I’m quite new to modelling (starting learning Blender by myself a few months ago in my sparse sparetime) and need some pointers as to how many vertices/faces are acceptable for a game character. I know this depends on the target platform, so for now, lets just say it is for a webplayer. The game I’m working on will not be a massive multiplayer, but will be quite simple with only a few assets on the screen at any time.
I have modelled a monkey (attached a picture) with 1349 vertices and 1354 faces once my mirror modifier will be applyed (info from Blender). If I add a level of subdivision surface this is bumbed up to roughly 5000 vertices/faces - this gives my the smoothness I’m looking for but is this too high?
I know I can bake a normal map from a higher resolution model - but will this help out smoothing the edges as well?
Finally, what is a good target size texture? Should I start high (like 2048x2048), and then scale it down (to say 512x512) once the texture is finished? And would it be best to use only a single material/texture for the entire model?
Hope for some good “rules of thumb” from some of you more experienced modellers. Any critique/advise on the model is also welcome - I’m here to learn ;o)
Thanks for your reply. You are right - most vertices are used on the head. Any one else have any input on the matter? Or anyone have comments on the model? I will continue work on him tonight, so feedback would be welcome.
W01ff: Part of this depends upon your target platform. For all games except mobiles, what you are hearing is true. For mobiles it’s much less. 500 to 1000 is a good target for mobile devices. For older phones (like iPhone 3G) the max viewable was about 7k-10k for everything, so in these cases you really have to squeeze.
The way I approached our current project is the same as all my projects: I model or ask for assets to be “almost perfect” at the edge level (where the edges are smooth enough not to jar the eye) for its purpose.
Then I shoehorn everything else in around that. It has served me well over the years. The monkey above is ridiculously too high in polygon count in my opinion, because you could get away with it looking identical at quarter the polygon count around the edges. But that is unimportant unless you’re aiming to sell the game on mobiles these days, so its a matter of what you are aiming to do with it.
Some numbers: my current ipad game had 40,000 polygons on the screen initially. I went in and decimated everything the user couldn’t see ruthlessly, and got that down to around to 32k. Then, I started culling things which the user can’t see at the same time with some fairly basic states, dropping it to around 10k visible on the screen at one time.
So its never suicide to just model for what looks acceptable. This the mysterious point where the eye is satisfied with the look, and it is just above “that looks too low poly”. Great texturing will transform this medium level polygon into something nobody sane could ever tell the difference from ultra-high poly.
just make a high poly model first, then create alow poly from that, with surface snapping, and bake the normalmap form the high poly onto the low one. And it will look high poly (details and such), but with way less poly’s.
Thanks all of you for the answers! Still learning here - this is my first 3D model I found worthy to be shown to people.
I have been following many tutorials from blendercookie.com, but most of these are not “game ready” characters - hence my question above. I guess it is a tradeof between haveing enough vertices around edges to not disturb the eye (as you say Hippo) and to allow the model to animate properly.
I guess this is all part of a learning process - in time I might be able to create better models. Just a note - I’m a programmer so not trying to reach a level where I can make a living form my models - just to use them in my games.
Actually if you say you’re really a beginner, that monkey is really quite cute and looks well made
I’m also trying to learn how to model things that can be used in games by using practical retopology techniques that I see on tutorials as well as learning C#. I always feel good whenever I see someone who is having success in his/her lessons
I myself have done some 3D Modeling, but I’m more of a programmer. As my only available tool is SketchUp, I model buildings and vehicles for my games. I’ve tried Blender, but could never really understand or pick it up, as there was a lack of tutorials to get me started. SketchUp though, was extremely easy to pick up, and easy to learn. I can make pretty detailed models, and get around 1000 poly’s. My target range for buildings is somewhere around 2k-4k poly’s, but if they have interior details, I’m aiming 5k-7k. A lot of the details are actually done using 2D textures. For bricks, wood carvings, and little details that would be too expensive to model, I would just use a texture. It all depends on what you’re modeling, and you’re target platform. It would be great to learn Blender, but I haven’t found enough tutorials to get me as far as designing the basic shape of plane. And when I say basic, I mean VERY basic :(.
I would also take a look at Blizzard, there is a “model viewer” that lets you see the models and export them. By doing this you can see how many polys you character has (as long as its not a mop character, they arnt on the expansion but there working on it) and see how they made it. This can help new comers get to see how a “professional” company makes game models.
