Hi,

Excuse me but I again have problems with OpenGL.

1) I cannot get correct result from method EntitySelector.GetEntitiesInRectangle() but method EntitySelector.GetClosestEntities() works correct. I create Rectangle2D by using mouse then transforms points of rectange how you show me in example then pass this Rectangle2D to EntitySelector.GetEntitiesInRectangle() but method returns IList<IList<DxfEntity>> closestEntities.Count = 0. I don't undestand why method EntitySelector.GetClosestEntities() works correct and method EntitySelector.GetEntitiesInRectangle() not correct. Can you check this in your example and answer me about results?

2) I cannot set white background in OpenGL? How I can do it?

Hi,

1) Make sure the rectangle's first coordinate is the bottom left, and the second coordinate is the top right.
2) Use e.g. GL.ClearColor(0.2f, 0.7f, 0f, 0f); to change the background color where OpenGL is initialized (see the OnHandleCreated override).

Wout

Hi,

Thanks for your anwer. It's really work but I have question for you.
How calculate real CAD point's coordinates (DxfModel points coordinate) from mouse click point in OpenGL? In GDI I use next calculations
Point3D pointInScreenSpace = new Point3D(mousePoint.X, mousePoint.Y, 0d);
Point3D pointInModelSpace = to2TTransform.GetInverse().Transform(pointInScreenSpace);
How about same calculations in OpenGL?

It is the same with OpenGL, with the minor modification that the screen point must now be defined in clip space (so normalized to -1, 1 as you have seen already). Alternatively you could correct for the viewport transformation in stead of normalizing the screen point, as I've explained before. Note that you should be using a orthographic transformation (not perspective).

Wout

Sorry but I cannot get correct result from calculating real CAD point's coordinates (DxfModel points coordinate) from mouse click point in OpenGL. I am using next code:

Matrix4D modelViewTransform = GetModelViewTransform();
Matrix4D projectionTransform = GetProjectionTransform();
double clientWidth = (double)ClientSize.Width;
double clientHeight = (double)ClientSize.Height;
Point2D startPoint = new Point2D((double)mouseClickLocation.X / clientWidth * 2d - 1d,
   -(double)mouseClickLocation.Y / clientHeight * 2d + 1d);
Point2D endPoint = new Point2D((double)e.X / clientWidth * 2d - 1d,
   -(double)e.Y / clientHeight * 2d + 1d);
DxfPolyline2D createdPolyline = MathDxf.ConvertToPolyline(
   new Rectangle2D(startPoint.X, startPoint.Y, endPoint.X, endPoint.Y),
   modelViewTransform * projectionTransform);

public static DxfPolyline2D ConvertToPolyline(Rectangle2D rectangle, Matrix4D matrix)
    {
      Matrix4D normalizedMatrix = Matrix4D.Identity;    
      matrix = matrix * normalizedMatrix;      
      Point3D point1 = new Point3D(rectangle.Corner1.X, rectangle.Corner1.Y, 0d);
      point1 = matrix.GetInverse().Transform(point1);   
      Point3D point2 = new Point3D(rectangle.Corner1.X, rectangle.Corner2.Y, 0d);
      point2 = matrix.GetInverse().Transform(point2);  
      Point3D point3 = new Point3D(rectangle.Corner2.X, rectangle.Corner2.Y, 0d);
      point3 = matrix.GetInverse().Transform(point3);   
      Point3D point4 = new Point3D(rectangle.Corner2.X, rectangle.Corner1.Y, 0d);
      point4 = matrix.GetInverse().Transform(point4);    

      DxfPolyline2D resultPolyline = new DxfPolyline2D();
      resultPolyline.Closed = true;
      resultPolyline.Vertices.Add(point1.X, point1.Y);
      resultPolyline.Vertices.Add(point2.X, point2.Y);
      resultPolyline.Vertices.Add(point3.X, point3.Y);
      resultPolyline.Vertices.Add(point4.X, point4.Y);
      return resultPolyline;
    }

Can you say me what I've wrote wrong?

Hi,

Looks like you have reversed the order of model view and projection transform. The projection transform takes place last, which means it has to be first in the matrix multiplication:

DxfPolyline2D createdPolyline = MathDxf.ConvertToPolyline(
   new Rectangle2D(startPoint.X, startPoint.Y, endPoint.X, endPoint.Y),
   projectionTransform * modelViewTransform
);

Also I suggest calculating the inverse matrix just once in stead of for each point again, because it's a fairly expensive operation. Probably you can cache it somewhere.

Wout

Thanks I think order of matrix not importent. :)