PrusaSlicer-NonPlainar/tests/libslic3r/test_quadric_edge_collapse.cpp
2021-10-19 13:34:26 +02:00

243 lines
8.9 KiB
C++

#include <catch2/catch.hpp>
#include <test_utils.hpp>
#include <libslic3r/QuadricEdgeCollapse.hpp>
#include <libslic3r/TriangleMesh.hpp> // its - indexed_triangle_set
#include <libslic3r/SimplifyMesh.hpp> // no priority queue
#include "libslic3r/AABBTreeIndirect.hpp" // is similar
using namespace Slic3r;
namespace Private {
struct Similarity
{
float max_distance = 0.f;
float average_distance = 0.f;
Similarity() = default;
Similarity(float max_distance, float average_distance)
: max_distance(max_distance), average_distance(average_distance)
{}
};
// border for our algorithm with frog_leg model and decimation to 5%
Similarity frog_leg_5(0.32f, 0.043f);
Similarity get_similarity(const indexed_triangle_set &from,
const indexed_triangle_set &to)
{
// create ABBTree
auto tree = AABBTreeIndirect::build_aabb_tree_over_indexed_triangle_set(
from.vertices, from.indices);
float sum_distance = 0.f;
float max_distance = 0.f;
auto collect_distances = [&](const Vec3f &surface_point) {
size_t hit_idx;
Vec3f hit_point;
float distance2 =
AABBTreeIndirect::squared_distance_to_indexed_triangle_set(
from.vertices, from.indices, tree, surface_point, hit_idx,
hit_point);
float distance = sqrt(distance2);
if (max_distance < distance) max_distance = distance;
sum_distance += distance;
};
for (const Vec3f &vertex : to.vertices) { collect_distances(vertex); }
for (const Vec3i &t : to.indices) {
Vec3f center(0, 0, 0);
for (size_t i = 0; i < 3; ++i) { center += to.vertices[t[i]] / 3; }
collect_distances(center);
}
size_t count = to.vertices.size() + to.indices.size();
float average_distance = sum_distance / count;
std::cout << "max_distance = " << max_distance << ", average_distance = " << average_distance << std::endl;
return Similarity(max_distance, average_distance);
}
void is_better_similarity(const indexed_triangle_set &its_first,
const indexed_triangle_set &its_second,
const Similarity & compare)
{
Similarity s1 = get_similarity(its_first, its_second);
Similarity s2 = get_similarity(its_second, its_first);
CHECK(s1.average_distance < compare.average_distance);
CHECK(s1.max_distance < compare.max_distance);
CHECK(s2.average_distance < compare.average_distance);
CHECK(s2.max_distance < compare.max_distance);
}
void is_worse_similarity(const indexed_triangle_set &its_first,
const indexed_triangle_set &its_second,
const Similarity & compare)
{
Similarity s1 = get_similarity(its_first, its_second);
Similarity s2 = get_similarity(its_second, its_first);
if (s1.max_distance < compare.max_distance &&
s2.max_distance < compare.max_distance)
CHECK(false);
}
bool exist_triangle_with_twice_vertices(const std::vector<stl_triangle_vertex_indices> &indices)
{
for (const auto &face : indices)
if (face[0] == face[1] || face[0] == face[2] || face[1] == face[2])
return true;
return false;
}
} // namespace Private
TEST_CASE("Reduce one edge by Quadric Edge Collapse", "[its]")
{
indexed_triangle_set its;
its.vertices = {Vec3f(-1.f, 0.f, 0.f), Vec3f(0.f, 1.f, 0.f),
Vec3f(1.f, 0.f, 0.f), Vec3f(0.f, 0.f, 1.f),
// vertex to be removed
Vec3f(0.9f, .1f, -.1f)};
its.indices = {Vec3i(1, 0, 3), Vec3i(2, 1, 3), Vec3i(0, 2, 3),
Vec3i(0, 1, 4), Vec3i(1, 2, 4), Vec3i(2, 0, 4)};
// edge to remove is between vertices 2 and 4 on trinagles 4 and 5
indexed_triangle_set its_ = its; // copy
// its_write_obj(its, "tetrhedron_in.obj");
uint32_t wanted_count = its.indices.size() - 1;
its_quadric_edge_collapse(its, wanted_count);
// its_write_obj(its, "tetrhedron_out.obj");
CHECK(its.indices.size() == 4);
CHECK(its.vertices.size() == 4);
for (size_t i = 0; i < 3; i++) {
CHECK(its.indices[i] == its_.indices[i]);
}
for (size_t i = 0; i < 4; i++) {
if (i == 2) continue;
CHECK(its.vertices[i] == its_.vertices[i]);
}
const Vec3f &v = its.vertices[2]; // new vertex
const Vec3f &v2 = its_.vertices[2]; // moved vertex
const Vec3f &v4 = its_.vertices[4]; // removed vertex
for (size_t i = 0; i < 3; i++) {
