Screen to world transformation

[sumitkulkarni23]

Hi,
I am working on a sample application, where I have embedded a Mogre window inside WPF. I am trying to draw a interactive line on the canvas there. In this I am facing a problem where I am not able to convert the selected screen co-ordiantes to world co ordinates. I tried lot of things to perform this conversion (including the hints given in other forums) but its not helping. Does Mogre have any API whcih does this for me ? I simply want to convert the mouse positions which are in pixel units (ex. 176, 174 etc) to world co - ordinates. The attachment shows the problem.

shows the screen to world transform not working

screenToworld.JPG (2.18 KiB) Viewed 1269 times

Regards
Sumit

[syedhs]

I think you can use Camera::getCameraToViewportRay. For an example of this, look at Mouse Picking Wiki Tutorial

[sumitkulkarni23]

Thanks for help.
I tried that, basically the "result" contains world geometries. it has some thing in X, Y, Z but those values are not useful.
It doesn't show me a generic method for converting the pixel co-ordinates to world/object ?
I need those methods. is there anything like that in OGRE/MOGRE ?

Regards
Sumit

[syedhs]

Actually we are not going to do mouse-picking, the sample just show the usage. What you are interested is to pick up the 3d coordinate based on the 2d position in the viewport. So it means that you need to pick up the end point of the ray, which is okay not possible just by calling the function because it just returns the origin and direction. But looking at the source code (simplified):-

Code: Select all

void Camera::getCameraToViewportRay(Real screenX, Real screenY, Ray* outRay) const
{
		Matrix4 inverseVP = (getProjectionMatrix() * getViewMatrix(true)).inverse();

		Real nx = (2.0f * screenX) - 1.0f;
		Real ny = 1.0f - (2.0f * screenY);
		Vector3 midPoint (nx, ny,  0.0f);

		// Get ray origin and ray target on near plane in world space
		Vector3 rayOrigin;
		
		rayTarget = inverseVP * midPoint;

}

What you want is rayTarget variable. So from here, it is not that hard to tailor your own function.

[sumitkulkarni23]

Hi Syedhs,
Thanks a lot for your help. I tried the second approach as suggested by you, but still I am not getting the correct results.
Just to highlight a point, I was reading the documentation and it says getCameraToViewportRay needs "normalised screen coordinates [0,1]"
which I don't have ! I have Pixel screen co-ordinates. Does that make any differance ?

Regards
Sumit

[syedhs]

Normalised in this case means mousex/windows_width and similarly, mousey/windows_height. So both parameter should not exceed 1.0.

[sumitkulkarni23]

Thanks syedhs,
I was able to solve the problem with your suggestions.Here is the code of mine which works.

Code: Select all

public Point Unproject(float screenX, float screenY)
      {
          // Actual width of Viewport - I have set my camera(mCam) as viewPort
          float scrx = screenX / mVPort.ActualWidth; 
          float scry = screenY / mVPort.ActualHeight;
          Ray ray = mCam.GetCameraToViewportRay((float)(scrx), (float)(scry));
          Vector3 vect = ray.GetPoint(mCam.NearClipDistance);
          Point ret = new Point(vect.x, vect.y);
          return ret;

      }

Earlier I was doing the same thing without normalizing the screenX/Y.
Thanks for all your help once again !

Regards
Sumit