[Bf-blender-cvs] [f5e098dd113] soc-2019-adaptive-cloth: Cloth: added derivative and compression metric support

[ishbosamiya]

Commit: f5e098dd113b6f2eba67135a970a953c50cc61f5
Author: ishbosamiya
Date:   Thu Jul 18 19:36:48 2019 +0530
Branches: soc-2019-adaptive-cloth
https://developer.blender.org/rBf5e098dd113b6f2eba67135a970a953c50cc61f5

Cloth: added derivative and compression metric support

===================================================================

M	source/blender/blenkernel/intern/cloth_remeshing.cpp

===================================================================

diff --git a/source/blender/blenkernel/intern/cloth_remeshing.cpp b/source/blender/blenkernel/intern/cloth_remeshing.cpp
index e3223135d99..dbb212758bd 100644
--- a/source/blender/blenkernel/intern/cloth_remeshing.cpp
+++ b/source/blender/blenkernel/intern/cloth_remeshing.cpp
@@ -130,6 +130,7 @@ static void cloth_remeshing_update_active_faces(vector<BMFace *> &active_faces,
                                                 BMesh *bm,
                                                 BMEdge *e);
 static ClothSizing cloth_remeshing_find_average_sizing(ClothSizing &size_01, ClothSizing &size_02);
+static void mul_m2_m2m2(float r[2][2], float a[2][2], float b[2][2]);
 
 static CustomData_MeshMasks cloth_remeshing_get_cd_mesh_masks(void)
 {
@@ -520,7 +521,6 @@ static void cloth_remeshing_flip_edges(BMesh *bm,
       /* BM_EDGEROT_CHECK_SPLICE sets it up for BM_CREATE_NO_DOUBLE */
       BMEdge *new_edge = BM_edge_rotate(bm, edge, true, BM_EDGEROT_CHECK_SPLICE);
       BLI_assert(new_edge != NULL);
-      /* TODO(Ish): need to update active_faces */
       cloth_remeshing_update_active_faces(active_faces, bm, new_edge);
     }
   }
@@ -1597,6 +1597,18 @@ static void cloth_remeshing_dynamic(ClothModifierData *clmd)
 #endif
 }
 
+static void cloth_remeshing_face_data(BMesh *bm, BMFace *f, float r_mat[2][2])
+{
+  BMVert *v[3];
+  float uv[3][2];
+  BM_face_as_array_vert_tri(f, v);
+  for (int i = 0; i < 3; i++) {
+    cloth_remeshing_uv_of_vert_in_face(bm, f, v[i], uv[i]);
+  }
+  sub_v2_v2v2(r_mat[0], uv[1], uv[0]);
+  sub_v2_v2v2(r_mat[1], uv[2], uv[0]);
+}
+
 static float cloth_remeshing_calc_area(BMesh *bm, BMFace *f)
 {
   /* TODO(Ish): Area calculation might be with respect to World Space, need
@@ -1660,6 +1672,35 @@ static void cloth_remeshing_eigen_decomposition(float mat[2][2], float r_mat[2][
   }
 }
 
+static void diag_m2_v2(float r[2][2], float a[2])
+{
+  zero_m2(r);
+  for (int i = 0; i < 2; i++) {
+    r[i][i] = a[i];
+  }
+}
+
+static void cloth_remeshing_compression_metric(float mat[2][2], float r_mat[2][2])
+{
+  float l[2];
+  float q[2][2];
+
+  cloth_remeshing_eigen_decomposition(mat, q, l);
+  float q_t[2][2];
+  copy_m2_m2(q_t, q);
+  transpose_m2(q_t);
+
+  for (int i = 0; i < 2; i++) {
+    l[i] = max(1 - sqrtf(l[i]), 0.0f);
+  }
+
+  float temp_mat[2][2];
+  float diag_l[2][2];
+  diag_m2_v2(diag_l, l);
+  mul_m2_m2m2(temp_mat, q, diag_l);
+  mul_m2_m2m2(r_mat, temp_mat, q_t);
+}
+
 #define NEXT(x) ((x) < 2 ? (x) + 1 : (x)-2)
 #define PREV(x) ((x) > 0 ? (x)-1 : (x) + 2)
 
@@ -1729,13 +1770,55 @@ static void cloth_remeshing_curvature(BMesh *bm, BMFace *f, float r_mat[2][2])
   mul_m2_fl(r_mat, 1.0f / cloth_remeshing_calc_area(bm, f));
 }
 
