VivaRado
commented
4 years ago Resolved:
- Multi-Instance (More than two instances can be imported to variomatic comparable format).
- Debug Plot windows now are appropriately sized to the actual size of coordinates.
Hello and welcome to Variomatic. Here are the bugs )
the contour sorting fails for non line travel distance sortable, as it tries to calculate the travel distance, of each contour and sort. For example the inner part of the letter "O" from the outer, is a smaller travel distance in both instances. Where this is not true for glyphs like "=". A approximate position or centroid or barycentric of the contour points, in the glyphs will probably have to be taken into account.
if end start is in the uncertain lines, final green matching line doesn't visually close.
if triangle matching and inline check fails for uncertain lines... well there is no matching.
at the moment there is no final line in a usable list, it is though plotted as green for certain and lime for uncertain basic_cleanup and match_uncertain_tris.
this script doesn't do curved bezier contour analysis, it does straight point position comparison across simplification levels. In further iterations we will improve to do bezier contour comparison.
as it doesn't do curved bezier contour analysis, there is a very small chance the certainty matching gives incorrect matching. But if this happens, there is a chance there are move issues than just that and those issues will be directly visible.
adding curved bezier contour analysis will result in 100% matching and cleanup. In respectfully similar contours. Do not expect this to work with two random letters, there is no morphing.
the script solves only two instances against each other, there is no multiple instance support at this moment.this code needs a strong cleanup.networkx should be removed.
Thank you!