Hello,
One of my friends recently went to apply for a game company and was shot down because they said he needed to model using subdivision surfaces. We are both at the same university (he just graduated and I have 2 years to go) and our 3D modeling course trained us to model in maya and 3DS max using polygons, and we also had a very brief training in NURBS. After hearing about his interview, I went out on the internet to search for documentation on subdiv surfaces. Everything I found said that subdivision surfaces weren't really plausible for games because it takes such a great amount of cpu for them to process. I've only been modeling for about a year, and a lot of this stuff still goes way over my head. Could anyone possibly explain to me the current presence of sub d's in games and why a game company would say that it was the best way to model? Also, if I were to use them in a game, how would I go about determining a sort of polycount in terms of how many my game could handle. Would it be best to have a balance of both polygons and sub d's?
Thanks for your help!
Replies
People do not use sub-d's for the final low poly model.
You need to be able to High poly model in sub-d's. So you can bake normal and AO maps from that.
here's a sub-d highpoly model:
RH_roboTrophy_render051310 by sprunghunt, on Flickr
and here's the lowpoly:
so when the company says you should learn how to create subdivision surfaces, they want you to know how to create subdivision surfaces (high poly models) to bake normal maps for your low poly model, not game ready subdivision meshes.
Not to mock you (you seem to have been let down by your teachers), but this is elementary knowledge for an artist these days. I see you have two more years left at this university, so I suggest you post here often and don't rely on your teachers to prepare you for the industry -odds are they've never worked in the industry themselves, or too long ago.
I have to ask what school are you guys going to that doesn't teach you sub-d modeling?
http://en.wikipedia.org/wiki/Subdivision_surface
any surface that is smoothed and tesselated at the same time is usually referred to as subdivisional surfaces.
what they mean is highpoly modeling that isn't created using zbrush or nurbs.
A proper subdivisional surface allows you to selectively tesselate areas for more detail and harden and soften edges without changing the base mesh - but few people actually use these features.
It really depends on what you're modeling. Generally mechanical stuff (like the robot I posted) is quicker to do with subdiv surfaces. But you might want to mix and match techniques.
Definitely learn both subdiv and zbrush if you can.
but its crazy, that I notice as bad as some "TRADE" schools are, the universities can sometimes be far worst.
I think they deal with a lot more theory at times and theri ntry level 3d class / 101 course teaches them so broad because you have people taking that course as an elective sometimes and not as a cg major quite often.
While trade schools move fast at least the teachers know going into the class 90% of those people will stay in the cg major for a while.
while a university has to sometimes teach broad until you take their master courses.
For instance, here in Dallas, SMU Guildhall is a great place to go....if you getting your masters. But i hear the minor course are not enough at all.
Its just really funny I think either way. If I was going to school now I would do gnomon of vfx phd, something more online and subscription base schooling, and community college.
Save money and invest in my hardware
http://wiki.polycount.com/SubdivisionSurfaceModeling