1 Replies - 2167 Views - Last Post: 24 June 2014 - 01:25 PM Rate Topic: -----

#1 marcusraz   User is offline

  • New D.I.C Head

Reputation: 0
  • View blog
  • Posts: 2
  • Joined: 24-June 09

orthographic projection

Posted 23 June 2014 - 03:30 PM

I'm trying to figure out orthographic projection. I am not using open GL or DX just VS C++

once I pro-ject the object onto the viewcube/view volume what do I do next to project
the side and/or top views onto the 2D screen?

Posted Image

The front view seems straight forward do you just set the Z vertices
to 0 and then draw on the computer screen?? But the side, back top need to be projected? rotated to the orientation
of the front view?

Do I need to 'unfold" the view cube and project the sideview to the same plane at the front view? if so do I just use a rotation matrix and simply rotate the side 90degree? and then project that on the 2d screen?

Posted Image

what is the preferred method

Is This A Good Question/Topic? 0
  • +

Replies To: orthographic projection

#2 BBeck   User is offline

  • Here to help.
  • member icon

Reputation: 792
  • View blog
  • Posts: 1,886
  • Joined: 24-April 12

Re: orthographic projection

Posted 24 June 2014 - 01:25 PM

I would presume you would use separate view matrices. Optionally, you could spin the world for each direction and use the same camera (actually that's what a view matrix does).

But if you have 3 separate camera positions, you should have 3 separate view cubes. And so you would have to do a separate projection for each.

The drawings are misleading and not what I would consider a "view cube".

Let's go with what's in the drawings. You have 6 planes you are pretending form a cube. They don't have to touch one another and therefore do not by their nature form a cube. They're just six mutually perpendicular 2D rectangles in 3D space that live on axis aligned planes.

Anyway, you can form them into a cube and project along every dimension of every axis and capture an orthographic projection of each. Then you can do whatever you like with those texture/surfaces. You could rotate them to all be on the same plane, for example, which is similar to what they are doing in the drawing.

In reality, you would project onto a render target along the given axis. The render target has an "imagined" position in the 3D space, but it's really just a memory buffer. If you were to draw that to the screen, you could texture it onto quad and position the quad into the scene any way you like.

If you're simply removing the Z value on the X,Y plane, then it stands to reason you would simply ignore the Y value on X,Z plane and ignore the X value on the Y,Z plane.
Was This Post Helpful? 0
  • +
  • -

Page 1 of 1