You will notice alot of the hard surface stuff you find is in max. Many of the Max techniques can be used in Maya but a lot do not translate to Maya very well, especially some of the stuff from the hard surface essentials from 3dtotal.

I followed a lot of the examples in there and although the results look fine in max, in Maya you get all kinds of artifacts and render problems and pinching. It seems turbo smooth is either smarter or more forgiving then smoothing in maya. That being said he offers some very sound techniques. You can get good hard surface models in Maya, but you have to work very hard and put a lot of thought into the way you model things.

The geometry in the vimeo video is an example of the kind of stuff that gets heavily criticized here because of all the messy geometry and edges that he terminate in ngons. So unless you have a thick skin I would avoid going down that path. Personally, I know where ngons cause problems and where they do not, so although I strive for all quads, I do not go nuts trying to run a metric butt ton of extra geometry all the hell over my model just to make everything quads! One way to kind of meet in the middle is to split the model into multiple parts so you can localize the geometry that needs a lot of support edges.

I have spent a lot of time studying hard surface modeling and I have yet to find any really outstanding tutorials on it using Maya. My background is in automotive surface modeling and structural analysis of parts and assemblies using applications like solid works, inventor, catia, alias studio, etc, which are all primarily parametric nurbs based modelers and the techniques they use are not easily translated to Maya, especially with the state of Maya's NURBS tool set (having remained virtually unchanged for more then ten years). Poly modeling unified complex hard surface shapes in Maya is, quite frankly, a bitch.

It seems the focus lately is on organics where most people are only interested in making blobby nurbs shapes and converting them a basic poly cages, then sending them to zbrush and subdividing a bazillion times and then sculpting like clay which is a workflow not well suited to hard surface modeling or to engineers that are used to working with precision and accurate drawings. I have seen a few decent looking hard surface zbrush models and I have also watched the tedious, very inaccurate, and almost painful workflow that I can only describe as trying to do brain surgery while wearing boxing gloves!

What is needed are tutorials that go over when to smooth and when not to smooth and how to optimize geometry when both smoothing and not smoothing. What I mean here is many times I see people saying "I intend to smooth this so I will use the minimum possibly geometry" and then after smoothing you get faceting at close camera distances and then have to subdivide even more driving the poly count way beyond what you would have had if you started with high resolution in the first place. So you have to know how close the camera is intended to get in order to resolve this (yes you can use the approximation node in mental ray). Then there is the issue of unifying many compound shapes into a complex unified part, how to consistently add edge bevels at junctions, while avoiding creases, pinching, render artifacts, and crappy or excessive geometry. I have yet to see any demonstrated consistent and REPEATABLE workflows for this problem. Then there is the issue of holes! Cutting holes, especially with sharp corners, into curved surfaces and then trying to smooth can be a big problem with no real consistent solution. Most of what I see is trial and error, touchy feely, stuff which is very unnatural to someone coming from a 3D CAD application background.