You picture "Raster.png" shows, in fact, the common form of a Raster. It consists of a unidirectional (we'll say Left-to-Right) horizontal scan at a controlled rate, followed by a "fly-back" quick Right-to-Left return to the left edge while simultaneously moving down one scan line to begin the next sample.
If you are creating a raster using movement along Cartesian (X, Y) axes, this horizontal scan/diagonal "fly-back" is probably the most common model of scans, used in (analog) Video devices (the electron beam moves this way). The "catch", of course, is that you generally move the beam at two speeds, "Scan" and "Fly-Back", two "time" intervals, one where data are taken/displayed, and another where it is "ignored", and a "busy" Axis (X, generally) and one that moves a scan line every sweep (Y). [Note -- Videos often "interlace", doing even scan lines in one pass, followed by the odd scan lines, but the principle of a "fast" and "slow" axis is still valid].
It occurs to me that if you are doing the scan, yourself, you might try scanning back and forth. The Odd Scans go horizontally L->R. At the end, you stop the mirror, step down one Scan line on the Y axis (a short step), then on the Even Scan go R->L. This minimizes the large "Fly-Back", though you'll still have the same situation when you get to the bottom of the Scan, and need to move Y back to the top. But ... no reason you can't use the following scheme:
- Start Y at 0, X at 0, Scan L-R to XEnd. Step Y to Y+2.
- Scan R-L back to 0. Step Y to Y+2. If Y<YEnd, go back and repeat to do all Even Scans.
- Step Y to Y-1 to start Odd Scans. I'll assume Y is divisible by 4 to keep the Math simple.
- Scan L-R to XEnd. Step Y to Y-2.
- Scan R-L back to 0. Step Y to Y-2. If Y>0, go back one step to finish Odd Scan.
- Go back to the top to do another Even/Odd Left-Right-Left, Top-Bottom-Top scan.
This "messes up" the timing some, but saves "wear and tear" on the mechanics, and if there are multiple scans per "event-of-interest", this might be a useful tradeoff.
To implement a conventional raster, the algorithm goes like this:
- Start at Y=0, X=0.
- Move the X mirror at constant velocity to move beam from X=0 to X=XEnd.
- Turn beam off, move mirror back to X=0, at the same time moving Y to Y+1.
- If Y<YEnd, go back to Step 2. Otherwise turn beam off, move mirrors back to X=0, Y=0, and go back to Step 1.
Note that you'd need to test the timings -- galvo mirrors are pretty good, but they don't instanteously start and stop.
Bob Schor
P.S. -- I haven't (yet) looked at your code. This is in response to your pictures, to clarify that your Raster picture is what you should expect to see. You need to know "when to sample" and "when to wait", which will require a more complex sampling scheme.
