Yes sure. but again it depends on what he wants to get at.
Theres a easier way to do those passes and its called label passes, those combined with the coverage gives you a automatic access to 3*256-1 objects. All you need to do is add a attribute named miLabel and a number to a object or shader and your set, mr auto generates the color for you and since theres no keying involved you dont get artifacts for the masks*. Some comp apps like fusiunn automatically list you the objects too so all you need to know is the number or point to object and thats it the comp app separates any object / or part form the others.
So you get the same effect awfully lot faster if you care to read the manual.
In any case not much of an issue, if you KNOW what parts you need to mask you can always just spit that pass out its always better. The real problem is that most times you dont know what the comp is going to be exactly because it gets done after you leave the job. So more data is better if you have budget for that disk space. Thats what the color areas address, if you know you just want a mask for rims then spit out that exact mask if you know you need theese 20 masks then spit out those 20 if you DONT know what they need THEN spit out the label masks.
without the use of masking.
keying is masking. Label passes arre the way to go, does same thing but technically better.
Both of these too have big sets of problems, try to do a key pass when you have very fast moving mtion blurred objects for example. Suddenly it doenst work at all. Also thsy fail when having semitransparent shaders on top.
Theres a solution for this too. Its called a deep raster, deep rasters are GREAT they pretty much give you all the render info to play with with no comp artifacts to speak of, your even getting the subpixels right. The problem with deep rasters are that the scenes get very heavy. Theres 3rd halfway alternative and that only works well with prman and thats to load tha samples before filtering this is also great you get subpixel accuracy, with no render time overhead increase, just bigger disk spaces-
*this saves time but not only that the style in the blog gives less color to act on change. Label passes get better here.
PS: anybody who plans to do rendering for a living SHOULD work a rotation on the comp stage to get the feel what the guys can and what they cant do easily. Sometimes you get dumpped so much crap that its impossible. Thing si tough most times comp guys are more competent than the render techs so they can pull of impossible odds. But its a cost issue if its easy they do it in 10 minutes if not they will work on it for loong time. So ALWAYS talk to the guy who does the comp FIRST.