Yep, binaries are on the releases page: https://github.com/TannerRogalsky/lua-poly2tri/releasesAntonioModer wrote:Please, if not too much trouble, could you download here(on this forum topic) the binary library files to link with Lua for Windows(32 and 64 bit versions).
It would save a lot of time for me in future, in develop my Navmesh library. Because I'm a noob in C and C++.
Cool, thanks. I was do a simple test, it work.Tanner wrote: Yep, binaries are on the releases page: https://github.com/TannerRogalsky/lua-poly2tri/releases
Code: Select all
function love.draw()
local polys = { {100,100}, {200,200}, {300,300}, {400, 100}, {300, 200} }
local triangles = math.DelauneyTriangulate(polys)
love.graphics.print(strformat("triangles=%s",#triangles),5,0)
love.graphics.print(strformat("vertices=%s",#polys),5,15)
if (triangles) then
for i,tri in ipairs(triangles) do
love.graphics.polygon( "line", polys[ tri[1] ][1],polys[ tri[1] ][2], polys[ tri[2] ][1],polys[ tri[2] ][2], polys[ tri[3] ][1],polys[ tri[3] ][2])
end
end
end
Unfortunately, if you don't satisfy the constraints that poly2tri has (no same points, no self-intersection, no holes touching other holes or the countour) then it's method of letting you know that is to segfault (at least on OSX). I'm working on a separate library to help with sanitizing data before it's passed to poly2tri and I will post it here when it's done.alundaio wrote:Some reason I crash *rarely* using those poly2tri dlls, not sure why.
It does. At least the version that I've wrapped does.ivan wrote:Poly2Tri is fast but I don't believe it supports constrained Delaunay triangulation (at least it didn't a while ago, not sure about the more recent versions).
I allude to that in the readme for my wrapper. The approach I'm using for sanitizing the data is to union the holes with NonZero winding along with some other robustness tweaks. Clipper already has a nice FFI-powered wrapper: https://luapower.com/clipperivan wrote:"self-intersection, no holes touching other holes or the countour" - polygon union/difference/offsetting are handled by the "Polygon Clipper" library or "GPC".
love.math.triangulate didn't solve my needs as it was creating overlapping edges. Poly2Tri had the results I was looking for, but I can't control the order the vertices I input to make sure I satisfy the constraints since I'm making a procedurally generated graph. The delaunay algorithm I'm using now works perfectly, though. I just needed nice non-overlapping edges so that I can calculate the adjacent neighbors for each 'room cell' in my picture. But thanks.Tanner wrote:Unfortunately, if you don't satisfy the constraints that poly2tri has (no same points, no self-intersection, no holes touching other holes or the countour) then it's method of letting you know that is to segfault (at least on OSX). I'm working on a separate library to help with sanitizing data before it's passed to poly2tri and I will post it here when it's done.alundaio wrote:Some reason I crash *rarely* using those poly2tri dlls, not sure why.
You might just be looking for https://love2d.org/wiki/love.math.triangulate, though. It's fast and good for simple cases. I believe it uses Kong's algorithm. http://www.sunshine2k.de/coding/java/Po ... /Kong.html
