I wonder what exactly happens if i draw a scene onto a 1x1 Framebuffer.
Imagine the following Scene:
(a ) shows the scene in 3D : the blue triangle is nearer to the eye point.
(b ) shows the scene just before rasterization (after viewpoint transformation)
My Question: what is the outcome.
Though the image middle doesn't hit any of the triangles, the blue triangle is drawn. Sound weird to me, but i read the following Article, which i understood saying:
If you render an object to a 1x1 Render-Target, your depth buffer contains the depth value of the nearest of its vertices. This would mean, that ALL Vertices would (try to) write to the depth buffer, no matter whether they exactly hit the pixel-middle.
Citation: "Since the nearest point always trumps the points that render behind it, the depth buffer of our 1x1 render target should hold the nearest Z value. "
The Framebuffer stays empty, because no object hits the pixel middle.
In http://www.gamedev.n...ost__p__4208505 it is said, that this way you can implement a 3D-Picker.
Did i get it right, that this is achieved if the viewing direction is aligned with the picking ray?! So the object that is nearest to the viewer AND hits the pixel Middle would be the last one to write to the depth (and so to the frame)buffer.
Which (if any) of these ideas is right? Can someone quickly explain why, considering the rasterization process?