The best principle to follow for creating a perfect asset for use in a game engine or game in general are Using reference, Knowing the limits of what you can do and what is needed, Simplification and Topological perfection. Read on for my summary of these four.
Or Alternatively for the website with the four principles in greater detail.
Principle #1 – Model after the reference
All of my models are created using a reference because otherwise it would look off, wrong even but it would definitely be noticeable by someone who is looking for it. If reference is not used it would look like it was thrown together. I always always use many different reference things even though it may not have anything to do with what i am modeling because this way i have different ideas to use or merge or throw away if not needed. Everything can be used for reference, concept art, photos paintings or video.
My way of working would definitely include using these four different principles to create an asset because it just works exactly as planned from the get-go. I have tried to use minimal reference and it is definitely hard to accomplish but it can be done, it should be done with reference which is what happens because it did have to go over it again with some kind of reference to make it better fit how my vision would become.
Principle #2 – Knowing the limits
The limits. These limits are what would hinder the model if gone too far or off the track, a game engine is usually always going to be used since it is a game development pipeline i am pursuing which would mean that it has a limit to what can be used and since that would have to be worked around then a limit is what is best for it.
The many game engines now-a-days are all definitely capable of having a massive amount of polygons and triangles and models in general shown on screen but the game that is being made is always going to have a limit, be it a low poly game for example would have its own limit to what the game engine would have. So there are different limits in the games industry, so for the main game engines and a Triple A title the following engines have limits like the Unreal Engine, The games created on this one are big like really big and they are very detailed which would mean that unreal doesn’t have that many limitations if you were to follow the game pipeline and have lower poly or medium poly assets to use. Unity on the other hand does sound like it is gaining up on Unreal but I don’t think it will overtake unreal in terms of limitations. For me the limitations would include how many assets can be used or the amount of polys can be seen, but for game devs that’s a whole other story.. Cryengine apparently can handle and has everything that unreal has/can do but the lighting is ‘different’ not sure how but it’s definitely not for beginners to use. Other engines that are free are probably great for smaller games but the free ones work well for most game types.
Principle #3 – Simplification
Simplification is pretty much normal maps, or high poly versions of low poly assets, but it takes a different turn, like making a decent looking model and taking it down a notch by cutting certain edges or vertices to make it look cleaner and more simple and then making a higher more detailer model to bake onto the original one. One of the main practises of making an asset for a game is to reduce the amount of detail like cuts, scratches, holes, edges and quads that are created to instead form a texture which has those details on them instead. One mustn’t lose sight of their original sight or image that they had in mind.
Principle #4 – Topological Perfection
This last one is quite intensive with information, but it includes having a nice and clean topology so that the model looks good while still saving on polys, tris. But it can be very different to a cinematic model that is used in movies or even game cinematics.
There are many differences to cinematic models and in between those differences there are problems that one must also look out for when creating any type of model or asset, like N-GONs, Holes, Non planar surface, Non Manifold creation and stack and Intersect Meshing.
To keep it short and simple, an N-GON is a five sided face, in a game a five sided face would cause heaps of problems so it’s always best to have only 3 or 4 sided faces so that it looks good while also keeping the topology good.
Holes are bad, here mustn’t be any holes in a mesh because it causes big big problems like the render not showing anything there, an invisible wall i could say, it’s always best to have the vertices merged at the point where a hole might be. But if when rendering and or making higher poly versions of something that has a hole in it, this would cause heaps of lighting issues and problems that should be fix otherwise the results would not be what you wanted.
The next is non-manifold creation, I’ve never heard of this but it occurs quite a lot throughout my processes and pipelines that i use. One way that this is seen is when three or more faces share a single edge which would cause weird problems when trying to flatten an image into an unwrap or texture. Another problem caused is when two faces share a single vert but no edge, this cause a whole list of wacky do problems when moving stuff or unwrapping too. it’s also bad if two faces are joined, but one of those faces normal was flipped so that there are two faces facing opposite of each other causing the backside of one of the faces to be see through.
Principles of 3D Modelling for Games. (2014). GAME ART Workbook. Retrieved 10 May 2016, from http://blog.gameartworkbook.com/game-art-theory/principles-of-3d-game-art-modelling/