question about high res to low res for mobile games

kmccarthy197

Hello, I'm new to the forums here but have recently got back into 3d modeling and had a couple questions about the best way to take a high res model in zbrush and create a low poly model suitable for a mobile game using the unity engine.

Is the best workflow to take your high res mesh and retopologize it to create a "game res" mesh and bake normal maps onto it? Also, can anyone give me a rough range of what a good poly count is to stay between for something being optimized for android and ios graphics?

Ged

I have to say I havent seen many mobile games actually use normal maps. Retopologising the high res would be a good plan but you will have to make a seriously low poly model depending on your game, a whole mobile environement including all models can usually weigh in around 8000-10000 triangles. So just as a hypothetical idea... if you only have one or 2 characters on a simple background you can go nuts and have each character 3000-5000 tris but if you have 10 characters maybe about 300-800 tris

kmccarthy197

I was thinking something along the lines of Shadowgun for android (possibly ios, i'm not sure). Unity 3.5 has some pretty awesome rendering capabilities for mobile games with lightmapping, occlusion culling and the new addition of light probes. I've actually just found ben mathis' portfolio and seeing the topology of his wireframes he designed for ios gives me a good idea of what kind of resolution I can work with as far as the low poly models.

As for baking a normal map, just a question: Is it better to try and build the low poly model within the high poly for the purpose of baking the texture or does it not matter? The best way I can explain this is to use the example of a high poly head in zbrush. When I retopologize a low poly model onto it sometimes the edges will intersect the geometry of the high poly. This is especially noticable on the rounded top of the head where I am trying to save polygons, where the vertices are lined up with the high res model but the edges from them intersect the rounded parts of the head to where they are slightly inside the model. Is this ok?

Thanks for the help so far, and sorry for the long winded response, just trying to explain things the best I can

Ged

Yeah we make mobile games in unity at work, it is powerful but those big features like normal maps and light probes do hit performance quite heavily. You would have to be very careful how you optimise your game if you are using per-pixel lights and normal map shaders. You can use unity vertex lighting and lightmaps and have a very beautiful game and not use any of those features and it will run well without much pain.

I think shadowgun took 6 months to a year to make? and they had unity engine developers helping them.

kmccarthy197

It sounds like the best thing for me to do might be to model all the low poly stuff in maya and then detail it and bake out the normal maps if I decide to scale the game up for the newer high end mobile devices or possibly different platforms

Goeddy

on the projects i´ve worked on i recognized, you often dont see the influence of the normal maps, escpecialy on phones with theyr small screens. so i wouldnt realy recommend that to you.
also normal maps work best on a mesh that has an even topology and matches the shape, and you will most likely not have the topology to achieve this.
in the end it realy depends on the game you are making.
for a 2D fighting game it might work, but with almost all other types of games you will run into walls, performancewise.