Hello Alexei,
> Can you describe discontinuities with the same sign if I overlook them?
So, I found there are two types of discontinuities due to overlapping
contours:
A) Discontinuity in sign. (i.e. there is a sudden jump from negative to
positive)
B) Discontinuity in distances. (i.e. the points that should be at
infinity/spread have smaller distances)
Here is an example of a SDF generated from glyph with overlaps:
https://i.imgur.com/3pqHZWd.png
In that image the right side is correct (generated from bitmap), and the
left side has both A and B
types of discontinuities.
What you are saying about detecting sudden jump from large negative to large
positive can work for discontinuities in sign. I tried that and here is an
example: https://i.imgur.com/unRB23Z.png
(ignore the weird lines in the image, that can be fixed). But even after
fixing the
sign there are pixels inside the outline which have shortest distance to
the curves
inside the outline. (basically the pixels around the overlap points to the
overlapping
curve, which causes discontinuous distances). Moreover they interpolate
smoothly
so there is no way to find the actual distance which will be to some curve
at the border
of the outline instead of inside the outline.
Also, I looked at a library (msdfgen) which handles overlapping contours
without
removing the overlaps. It works well, but it doesn't work for
self-intersecting contours,
that is why I'm not sure whether it's worth implementing.
Finally, I can find and fix signs such that all points inside a shape are
positive and
all points outside are negative. But I don't know how I can fix the
incorrect distances
caused by the overlaps. Here is one final example:
https://i.imgur.com/jiaqiBS.png
In this the point `P' has a positive sign and it points to a line inside
the shape, which
is not correct, instead it should point to the corner. Moreover there is no
corner in the
shape because they are two separate contours, so I think removing the
overlaps is
the best way to handle this.
Thanks,
Anuj
