PDA

View Full Version : Modeling for the camera shots, is it necessary?

redwhitejacket
10-04-2009, 06:27 PM
From what I've heard, you model for your camera shots. So if you don't show the entire building, you don't model the entire building. If a model is only to be seen at a distance, you either model very low poly or you rely more on texture mapping if you can get away with it. If you're not going to show the back of a couch, or computer or tree or whatever model, you don't model it.

And the point for all of this is to save on render time. However, doesn't that essentially mean that those models are only good for those shots? So if you have a different scene or animation that needs those models, and shoots them from a different angle, you you have to edit the models, adding the parts you'll see and deleting the parts you want in order to save on render time.

However, I'm not exactly certain why the renderer doesn't take on the task of determining relevant polys up front. That is, couldn't you run a "pre-render" in which the renderer maps out which polys will be visible primarily, in shadow, in reflections, in refractions, etc. and then stores that information to speed up later renders by ignoring all other polys in the scene?

I guess what I'm asking is, why is this the burden of the modeler and not the renderer? Or am I missing something? Thanks.
eldee
10-04-2009, 06:54 PM
you've got it backwards.. the render already does frustum culling and ignores geometry that is completely obscured by another object (for primary rays anyway). The "model for the camera" concept is for the benefit of the modeler and the scene assembler (or layout artist, the title varies). If you're working with a very dense scene, it starts becoming very difficult to manage in your 3D package.. viewport slowdowns, etc. Additionally, knowing what the camera is going to be looking at saves you from modeling the back side of a building, or say, the bottom of a car.
redwhitejacket
10-04-2009, 07:30 PM
you've got it backwards.. the render already does frustum culling and ignores geometry that is completely obscured by another object (for primary rays anyway). The "model for the camera" concept is for the benefit of the modeler and the scene assembler (or layout artist, the title varies). If you're working with a very dense scene, it starts becoming very difficult to manage in your 3D package.. viewport slowdowns, etc. Additionally, knowing what the camera is going to be looking at saves you from modeling the back side of a building, or say, the bottom of a car.

Hmm...ok, well if that's the reason, Maya already provides bounding box shading and points shading (if wireframe and other shading is too much). They also have backface culling, and you can choose the type of shading method used on a layer basis, so you smooth shade one layer, and bounding box shade another layer containing more complex geometry.

The times I've heard this topic discussed, it had to do with the renderer.
Mahje
10-10-2009, 11:43 AM
It's all about saving time and staying efficient, why model the backside of something if u know it won't show anyway?
musashidan
10-10-2009, 12:10 PM
The straightforward answer is: If assets are for a one off shot,model/texture what you need for the shot. If assets are to be reusable for multiple shots/scenes/angles,complete the modeling/texturing.
CGTalk Moderation
10-10-2009, 12:10 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.