I've been trying to model very low polygon figures that can work in Poser (Stop moaning.), or be exported to a game engine. >5000 Polygons.
The trouble is I've used to working mostly in quads and avoiding Tri's like the plauge, but game engines seem to require only tri's.
A 2000 polygon figure istantly becomes 4000 Tri's the moment you want to make it game compatible.
Why is it that Poser Prefers Quads while game engines need tris again, while the Game engines prefer tri's? I know it has something to do with shading problems.
Are there any game engine that accept quads at this point?
Would be people interested in game development who DO use Poser prefer a low quad OBJ figure, or would you rather have it tris in Poser?
Replies
your poser model will be turned into triangles as well when you render it.
eventually people model quad or tris whatever they like, and then on export or import by engine whatever its turned to tris anyway...
I would stick to whatever is easier to work with for you. after all the internal representation might differ depending on each engine anyway.
All you need to do is find a good limit for your work be it tris or quads and check if it doesnt explode your target engine
if you have just quads, you can subdivide them cleaner (there is more algos based on quads), hence quads are preferred for most modelling these days.
Subdivision is primarily only important when doing high-polygon work (for prerendered work, or normal mapping).
Some low poly artists (myself included) stick to modeling in Quads for 2 reasons:
1: Its good practice still for the high-poly work
2: It looks cleaner when viewing the wireframe.
There are a number of tools out there to count the number of triangles in your model, even if youre modeling in Quads.
Not knowing what application you're using, I can't tell you which tool to download or enable, however. Ultimately all your Quads are really just made up of 2 triangles anyway.
When somebody mentions the polygon budget of a model they ALWAYS ALWAYS mean triangles.
So, the answer to your question is thus. Model how you feel comfortable, but realize that its the triangle count only that matters(some engines the Vertex count is important though)
Also, some engines do support quads, but really only as a optimization method, the surface always gets rendered as tris like CB says. When I worked on Star Fleet Command 2, the programmers requested we use quads wherever possible. But that's old school, most engines these days just use tris.
Another tidbit. When people ask for the polygon count, they're really asking for the triangle count since that's what (partly) matters when the model hits the card.
A cool couple articles on my site that might be helpful for you.
I shall C&P from something I wrote:
Computers and games consoles draw tris by plotting 3 vertices in turn. Once you have those 3 vertices plotted in 3d space, you simply connect the dots and you have a polygon.
. . . . : = polygon
Once you have 1 polygon drawn, you need to draw the next one. But unlike the first triangle, you dont need to plot 3 points and connect the edges.
Why not? Well, generally triangles are next to one another You have drawn one triangle, so for the next triangle you already have 2 of its vertices plotted, so you simply add the third vert and connect the dots.
<| <| . <|>
For the computation of one vert you have a new poly. For most of the model this holds true, making them very quick to compute. This method is often referred to as Triangle Strips. The method works best when you have a neat mesh, with rows of triangles - something that comes naturally when you work with quads. When you dont model in clean rows, you create orphan triangles which slow down the triangle stripping algorithm.
I'm using Lightwave 7.5 for most of my modeling work, and trying to get into Zbrush for my texture and Normal/Displace mapping.