View Full Version : Questions on game modelling

09 September 2010, 09:27 PM
I'm a student in 3D design and animation (first year) and will later on continue with game design as this is what I've always wanted. I don't have much experience in 3D yet, but after only two weeks of school I feel like I've learned a lot and I'm hungry for more and more.

Now onto what this thread is really about: game modelling. While practicing modelling low-poly, several questions have popped up into my mind. And hope that some more experienced modellers can enlighten me!

1. The term low-poly has in recent times confused me. What IS low-poly nowadays. Say God of War 3, or Final Fantasy XIII, I now normal mapping can hide a lot of rough edges, but I wonder - what kinda poly limit is there on modern day games? And what would it have been for say the PS2/GC/Xbox

2. I do have a personal game design project on my hands, and for that I require character models. Is there a general way of doing characters for a game? Say I was to make, call it a base mesh of some kind, maybe one for female and one for male and any other like kids. Is this a good way to go about things?

3. Costumes: Do you model them directly onto the character, or model them seperately then put it on the character and delete anything that is unneccessary?

4. Is there a favorable way of starting a character model? Up until now I've started with an 8-sided cylinder and worked my way around my reference images with that.

That is all for now, I might have more questions later on though.

09 September 2010, 11:21 PM
Hi Waiise

1. In my opinion the term 'low-poly' hasn't applied to game models for PC/High-end consoles for some years now with the exception of games with stylised art like WOW, which is indeed low-poly and very well done at that. The more graphics driven games push ridiculous amounts of polygons around.
I would suggest browsing through Dominance War entrys and regulations, current and archives from past years, and see how the polycount limit changes. In the last one it was up to 10 000 tris max for everything.

2. Character modeling for games, especially these days allowing lots of polys, is not really different from modeling for cinematics. Many people start from some sort of base mesh, sculpt for final overall shape, retopologise for poly limit/reasonable amount below it, sculpt fine detail in and then export relevant maps to use with the retopologised mesh. Difference to cinematics may be just that there you use displacement maps as well to create the fine detail for real at rendertime.

Creating bases meshes to speed up your workflow is definitely sensible. When it comes to paying work, time is money and the client doesn't care how you got to the final. However, as you are on first year, for practises sake I would build models often from scratch, too.

3. I often make a copy of my character, delete non-cloth parts, and model clothes on that. The benefit of modeling from character mesh(copy), in addition to the obvious ready base shape you get from the body, is that if the character had good polygon flows for animation, so will the clothes(unless you change it). Some more elaborate or odd clothing shapes may be easier to model from scratch, but have your character in background as a guide.
As to wether or not to leave parts of a character inside the cloth, I'd say always delete unseen things if they serve no purpose.

4. Up to personal preference, I think. There have been discussions about this on varied modeling forums. I tend to start from a box regardless of where I begin(head or torso), like seen in a head timelapse here ( (timelapse starts at 0:35). Really I think it is up to what you feel most comfortable with and what is the fastest way for you.

- Niko

09 September 2010, 07:45 AM
Thanks a lot for the long reply!

09 September 2010, 09:38 AM
You are welcome. Ask away if you are wondering about something else.

01 January 2011, 10:25 PM
hey folks,i have a question about modeling for games and cannot seem to find a clear answer and i think its because im not sure what exactly to ask.

im modeling a building and it will be explorable without loading times as in when you open the door to this building there won't be a seperate model that serves as "the inside". i've been researching on various techniques of modeling for a game and mostly just find character modeling.and the few building modeling tutorials i've found are of too simple of buildings to really answer my question such as a house with only walls,roof,a door, and some windows. there really is not much tutorial for the techniques game developers use for more detailed building models(aside from the use of normal maps)

so i've heard two things that im not sure about:

1.connect all faces,polygons,etc. and dont overlap

2. low poly!

but,i've seen games with a less graphically inclined engine(mount & blade) with seperate models placed together to make a town building and they overlap so therefore they are not connected and have extra polygons that you dont see,but are there. i apologize that this is lengthy as im having trouble describing what i want to know.

ok so for example, im modeling a building with a wooden frame on the inside and some spaced wood planks on the outside. when i model this building,should i really model this out to be a single model because thats what a game engine requires,or can i make the walls one model and the frame another and in the game editor place them together?

this is such a problem for me because i modeled it all out as one model but after unwrapping the entire model and fitting the UVs in the UV space and rendered the UVW template with a 1024x1024 resolution to import into photoshop but when i make my photoshop map its blurry with bad what to the game developers do?

if you've read all this,thank you for your time!

CGTalk Moderation
01 January 2011, 10:25 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.