Back in the spring I scanned my lumber grading training materials into PDF and posted them here on AS. The thread is
HERE and the direct link to the PDF is
HERE. It's more geared towards softwood structural lumber, but there is still a ton of information concerning grading technique, defects & lumber characteristics, species identification, etc. I can't scan the actual NLGA rule book since it's copyrighted, but they can be had for a nominal fee straight from NLGA. As previously stated though, if you're dealing with hardwoods mostly, you'll likely want to get NHLA materials since hardwood grading is a fair bit different, because shop/cabinet grades are based mostly on appearance rather than structural integrity.
Curlycherry had some good points a few posts back too. Most notably that it's important to "read" a log before you even touch steel to it. This is a skill that must be acquired through a lot of experience and trial-and-error. Training will go a long way, but every log is unique and potentially full of surprises. Once you learn what to expect from the wood grain by looking at the outside characteristics of the log, it's much easier to get the highest yield at higher grades. Back in August I posted
THIS THREAD with many pictures of a log I milled various different dimensions out of, with descriptions of how/why I made the cuts I did. I should note that I'm by no means an expert sawyer compared to many others here; rather I'm a trained lumber graderman with some sawing experience on the side. I like to take a lot of pictures of the logs I do even if I don't share them here - it helps me remember what did and didn't work well, and why. I try to do it with every log - since I don't do that many it isn't a big inconvenience, but it wouldn't make much sense for someone trying to seriously produce to photograph everything. Photo and video are the next best thing to first-hand experience though and can teach a lot. Like a sports team studying replays of their last game I guess.
All "grade sawing" or "sawing to grade" means is that you are processing the log with the objective of achieving the highest grade possible in all pieces you cut, which may not necessarily allow the highest board footage yield. In a structural dimension sawmill, logs are just cut to get the highest yield at the fastest speed, since in the long run it doesn't matter that much whether a 2X4 stud is flatsawn, riftsawn, or quartersawn (other than differences in warping tendency). Finishing work requires much more attention to the wood grain, not just for aesthetic reasons but also (and possibly more importantly) for stability and integrity, so more time is taken to make sure that each board is cut in the best possible way.
One might assume that (in hardwoods at least) quartersawing would achieve the best grade result. This would likely be true with a good, straight, clear log. But say the log has a lot of small knots, such that it won't really produce clear lumber anywhere, but the knots aren't large enough to make the log a writeoff. If you were to quartersaw the log, you would end up with a lot of "spike knots" in the lumber since the boards are being cut radially from center to edge:
I've admittedly exaggerated things slightly in that drawing, but you can see that a spike knot can occupy the vast majority of the cross section of a quartersawn board, whereas a knot of the same size would occupy just a fraction of the cross section of a flatsawn board, and result in a "round knot" defect. This is why most structural lumber mills actually prefer to flatsaw the lumber if possible - it's just that it isn't worth it for them to slow down and continually rotate and scan logs. For example, in Stud grade a knot can occupy half the cross section of a board (slightly more if it's away from the edges). It takes a round knot of 1-3/4" diameter to occupy that much in a 2X4 (nominal 3-1/2" wide) - HOWEVER it only takes a 3/4" diameter spike knot to do the same damage since the boards are only 1-1/2" thick. But I digress, these concepts are discussed in the grading materials I linked to above.