Game Development: Background information

Firstly when creating game assets, main thing in mind is that you use as much or few as needed to get the result you want. I know this sounds copout answer, so i'll elaborate a bit.

you'll have to answer these question aswell. What is platform you are planning to make, xbox/ps4/mobile or pc and on pc is it high end/low end. granted both unity and ue4 has tools to make the optimization for each platform. however you still need to know this to reduce the overheard on creating assets. if you target mobile, there is no reason to make extremely highpoly meshes to be baked on lowpolys that you would even see the details. and when going for consoles there are limitations on various things to get performance and that the game runs on 30fps minimum. like 2k textures on consoles is max, on pc:s you can use 4k.

1. Vertice count is always hot potato, it depends. it really does, some games have relatively high poly count on their chars just cause its only one in the screen all the time. then there is games like wow where you have to have lot less to have that 200 of them in screen plus the enviroment. so use as much you need to achieve 30fps. 

2. you don't texture the highpoly, what you bake from high to low is normal data, and paint the lowpoly. the normal data drives everything else.

3. all game engines changes the meshes to triangles when importing, its just how the renderer works. so you don't need to worry about that, just have good topology and the engine will handle it.

4. I'd recommend you download unreal engine 4 and check their third person template, it has everything you need to know how skeleton are built for that. I dont think unity's differ that much.

5. consoles 2k, pc.s what you can throw them. bit is something that depends on the project aswell.

i'll see later today when i get home if i had some links to good game asset tutorials. the cgcookie rifle is good one for example.

hope this helps. :)

1. How many vertices should a character have at maximum when I'm planing of haven very packed scenes where about 200++ of them appear at the same time?

When it comes to game assets a general rule is to provide the minimum amount of geometry needed to create the look you want. There is no specific amount of vertices you should aim for as it's highly dependent on how that model should look. Keeping verts (and tri's really) low should be considered, but not at the cost of creating a good looking game model.

2. If I want to make these characters good looking would I model out a high-poly model, texture the high poly model and then bake the normals to the low-poly one and apply the high-poly texture on it?

Yes you can do that. A lot of times people will create a high poly version to create all the unique details and overall look, then retopologize or recreate that same asset with much lower poly count. The high poly model gets baked and applied to the low poly model.

3. What topology is preferred for animated game assets? I've read that triangles are the best for game assets since the quads will be translated to triangles anyway, but on the other hand quads are better for animating. can somebody point this out to me?

Quads get turned into triangles when you bring it into a game engine.So I would just stick with quads in Blender when animating.

4. When it comes to animating the assets, how complex should the rig be to make some good looking movements?

This is mostly based on what you're trying to achieve. What is the minimum rig needed to make the animations you want?

5. How big is the texture allowed to be? What color-depth is to preferre? 8-bit or 16-bit? 

I'm not sure what texture you're referring to, but I'll assume you mean any texture. Similar to the "min amount of geometry", with textures it's a good idea to pack in as much information into one larger texture atlas than to have many small textures separately. I talk a bit about the process in this article: https://cgcookie.com/articles/maximizing-your-unity-games-performance 

Another common thing to do with textures is to build trim sheets: https://www.youtube.com/watch?v=IziIY674NAw&t=662s  Anything you can do to reduce the amount of textures you use separately will be beneficial.

you can check out polygon academy channel, if you are planning for ue4. though it does apply for unity aswell, but he is using ue4.

https://www.youtube.com/channel/UCGXr6E_g91ue1rfhA9j4TLA 

Hey guys,

your answers halp me out a lot! But I'm still pretty unsure about the look of an RTS-character. Do you guys think this one would be possible with a retopology or would that probably still be too high-poly?

Will I have to go as low poly as this one?

                                                source: https://blenderartists.org/t/kangaroo-warrior-low-poly-game-character/556119

that depends on few things.

How far you are keeping the camera? that pretty much defines how much you need polygon versus texture detail. for example the xcom series even though is not rts, the character are high quality but for the reason camera gets zoomed in and out. if the characters are far, then there is no point putting in details no one is going to see, just the hint of details.

looking at the two pictures, you could propably manage to bake the details from the upper one to the lower one, when we are talkin just polygon count and textures. again though, it depends on the fact what you are showin to player.

louhikarme   I forgot to mention that I am targeting PC as platform.
For the camera zoom I plan to be able to go pretty close and far away. To get things along I want to use LoDs. The retopologized version of the picture above would be the zoomed in version. The zoomed out versions should be decimated versions (with blender decimate modifier) from the retopologized picture.

are using unity or ue4? because if ue4, then you don't have to worry about lods that much since they now have automatic lod system implemented for skeletal meshes. and it works pretty damn good. the sniper rifle that i'm working on currently is using skeletal meshes due the animations and easier usage with other skeleton. and i used ue4:s automatic lod system on it and it work brilliantly. and while i say automatic, you ofcourse need to make the parameters for the decimation settings when they kick in and how it will decimate. but you don't have to manually do that in blender and export/import multiple files. just create the mesh that is closeup. and then let the tool worry rest.

if its unity, then i don't know. maybe jgonzalez can answer that one?