Fix #32198: UV unwrap of half a UV sphere would not give a perfect sphere as a

result. This wasn't really guaranteed anyway, because of numerical precision and
possible asymmetry in the triangulation, but we can do a bit better.

Now we bias the choice of how to split a quad into two triangles slightly
towards one of two possibilities, so that in case they are equal, floating point
errors do not decide the direction and symmetry is preserved.

1 files changed, 5 insertions, 1 deletions

diff --git a/source/blender/editors/uvedit/uvedit_parametrizer.c b/source/blender/editors/uvedit/uvedit_parametrizer.c
index 2161d4989db..72c38be5398 100644
--- a/source/blender/editors/uvedit/uvedit_parametrizer.c
+++ b/source/blender/editors/uvedit/uvedit_parametrizer.c
@@ -1135,7 +1135,11 @@ static PFace *p_face_add_fill(PChart *chart, PVert *v1, PVert *v2, PVert *v3)
 
 static PBool p_quad_split_direction(PHandle *handle, float **co, PHashKey *vkeys)
 {
-	float fac = len_v3v3(co[0], co[2]) - len_v3v3(co[1], co[3]);
+	/* slight bias to prefer one edge over the other in case they are equal, so
+	 * that in symmetric models we choose the same split direction instead of
+	 * depending on floating point errors to decide */
+	float bias = 1.0f + 1e-6f;
+	float fac = len_v3v3(co[0], co[2])*bias - len_v3v3(co[1], co[3]);
 	PBool dir = (fac <= 0.0f);
 
 	/* the face exists check is there because of a special case: when