This is probably a really stupid question but here goes: The typical example for a view frustrum is 90 deg field of view pointing down the z axis. This makes clipping very easy. So in this situation is it standard practise to not move the field of view from pointing down the z axis? So basically you just transform your model and camera coordinates so your viewable area is in that frustrum? This sounds to simple to be the case. So how is it really done?

I''ve only done a little work with software renderers, but here are a couple of thoughts. You''re right about clipping being easier in view space, and it would definitely work to transform all your geometry into view space and go from there. The problem is that you will probably want to do frustum culling first, where you eliminate geometry not in the view frustum and therefore avoid processing it. To do this, you''ll need your frustum in world space first, to carry out your culling algorithm (octree, BSP, etc.). Once you''ve identified the geometry you''re going to render, then you can go ahead and transform it, and do all the clipping in view space.

I think that would be a reasonable approach, but again, I''m no expert on software rendering.

Actually you''re very close to the truth...

I''ve explained a few things in a recent thread: http://www.gamedev.net/community/forums/topic.asp?topic_id=186128

Thanks all!