1. Preparation
Use Std::vector to know the points:
(1) Memory contiguous container, sort of like an array
(2) Insert and delete elements are slower than std::list-because data migration
(3) Adding elements may cause memory allocations and data migrations.
2. Questions
ANYCAD::API::P Ointcloudnode uses floatlist and std::vector<float> to store some column points [x0, y0, z0, x1, y1, Z1, ...]:
void |
Setpoints (const floatlist &buffer) |
If you want to display n points, you need 3n of length:
In order to simulate the motion trajectory of an object in space, which is a series of points, how can the dynamic drawing be realized efficiently?
3. The programme
The most basic approach:
Variables defined:
Pointcloudnode m_pointcoud;std::_vector<float> m_points;
Each time you call Push_back add point:
void Onaddpoint (x, Y, z) { m_points.push (x); M_points.push (y); M_points.push (z); M_pointcloud.setpoints (m_points); Render ();}
"Optimization 1": To reduce the number of points displayed to cause memory problems and efficiency problems, you can limit how many points to display
int max_point3_count = Max_point_count * 3;
void OnAddPointV1 (x, Y, z) { if (m_points.size () > Max_point3_count) { m_points.erase (M_points.begin () ); M_points.erase (M_points.begin ()); M_points.erase (M_points.begin ()); } M_points.push (x); M_points.push (y); M_points.push (z); M_pointcloud.setpoints (m_points); Render ();}
What is the problem introduced by onAddPointV1?
Optimization 2: Avoid allocating memory every time the vector is allocated, specifying the initial memory size of the vector
M_points.reserve (Max_point3_count);
Optimization 3: Avoid data migration due to deletion of head element engine
The new points are placed at the end of the team or the team head, and the results are the same for the display. So you can overwrite the "expired" point.
int m_totalcount = 0;
void OnAddPointV3 (x, Y, z) { ++m_totalcount; if (M_totalcount <= max_point_count) { m_points.push_back (x); M_points.push_back (y); M_points.push_back (z); } else { int pos = (m_totalcount% 100-1) * 3; M_points[pos] = x; M_points[pos+1] = y; M_POINTS[POS+2] = Z; } M_pointcloud.setpoints (m_points); Render ();}
4 Summary
Slightly
Copyright NOTICE: This article for Bo Master original article, without Bo Master permission not reproduced.
An example of using std::vector to optimize the animation of point clouds