alexandre.prokoud...@gmail.com (2011-03-20 at 2227.33 +0300):
> On 3/20/11, Bill Skaggs wrote:
> > I think it would be pretty difficult to figure out the algorithm by looking
> > at the Gimp source code.  The algorithm that Gimp uses is based on
> > papers by Todor Georgiev, and you can find a description of the algorithm
> > in a paper he wrote called "Photoshop Healing Brush: a Tool for
> > Seamless Cloning" -- you can access a PDF version at
> >
> > http://citeseerx.ist.psu.edu/viewdoc/download?doi=
> >
> > Note that Gimp only uses the Laplacian method that he describes, not
> > the more sophisticated fourth-order version.
> GIMP's implementation is prone to errors which is especially visible
> when you heal near a visible border. I'm not entirely sure it's a good
> idea to learn from it.

Was not that due to running the algorithm per brush stamp instead of
for the full paint stroke? IIRC when using a big enough brush tip and
just clicking once to cover the error, it worked. And PS applied/s the
solver once you end the stroke. In other words, the system is correct
internally, but the final touch is missing.

Gimp-developer mailing list

Reply via email to