatly :) great, 
with this new version of the patch, it becomes really hard to lose the mouse 
cursor. Now 70% of the time is spend in the ++ operations, but that's too 
late for trying optimizations on it in 1.6.

> But iirc, what really slows down 
> stuff is the rescaling (at >100%) of the mask when pushing your tablet to
> 100% pressure (which will make the dab bigger with the default options).
> Especially noticeable with big brush sizes iirc.
hum I don't have my tablet with me, I will try it tonight.

-- 
--- Cyrille Berger ---
-------------- next part --------------
A non-text attachment was scrubbed...
Name: optimization.paintop.diff
Type: text/x-diff
Size: 1966 bytes
Desc: not available
Url : http://mail.kde.org/pipermail/kimageshop/attachments/20060927/27e69cb6/attachment.bin