Merge branch 'lm_fdm_gizmo_experiments'

This commit is contained in:
Lukas Matena 2020-05-18 16:08:29 +02:00
commit d0e0387b82
4 changed files with 160 additions and 116 deletions

View file

@ -15,6 +15,8 @@
namespace Slic3r {
namespace GUI {
static constexpr size_t MaxVertexBuffers = 50;
GLGizmoFdmSupports::GLGizmoFdmSupports(GLCanvas3D& parent, const std::string& icon_filename, unsigned int sprite_id)
: GLGizmoBase(parent, icon_filename, sprite_id)
, m_quadric(nullptr)
@ -123,10 +125,9 @@ void GLGizmoFdmSupports::render_triangles(const Selection& selection) const
// Now render both enforcers and blockers.
for (int i=0; i<2; ++i) {
if (m_ivas[mesh_id][i].has_VBOs()) {
glsafe(::glColor4f(i ? 1.f : 0.2f, 0.2f, i ? 0.2f : 1.0f, 0.5f));
m_ivas[mesh_id][i].render();
}
glsafe(::glColor4f(i ? 1.f : 0.2f, 0.2f, i ? 0.2f : 1.0f, 0.5f));
for (const GLIndexedVertexArray& iva : m_ivas[mesh_id][i])
iva.render();
}
glsafe(::glPopMatrix());
}
@ -205,8 +206,14 @@ void GLGizmoFdmSupports::update_from_model_object()
++num_of_volumes;
m_selected_facets.resize(num_of_volumes);
m_neighbors.resize(num_of_volumes);
m_ivas.clear();
m_ivas.resize(num_of_volumes);
for (size_t i=0; i<num_of_volumes; ++i) {
m_ivas[i][0].reserve(MaxVertexBuffers);
m_ivas[i][1].reserve(MaxVertexBuffers);
}
int volume_id = -1;
for (const ModelVolume* mv : mo->volumes) {
@ -226,7 +233,8 @@ void GLGizmoFdmSupports::update_from_model_object()
for (int i : list)
m_selected_facets[volume_id][i] = type;
}
update_vertex_buffers(mv, volume_id, true, true);
update_vertex_buffers(mesh, volume_id, FacetSupportType::ENFORCER);
update_vertex_buffers(mesh, volume_id, FacetSupportType::BLOCKER);
m_neighbors[volume_id].resize(3 * mesh->its.indices.size());
@ -325,7 +333,7 @@ bool GLGizmoFdmSupports::gizmo_event(SLAGizmoEventType action, const Vec2d& mous
Vec3f closest_hit = Vec3f::Zero();
double closest_hit_squared_distance = std::numeric_limits<double>::max();
size_t closest_facet = 0;
size_t closest_hit_mesh_id = size_t(-1);
int closest_hit_mesh_id = -1;
// Transformations of individual meshes
std::vector<Transform3d> trafo_matrices;
@ -368,17 +376,22 @@ bool GLGizmoFdmSupports::gizmo_event(SLAGizmoEventType action, const Vec2d& mous
}
// We now know where the ray hit, let's save it and cast another ray
if (closest_hit_mesh_id != size_t(-1)) // only if there is at least one hit
hit_positions_and_facet_ids[closest_hit_mesh_id].emplace_back(closest_hit, closest_facet);
some_mesh_was_hit = true;
if (some_mesh_was_hit) {
// Now propagate the hits
mesh_id = -1;
const TriangleMesh* mesh = nullptr;
for (const ModelVolume* mv : mo->volumes) {
if (! mv->is_model_part())
continue;
++mesh_id;
if (mesh_id == closest_hit_mesh_id) {
mesh = &mv->mesh();
break;
}
}
// Now propagate the hits
mesh_id = -1;
for (const ModelVolume* mv : mo->volumes) {
if (! mv->is_model_part())
continue;
++mesh_id;
bool update_both = false;
const Transform3d& trafo_matrix = trafo_matrices[mesh_id];
@ -389,89 +402,96 @@ bool GLGizmoFdmSupports::gizmo_event(SLAGizmoEventType action, const Vec2d& mous
const float avg_scaling = (sf(0) + sf(1) + sf(2))/3.;
const float limit = pow(m_cursor_radius/avg_scaling , 2.f);
// For all hits on this mesh...
for (const std::pair<Vec3f, size_t>& hit_and_facet : hit_positions_and_facet_ids[mesh_id]) {
some_mesh_was_hit = true;
const TriangleMesh* mesh = &mv->mesh();
std::vector<NeighborData>& neighbors = m_neighbors[mesh_id];
const std::pair<Vec3f, size_t>& hit_and_facet = { closest_hit, closest_facet };
// Calculate direction from camera to the hit (in mesh coords):
Vec3f dir = ((trafo_matrix.inverse() * camera.get_position()).cast<float>() - hit_and_facet.first).normalized();
const std::vector<NeighborData>& neighbors = m_neighbors[mesh_id];
// A lambda to calculate distance from the centerline:
auto squared_distance_from_line = [&hit_and_facet, &dir](const Vec3f point) -> float {
Vec3f diff = hit_and_facet.first - point;
return (diff - diff.dot(dir) * dir).squaredNorm();
};
// Calculate direction from camera to the hit (in mesh coords):
Vec3f dir = ((trafo_matrix.inverse() * camera.get_position()).cast<float>() - hit_and_facet.first).normalized();
// A lambda to determine whether this facet is potentionally visible (still can be obscured)
auto faces_camera = [&dir](const ModelVolume* mv, const size_t& facet) -> bool {
