Splitting quad into triangles and merging triangles into quad should

work correct with sculpting data now.

Joining two triangles could give incorrect sculpting result for
special topologies, but it's that case that can't be nicely handled
with our layers architecture.

1 files changed, 44 insertions, 0 deletions

diff --git a/source/blender/blenkernel/intern/multires.c b/source/blender/blenkernel/intern/multires.c
index 68ee0d0d0cd..84350127968 100644
--- a/source/blender/blenkernel/intern/multires.c
+++ b/source/blender/blenkernel/intern/multires.c
@@ -1981,3 +1981,47 @@ void mdisp_flip_disp(int S, int corners, float axis_x[2], float axis_y[2], float
 		disp[1] = 0;
 	}
 }
+
+/* Join two triangular displacements into one quad
+	 Corners mapping:
+	 2 -------- 3
+	 | \   tri2 |
+	 |    \     |
+	 | tri1  \  |
+	 0 -------- 1 */
+void mdisp_join_tris(MDisps *dst, MDisps *tri1, MDisps *tri2)
+{
+	int side, st;
+	int S, x, y, crn;
+	float face_u, face_v, crn_u, crn_v;
+	float (*out)[3];
+	MDisps *src;
+
+	if(dst->disps)
+		MEM_freeN(dst->disps);
+
+	side = sqrt(tri1->totdisp / 3);
+	st = (side<<1)-1;
+
+	dst->totdisp = 4 * side * side;
+	out = dst->disps = MEM_callocN(3*dst->totdisp*sizeof(float), "join disps");
+
+	for(S = 0; S < 4; S++)
+		for(y = 0; y < side; ++y)
+			for(x = 0; x < side; ++x, ++out) {
+				mdisp_rot_crn_to_face(S, 4, st, x, y, &face_u, &face_v);
+				face_u = st - 1 - face_u;
+
+				if(face_v > face_u) {
+					src = tri2;
+					face_u = st - 1 - face_u;
+					face_v = st - 1 - face_v;
+				} else src = tri1;
+
+				crn = mdisp_rot_face_to_crn(3, st, face_u, face_v, &crn_u, &crn_v);
+
+				old_mdisps_bilinear((*out), &src->disps[crn*side*side], side, crn_u, crn_v);
+				(*out)[0] = 0;
+				(*out)[1] = 0;
+			}
+}