maya help is good enough. But consider this:
Take a car, its manual isnt realy about how youd fix the car in a real production environment, nor does it specificaly go into HOW you should drive a car where whan and so on even tough nuch work is being done to make it fill even this need. What it does is it goes deeply into how the underlying stuff works is supposed to work and how its specified. The rest is up tp the customer.
Take a simple thing like hammer, write a manual on all that you can apply a hammer on and all tips and tricks... Youll habve VERY thick book. Now recheck what the manual should do! ist shoud give you the tools to solve problems YOU understand not tell to you how to solve them, because itd be so thorough youd fall asleep halfway across the search. Also maya would not jsut be more expesive it owuld be FENOMENALY expencive as itd teach everything about everything.
Now lets take a look on mayas manual, sure there are few tuorials on top. But theres nothing too specific. There are two reasons for this. fisrt a general example is often more productive than a very specific one. Second the developpers of maya have3 absolutely NO idea what the other end of the pipe is doing, so for games developpers thais means that alias does not tell its customers HOW they should do stuff. they are free to do it as they wish. Other progs manuyuals arent any better in the end.
Now herein lies most beginners problem, all the etchnology is built on top of other technology thats OBSCURE for the casual observer. The fact is that no individual piece of this puzzle is inintself a very complex thing, but the combination of them is quite hard to find. FURTHERMORE no such resource that can completely tell you what to do next exsists any where on the planet. Sure theres a lot of information, but the thing is that its never dstillied in quit eth eway YOUD want tit to be. Also most of the time even pros wont know what to do with a piece of the puzzle untill they come across it.
Younger people have this notion of how things would work out in perfection, but reality is not perfect. Infact theres no such thing as best, theres just adequate and slightly better. Even thesn this metric iis only defined as a optimisation around certain parameters. Something thats brilliant in one context WILL suck in another.
So now you need to fill your brain with knowlege fully avarae of the fact that everything will not ever be filled in for you because ist silent knowlege... not meaning somebody cant know it but it cant be passed on to others ecept trough experience.
OK now on prelighting... this too is a optimisation issue. Firtsof theres relay no need of highress modells unless your game model indeed is a representation of something thet IS a higher res model in the end, thetway atleast you could fool the audience better. Not so mouch about the computer needing this but you the user might not have the capability to see all the kinds of stuff intuitively and paint/fake them in.
Now wich is better is hard to say, infact ists allmost impossible untill you can tell yourself how the eigene your moving the stuff into works. IN general things like occlusion, radiance and fg are well suited for, vertex coloring, this is because the vertex arrays better describe change in tropology and denser areas in tropology change theese sorts of calculations value more. Thet is notto say you couldnt use amap for this. BU the thing is you realy want to avoid using maps for this if you can get by with vertex coloring.
Texture cant be realy gotten across with vertex coloring UNLESS the vertex mesh is very dense... wich defeats the purpose. Infact maps are fakes bacause we cant do it the optimal way around.
And failure allready covered specular. Wich needs to be handled by the realtime eingene itself As dfoes reflection in some other way.