bool is_between = (v[i] < v4[i] && v[i] > v2[i]) ||
(v[i] > v4[i] && v[i] < v2[i]);
CHECK(is_between);
}
Private::Similarity max_similarity(0.75f, 0.014f);
Private::is_better_similarity(its, its_, max_similarity);
}
TEST_CASE("Simplify frog_legs.obj to 5% by Quadric edge collapse", "[its][quadric_edge_collapse]")
{
TriangleMesh mesh = load_model("frog_legs.obj");
double original_volume = its_volume(mesh.its);
uint32_t wanted_count = mesh.its.indices.size() * 0.05;
REQUIRE_FALSE(mesh.empty());
indexed_triangle_set its = mesh.its; // copy
float max_error = std::numeric_limits<float>::max();
its_quadric_edge_collapse(its, wanted_count, &max_error);
// its_write_obj(its, "frog_legs_qec.obj");
CHECK(its.indices.size() <= wanted_count);
double volume = its_volume(its);
CHECK(fabs(original_volume - volume) < 33.);
Private::is_better_similarity(mesh.its, its, Private::frog_leg_5);
}
#include <libigl/igl/qslim.h>
#include "Simplify.h"
TEST_CASE("Simplify frog_legs.obj to 5% by IGL/qslim", "[]")
{
std::string obj_filename = "frog_legs.obj";
TriangleMesh mesh = load_model(obj_filename);
REQUIRE_FALSE(mesh.empty());
indexed_triangle_set &its = mesh.its;
double original_volume = its_volume(its);
uint32_t wanted_count = its.indices.size() * 0.05;
Eigen::MatrixXd V(its.vertices.size(), 3);
Eigen::MatrixXi F(its.indices.size(), 3);
for (size_t j = 0; j < its.vertices.size(); ++j) {
Vec3d vd = its.vertices[j].cast<double>();
for (int i = 0; i < 3; ++i) V(j, i) = vd(i);
}
for (size_t j = 0; j < its.indices.size(); ++j) {
const auto &f = its.indices[j];
for (int i = 0; i < 3; ++i) F(j, i) = f(i);
}
size_t max_m = wanted_count;
Eigen::MatrixXd U;
Eigen::MatrixXi G;
Eigen::VectorXi J, I;
CHECK(igl::qslim(V, F, max_m, U, G, J, I));
// convert to its
indexed_triangle_set its_out;
its_out.vertices.reserve(U.size()/3);
its_out.indices.reserve(G.size()/3);
for (size_t i = 0; i < U.size()/3; i++)
its_out.vertices.emplace_back(U(i, 0), U(i, 1), U(i, 2));
for (size_t i = 0; i < G.size()/3; i++)
its_out.indices.emplace_back(G(i, 0), G(i, 1), G(i, 2));
// check if algorithm is still worse than our
Private::is_worse_similarity(its_out, its, Private::frog_leg_5);
// its_out, its --> avg_distance: 0.0351217, max_distance 0.364316
// its, its_out --> avg_distance: 0.0412358, max_distance 0.238913
}
TEST_CASE("Simplify frog_legs.obj to 5% by simplify", "[]") {
std::string obj_filename = "frog_legs.obj";
TriangleMesh mesh = load_model(obj_filename);
uint32_t wanted_count = mesh.its.indices.size() * 0.05;
Simplify::load_obj((TEST_DATA_DIR PATH_SEPARATOR + obj_filename).c_str());
Simplify::simplify_mesh(wanted_count, 5, true);
// convert to its
indexed_triangle_set its_out;
its_out.vertices.reserve(Simplify::vertices.size());
its_out.indices.reserve(Simplify::triangles.size());
for (size_t i = 0; i < Simplify::vertices.size(); i++) {
const Simplify::Vertex &v = Simplify::vertices[i];
its_out.vertices.emplace_back(v.p.x, v.p.y, v.p.z);
}
for (size_t i = 0; i < Simplify::triangles.size(); i++) {
const Simplify::Triangle &t = Simplify::triangles[i];
its_out.indices.emplace_back(t.v[0], t.v[1], t.v[2]);
}
// check if algorithm is still worse than our
Private::is_worse_similarity(its_out, mesh.its, Private::frog_leg_5);
// its_out, mesh.its --> max_distance = 0.700494, average_distance = 0.0902524
// mesh.its, its_out --> max_distance = 0.393184, average_distance = 0.0537392
}
TEST_CASE("Simplify trouble case", "[its]")
{
TriangleMesh tm = load_model("simplification.obj");
REQUIRE_FALSE(tm.empty());
float max_error = std::numeric_limits<float>::max();
uint32_t wanted_count = 0;
its_quadric_edge_collapse(tm.its, wanted_count, &max_error);
CHECK(!Private::exist_triangle_with_twice_vertices(tm.its.indices));
}
TEST_CASE("Simplified cube should not be empty.", "[its]")
{
auto its = its_make_cube(1, 2, 3);
float max_error = std::numeric_limits<float>::max();
uint32_t wanted_count = 0;
its_quadric_edge_collapse(its, wanted_count, &max_error);
CHECK(!its.indices.empty());
}