return (mv->mesh().stl.facet_start[facet].normal.dot(dir) > 0.);
};
// Now start with the facet the pointer points to and check all adjacent facets. neighbors vector stores
// pairs of vertex_idx - facet_idx and is sorted with respect to the former. Neighboring facet index can be
// quickly found by finding a vertex in the list and read the respective facet ids.
std::vector<size_t> facets_to_select{hit_and_facet.second};
NeighborData vertex = std::make_pair(0, 0);
std::vector<bool> visited(m_selected_facets[mesh_id].size(), false); // keep track of facets we already processed
size_t facet_idx = 0; // index into facets_to_select
auto it = neighbors.end();
while (facet_idx < facets_to_select.size()) {
size_t facet = facets_to_select[facet_idx];
if (! visited[facet]) {
// check all three vertices and in case they're close enough, find the remaining facets
// and add them to the list to be proccessed later
for (size_t i=0; i<3; ++i) {
vertex.first = mesh->its.indices[facet](i); // vertex index
float dist = squared_distance_from_line(mesh->its.vertices[vertex.first]);
if (dist < limit) {
it = std::lower_bound(neighbors.begin(), neighbors.end(), vertex);
while (it != neighbors.end() && it->first == vertex.first) {
if (it->second != facet && faces_camera(mv, it->second))
facets_to_select.push_back(it->second);
++it;
}
// A lambda to calculate distance from the centerline:
auto squared_distance_from_line = [&hit_and_facet, &dir](const Vec3f& point) -> float {
Vec3f diff = hit_and_facet.first - point;
return (diff - diff.dot(dir) * dir).squaredNorm();
};
// A lambda to determine whether this facet is potentionally visible (still can be obscured)
auto faces_camera = [&dir, &mesh](const size_t& facet) -> bool {
return (mesh->stl.facet_start[facet].normal.dot(dir) > 0.);
};
// Now start with the facet the pointer points to and check all adjacent facets. neighbors vector stores
// pairs of vertex_idx - facet_idx and is sorted with respect to the former. Neighboring facet index can be
// quickly found by finding a vertex in the list and read the respective facet ids.
std::vector<size_t> facets_to_select{hit_and_facet.second};
NeighborData vertex = std::make_pair(0, 0);
std::vector<bool> visited(m_selected_facets[mesh_id].size(), false); // keep track of facets we already processed
size_t facet_idx = 0; // index into facets_to_select
auto it = neighbors.end();
while (facet_idx < facets_to_select.size()) {
size_t facet = facets_to_select[facet_idx];
if (! visited[facet]) {
// check all three vertices and in case they're close enough, find the remaining facets
// and add them to the list to be proccessed later
for (size_t i=0; i<3; ++i) {
vertex.first = mesh->its.indices[facet](i); // vertex index
float dist = squared_distance_from_line(mesh->its.vertices[vertex.first]);
if (dist < limit) {
it = std::lower_bound(neighbors.begin(), neighbors.end(), vertex);
while (it != neighbors.end() && it->first == vertex.first) {
if (it->second != facet && faces_camera(it->second))
facets_to_select.push_back(it->second);
++it;
}
}
visited[facet] = true;
}
++facet_idx;
visited[facet] = true;
}
++facet_idx;
}
// Now just select all facets that passed.
for (size_t next_facet : facets_to_select) {
FacetSupportType& facet = m_selected_facets[mesh_id][next_facet];
std::vector<size_t> new_facets;
new_facets.reserve(facets_to_select.size());
if (facet != new_state && facet != FacetSupportType::NONE) {
// Now just select all facets that passed and remember which
// ones have really changed state.
for (size_t next_facet : facets_to_select) {
FacetSupportType& facet = m_selected_facets[mesh_id][next_facet];
if (facet != new_state) {
if (facet != FacetSupportType::NONE) {
// this triangle is currently in the other VBA.
// Both VBAs need to be refreshed.
update_both = true;
}
facet = new_state;
new_facets.push_back(next_facet);
}
}
if (! new_facets.empty()) {
if (new_state != FacetSupportType::NONE) {
// append triangles into the respective VBA
update_vertex_buffers(mesh, mesh_id, new_state, &new_facets);
if (update_both) {
auto other = new_state == FacetSupportType::ENFORCER
? FacetSupportType::BLOCKER
: FacetSupportType::ENFORCER;
update_vertex_buffers(mesh, mesh_id, other); // regenerate the other VBA
}
}
else {
update_vertex_buffers(mesh, mesh_id, FacetSupportType::ENFORCER);
