Next-Gen game developing?

FEMTO

Could someone be kind enough to explain to me what are the key considerations one must take into account when designing a model for a next-gen game. I really don’t know a great deal about next-gen modelling, and really want to get started on modelling some stuff. Are there any tutorials on the net that explain Next-Gen modelling?

Anyway, any help given is much appreciated and if you are unable to help then thanks for taking the time to read my post!

Dio

Im yet not professional but I have to imagine normal mapping will have the largest effect on next gen model development. As a consequence high poly modeling is an important consideration.

Daz

Next gen is such a nebulous term. I assume you mean models for the Playstation 3 and Xbox360? Well, it obviously totally depends on the kind of game the asset is headed for.

If it's a character for a 3rd or 1st person view game, a good place to start looking for how it's currently being done by those that know is here:

http://www.unrealtechnology.com/flash/technology/ue30.shtml

Scroll down to the section that starts 'Distributed Computing Normal Map Generation Tool' and there is some info and polycounts there.

Other stuff to consider might be:

Dont paint light information into your texture maps anymore. Split them up into colour, normal, specular and perhaps bump and more. Oh and er, make them bigger ;-)

Joshua Stubbles

One thing I'm still unsure of, is the normal mapping setup. I mean, it's pretty straight forward I think, until I get to multiple object meshes.
Lets say you have an old wooden wagon, which the wheels needs to rotate on. Since these need to be seperate objects, what's the best way to set this up for normal mapping (in 3dsmax)?
Setup the RTT with multiple objects, all ready to go with cage projections? Or do one at a time?
I understand how to make the normal maps, just not in every circumstance.

MoP

You could do either of those routes, Vassago.
I'd maybe either move the pieces far apart (low and high-poly separate objects) but keep them all attached together, and use that setup to render out the normal-map all in one go.

I have in the past rendered out different segments and put them together using Photoshop, that works too, and means if you have any problems in one object, you don't have to re-render out the whole map.

I'd just say trial and error, see which routes work best for you.

Eric Chadwick

Vassago -- We do one at a time... in your wagon case that would mean dissecting the wagon into its repeatable parts (only need to do one hipoly wheel since you can reuse its normal map for both wheels, etc.). Then model those repeatable (or tilable) hipoly elements (and their ingame counterparts) as separate entities. Then cast each normal map separately, tuning as needed. Then combine them into as few bitmap sheets as possible (less is usually better).

FEMTO -- Lots of good advice on this board, just have to poke around a bit. For example, here's a link I bookmarked awhile back.
Blending Hi-poly Normal Maps with 2D Normal Maps

Eric Chadwick

Oh also, to add to Daz's post, I recall reading some more good info about UE3 from Rorshach...
Multiple positions open at Epic
(just stay on topic by not posting any tech questions there)

arshlevon

you can also render all the normals at once.. by slaping mat IDs on the seperate pieces(high and low)..there is an option on render to texture to only hit matching mat ids.. so for all the parts of the wheel on the highpoly you can give it a mat ID of 1 and the same for the low.. and give the wagon a mat id of 2 .. and when you render you dont have to worry about intersecting geometry because if it dosent share a mat id then it is ignored..

MoP

Ah, very good advice Arsh, I hadn't thought of using that!

FEMTO

Cool...thanks for all the info everyone!