You don't seriously expect the 805x-series MCU to interpret MPEG formatted video in real time, do you?

Now ... let's review ... the laser produces a pixel. It doesn't produce a line or any shape. Sweeping a pixel over the area to be selectively illuminated or not is a matter of timing. It doesn't matter what the encoding scheme is, because the data can be stored in ROM in bitmap form. It can also be stored in vector format, but that requires MCU involvment in its interpretation, as there have to be commands like

... move to x1,y1 ... figure out how to do that, wait for the (relatively slow) mechanism to finish.
... turn the laser on ... wait for it to turn on ...
... move to x2,y2 ... turn the laser off (or not) ...
... move to x3,y3 ...etc...

With a raster display, that's all almost trivially done with hardware, particularly since it can be preloaded as a bitmap.

Back in the early '80's, I built a laser soldering machine for a client who wanted to solder components glued to both sides of a PCB at the same time, so I'm slightly acquainted with what it takes to move a mirror around using vector-formatted commands. We used HPGL for that process. The distances were short, and the precision was high.

In this case, the spinning mirror is much easier since one can format the data in a simple bitmap, use a VCO to control the pixel rate as the light lever varies, and move the mirror with the (relatively quite slow) voice coil.

In a classically formatted 80x25 (e.g. IBM MONO display) text display, there are nominally 2000 characters of 8x12 pixels per frame. For what Chico is doing, I doubt he needs any more than 256 x 192 or at most, twice that density, of pixels for the entire frame. The rasterized display runs no risk of overheating the laser or, the illuminated area, because it moves the laser around and turns it on and off. Fills are no problem. It does have to hit every illuminated point on the page frame at a relatively high rate, though, so the retinal persistence (no phosphor to help us in this case) can make the image seem static.

A common average seek for a typical disk drive is still about 10 ms, and track to track, which, nowadays is a very short distance indeed, is still on the order of a couple of milliseconds.

With the rasterized display, the physical motion of the slow mirror can be constantly oscillating. The spinning mirror and the vco-controlled pixel clock handles the rest.

Once you attach numbers to the mechanical tasks, you will see that moving a fixed laser with a loudspeaker voice coil, which can move only a very short distance at rates at most well below the required sweep rate, I think you'll see that complex drawing will be highly impractical with the vectorized approach. Yes, you can draw the outline of a dog or a horse, or an airplane, but not a chessboard with the pieces present, when using vectors. With a raster, you can do that. I've seen it done.

RE