update_vertex_buffers(mesh, mesh_id, FacetSupportType::BLOCKER);
}
}
update_vertex_buffers(mv, mesh_id,
new_state == FacetSupportType::ENFORCER || update_both,
new_state == FacetSupportType::BLOCKER || update_both
);
}
if (some_mesh_was_hit)
{
if (m_button_down == Button::None)
m_button_down = ((action == SLAGizmoEventType::LeftDown) ? Button::Left : Button::Right);
// Force rendering. In case the user is dragging, the queue can be
// flooded by wxEVT_MOVING event and rendering would be skipped.
m_parent.render();
return true;
}
if (action == SLAGizmoEventType::Dragging && m_button_down != Button::None) {
// Same as above. We don't want the cursor to freeze when we
// leave the mesh while painting.
m_parent.render();
if (action == SLAGizmoEventType::Dragging && m_button_down != Button::None)
return true;
}
}
if ((action == SLAGizmoEventType::LeftUp || action == SLAGizmoEventType::RightUp)
@ -493,34 +513,54 @@ bool GLGizmoFdmSupports::gizmo_event(SLAGizmoEventType action, const Vec2d& mous
}
void GLGizmoFdmSupports::update_vertex_buffers(const ModelVolume* mv,
void GLGizmoFdmSupports::update_vertex_buffers(const TriangleMesh* mesh,
int mesh_id,
bool update_enforcers,
bool update_blockers)
FacetSupportType type,
const std::vector<size_t>* new_facets)
{
const TriangleMesh* mesh = &mv->mesh();
std::vector<GLIndexedVertexArray>& ivas = m_ivas[mesh_id][type == FacetSupportType::ENFORCER ? 0 : 1];
for (FacetSupportType type : {FacetSupportType::ENFORCER, FacetSupportType::BLOCKER}) {
if ((type == FacetSupportType::ENFORCER && ! update_enforcers)
|| (type == FacetSupportType::BLOCKER && ! update_blockers))
continue;
// lambda to push facet into vertex buffer
auto push_facet = [this, &mesh, &mesh_id](size_t idx, GLIndexedVertexArray& iva) {
for (int i=0; i<3; ++i)
iva.push_geometry(
mesh->its.vertices[mesh->its.indices[idx](i)].cast<double>(),
m_c->raycaster()->raycasters()[mesh_id]->get_triangle_normal(idx).cast<double>()
);
size_t num = iva.triangle_indices_size;
iva.push_triangle(num, num+1, num+2);
};
GLIndexedVertexArray& iva = m_ivas[mesh_id][type==FacetSupportType::ENFORCER ? 0 : 1];
iva.release_geometry();
size_t triangle_cnt=0;
if (ivas.size() == MaxVertexBuffers || ! new_facets) {
// If there are too many or they should be regenerated, make one large
// GLVertexBufferArray.
ivas.clear(); // destructors release geometry
ivas.push_back(GLIndexedVertexArray());
bool pushed = false;
for (size_t facet_idx=0; facet_idx<m_selected_facets[mesh_id].size(); ++facet_idx) {
FacetSupportType status = m_selected_facets[mesh_id][facet_idx];
if (status != type)
continue;
for (int i=0; i<3; ++i)
iva.push_geometry(mesh->its.vertices[mesh->its.indices[facet_idx](i)].cast<double>(),
MeshRaycaster::get_triangle_normal(mesh->its, facet_idx).cast<double>());
iva.push_triangle(3*triangle_cnt, 3*triangle_cnt+1, 3*triangle_cnt+2);
++triangle_cnt;
if (m_selected_facets[mesh_id][facet_idx] == type) {
push_facet(facet_idx, ivas.back());
pushed = true;
}
}
if (! m_selected_facets[mesh_id].empty())
iva.finalize_geometry(true);
if (pushed)
ivas.back().finalize_geometry(true);
else
ivas.pop_back();
} else {
// we are only appending - let's make new vertex array and let the old ones live
ivas.push_back(GLIndexedVertexArray());
for (size_t facet_idx : *new_facets)
push_facet(facet_idx, ivas.back());
if (! new_facets->empty())
ivas.back().finalize_geometry(true);
else
ivas.pop_back();
}
}
@ -553,7 +593,8 @@ void GLGizmoFdmSupports::select_facets_by_angle(float threshold_deg, bool overwr
? FacetSupportType::BLOCKER
: FacetSupportType::ENFORCER;
}
update_vertex_buffers(mv, mesh_id, true, true);
update_vertex_buffers(&mv->mesh(), mesh_id, FacetSupportType::ENFORCER);
update_vertex_buffers(&mv->mesh(), mesh_id, FacetSupportType::BLOCKER);
}
Plater::TakeSnapshot(wxGetApp().plater(), block ? _L("Block supports by angle")
@ -623,7 +664,8 @@ void GLGizmoFdmSupports::on_render_input_window(float x, float y, float bottom_l
if (mv->is_model_part()) {
m_selected_facets[idx].assign(m_selected_facets[idx].size(), FacetSupportType::NONE);
mv->m_supported_facets.clear();
update_vertex_buffers(mv, idx, true, true);
update_vertex_buffers(&mv->mesh(), idx, FacetSupportType::ENFORCER);
update_vertex_buffers(&mv->mesh(), idx, FacetSupportType::BLOCKER);
m_parent.set_as_dirty();
}
}

