How far does low polly objects/character reach ? By this i mean how many polygons does a object have to have while still being a low polly mesh. Is there a ration like if you have 5million polly’s then your mesh should be 3% of that ? I’v gotten into problems while animating because i simply had far to little polly’s and the animations turned out to be horrible.
It depends, because as video cards get faster, the threshold between what is considered low and high poly changes. I guess the deciding factor is if it will bog your game down or not.
low poly changes every few years because of industry standards, but to break it down some
the 2003 era is usually 500 or below, the 2007 era is something like 1200 or below, 2009 era models are 4000-8000 polys, thats what iv seen in the game industry. as far as animations go, i couldnt tell you
Low poly is a style rather than a specification.
A decade ago, low poly modelling basically had nothing to do with creating a model with a delibareate low polygon count. You just had a very low overall budget and had to make the best geometry possible with the least polygons possible to meet the required detail.
Today the overall polygon budget is so high you can make a model with a high polygon count “wasting resources” while you could make a model looking quite as good with fewer polygons as well.
Nowaday lowpoly’s coming again with OpenGL ES and all the mobile platforms though.
A rule of thumb is that real low poly stuff usually is triangulated, while high poly stuff and hybrid stuff (lowish poly with sub-d) is generally made of quads because dubdivision isn´t working properly with tris.
Your base mesh usually should always be low poly and quadbased.
If you need the model for realtime/gameengine, you optimize the topology towards triangles, but still can sub-d the quadbased basemodel for instance for rendered cutscenes.
If you need it for animation you keep it quadbased to be able to add a sub-d modifier and start cutting new loops for detail, as much as you need or as is required.
As you still got the base mesh you can always decide to bake high poly detail on a bump/normalmap and apply it to a lower detail version.
And with the computers getting more and more powerful todays approach usually is sculpting.
You create a basemesh, sub-d it, or with the new adaptive subdivision technology you just create anything with the worst topology ever to look like your basemodel, sculpt the detail and later retopo the whole model to a basemesh of your desired level of detail.
It´s always a matter of requirements:
If you need a male human for a mobile phone game and know you’ll never make an animation with him you can start off with a triangulated low poly mesh.
If you might need it for an advertising video as well, maybe make the base mesh quadbased, trim down the geometry and optimize topology for triangles for realtime usage, and simply apply a sub-d for the animation.
If you’re creating a main character targeting all markets you’ll most likely whip up some base mesh, sculpt a high detail version and retopo the mesh. You’d use the retopo’d mesh with a baked map for the PC/Console platform, use the same mesh subdivided for a CG animation, and take the retopo’d mesh without normal/bumpmaps just baked down to a static lighting texture map and cut down the polycount and optimize for triangles again to use it in a mobile platform spinout game.
http://www.polycount.com/forum/showthread.php?t=41232
good reference for low poly models