Nice presentation. Some of the sleep statements could be a little longer for us slow readers. ;)
I've found WINDOW to be a good way to predetermine how my graphics interacts with the data generated. I use it quite a lot, particularly for quickie applications. There is a tradeoff, I can define my space in R3 like I'm used to, but then I have to be aware to alter how any trig functions behave. It can be a bit of a pain working between vector and polar coordinates. This was how I handled the space flight program I've been endlessly working on. Generate a sub-image with custom coordinates using VIEW & WINDOW then _PUTIMAGE to the mainscreen.
One painful lesson I learned from that approach, if your "model space" is much larger than your "display space" and contains a lot of "off-screen" details, you're well served to use logical operators that avoid the processing time of drawing things that will not actually be displayed. Otherwise, things can slow to a crawl if the graphics are complex, such as liberal use of Steve's fcirc, in my case.
I'm going to have to get familiar with PMAP, it suddenly occurred to me that it might hold the key to doing some mouse selection bugs I've been dealing with.