View file

@ -33,14 +33,16 @@ private:
// individual facets (one of the enum values above).
std::vector<std::vector<FacetSupportType>> m_selected_facets;
// Store two vertex buffer arrays (for enforcers/blockers)
// for each model-part volume.
std::vector<std::array<GLIndexedVertexArray, 2>> m_ivas;
// Vertex buffer arrays for each model-part volume. There is a vector of
// arrays so that adding triangles can be done without regenerating all
// other triangles. Enforcers and blockers are of course separate.
std::vector<std::array<std::vector<GLIndexedVertexArray>, 2>> m_ivas;
void update_vertex_buffers(const ModelVolume* mv,
void update_vertex_buffers(const TriangleMesh* mesh,
int mesh_id,
bool update_enforcers,
bool update_blockers);
FacetSupportType type, // enforcers / blockers
const std::vector<size_t>* new_facets = nullptr); // nullptr -> regenerate all
public:
GLGizmoFdmSupports(GLCanvas3D& parent, const std::string& icon_filename, unsigned int sprite_id);

View file

@ -95,11 +95,9 @@ void MeshClipper::recalculate_triangles()
}
Vec3f MeshRaycaster::get_triangle_normal(const indexed_triangle_set& its, size_t facet_idx)
Vec3f MeshRaycaster::get_triangle_normal(size_t facet_idx) const
{
Vec3f a(its.vertices[its.indices[facet_idx](1)] - its.vertices[its.indices[facet_idx](0)]);
Vec3f b(its.vertices[its.indices[facet_idx](2)] - its.vertices[its.indices[facet_idx](0)]);
return Vec3f(a.cross(b)).normalized();
return m_normals[facet_idx];
}
void MeshRaycaster::line_from_mouse_pos(const Vec2d& mouse_pos, const Transform3d& trafo, const Camera& camera,
@ -218,12 +216,9 @@ Vec3f MeshRaycaster::get_closest_point(const Vec3f& point, Vec3f* normal) const
int idx = 0;
Vec3d closest_point;
m_emesh.squared_distance(point.cast<double>(), idx, closest_point);
if (normal) {
auto indices = m_emesh.F().row(idx);
Vec3d a(m_emesh.V().row(indices(1)) - m_emesh.V().row(indices(0)));
Vec3d b(m_emesh.V().row(indices(2)) - m_emesh.V().row(indices(0)));
*normal = Vec3f(a.cross(b).cast<float>());
}
if (normal)
*normal = m_normals[idx];
return closest_point.cast<float>();
}

View file

@ -108,7 +108,11 @@ public:
// The pointer can be invalidated after constructor returns.
MeshRaycaster(const TriangleMesh& mesh)
: m_emesh(mesh)
{}
{
m_normals.reserve(mesh.stl.facet_start.size());
for (const stl_facet& facet : mesh.stl.facet_start)
m_normals.push_back(facet.normal);
}
void line_from_mouse_pos(const Vec2d& mouse_pos, const Transform3d& trafo, const Camera& camera,
Vec3d& point, Vec3d& direction) const;
@ -140,10 +144,11 @@ public:
Vec3f get_closest_point(const Vec3f& point, Vec3f* normal = nullptr) const;
static Vec3f get_triangle_normal(const indexed_triangle_set& its, size_t facet_idx);
Vec3f get_triangle_normal(size_t facet_idx) const;
private:
sla::EigenMesh3D m_emesh;
std::vector<stl_normal> m_normals;
};