+static void transpose_m32_m23(float r_mat[3][2], float mat[2][3])
+{
+  r_mat[0][0] = mat[0][0];
+  r_mat[0][1] = mat[1][0];
+
+  r_mat[1][0] = mat[0][1];
+  r_mat[1][1] = mat[1][1];
+
+  r_mat[2][0] = mat[0][2];
+  r_mat[2][1] = mat[1][2];
+}
+
+static void zero_m32(float r[3][2])
+{
+  for (int i = 0; i < 3; i++) {
+    for (int j = 0; j < 2; j++) {
+      r[i][j] = 0.0f;
+    }
+  }
+}
+
+static void mul_m32_m32m22(float r[3][2], float a[3][2], float b[2][2])
+{
+  zero_m32(r);
+  for (int i = 0; i < 3; i++) {
+    for (int j = 0; j < 2; j++) {
+      for (int k = 0; k < 2; k++) {
+        r[i][j] += a[i][k] * b[k][j];
+      }
+    }
+  }
+}
+
 static void cloth_remeshing_derivative(
-    float m_01[3], float m_02[3], float m_03[3], BMFace *f, float r_mat[3][2])
+    BMesh *bm, float m_01[3], float m_02[3], float m_03[3], BMFace *f, float r_mat[3][2])
 {
-  /* TODO(Ish) */
+  float mat[2][3];
+  sub_v3_v3v3(mat[0], m_02, m_01);
+  sub_v3_v3v3(mat[1], m_03, m_01);
+  float mat_t[3][2];
+  transpose_m32_m23(mat_t, mat);
+
+  float face_dm_inv[2][2];
+  cloth_remeshing_face_data(bm, f, face_dm_inv);
+  transpose_m2(face_dm_inv);
+  mul_m32_m32m22(r_mat, mat_t, face_dm_inv);
 }
 
-static void transpose_m32(float r_mat[2][3], float mat[3][2])
+static void transpose_m23_m32(float r_mat[2][3], float mat[3][2])
 {
   r_mat[0][0] = mat[0][0];
   r_mat[0][1] = mat[1][0];
@@ -1785,22 +1868,25 @@ static ClothSizing cloth_remeshing_compute_face_sizing(ClothModifierData *clmd,
   float sizing_s[2][2];
   cloth_remeshing_curvature(bm, f, sizing_s);
   float sizing_f[3][2];
-  cloth_remeshing_derivative(cv[0]->x, cv[1]->x, cv[2]->x, f, sizing_f);
+  cloth_remeshing_derivative(bm, cv[0]->x, cv[1]->x, cv[2]->x, f, sizing_f);
   float sizing_v[3][2];
-  cloth_remeshing_derivative(cv[0]->v, cv[1]->v, cv[2]->v, f, sizing_v);
+  cloth_remeshing_derivative(bm, cv[0]->v, cv[1]->v, cv[2]->v, f, sizing_v);
 
   float sizing_s_t[2][2];
   copy_m2_m2(sizing_s_t, sizing_s);
   transpose_m2(sizing_s_t);
+  float sizing_f_t[2][3];
+  transpose_m23_m32(sizing_f_t, sizing_f);
   float sizing_v_t[2][3];
-  transpose_m32(sizing_v_t, sizing_v);
+  transpose_m23_m32(sizing_v_t, sizing_v);
 
   float curv[2][2];
   mul_m2_m2m2(curv, sizing_s_t, sizing_s);
 
-  /* TODO(Ish): compression metric */
   float comp[2][2];
-  zero_m2(comp);
+  float f_x_ft[2][2];
+  mul_m2_m23m32(f_x_ft, sizing_f_t, sizing_f);
+  cloth_remeshing_compression_metric(f_x_ft, comp);
 
   float dvel[2][2];
   mul_m2_m23m32(dvel, sizing_v_t, sizing_v);
@@ -1844,10 +1930,7 @@ static ClothSizing cloth_remeshing_compute_face_sizing(ClothModifierData *clmd,
   float result[2][2];
   float temp_result[2][2];
   float diag_l[2][2];
-  zero_m2(diag_l);
-  for (int i = 0; i < 2; i++) {
-    diag_l[i][i] = l[i];
-  }
+  diag_m2_v2(diag_l, l);
   float q_t[2][2];
   copy_m2_m2(q_t, q);
   transpose_m2(q_t);