(2019-01-25, 15:48)Chris Dee Wrote: ... I'd like to understand the relative development effort and rendering efficiency between a) building the text string with txt2dat and tweaking if necessary, versus b) scaling and combining individual character primitives.
I would say that the difference is not that great.
The difficulty lays in finding the correct, and free of charge, font, downloading it and tweeking the settings in txt2dat.
Once you have a good set of low enough resolution letters in a file, I import it into LPC and scale and resize it to fit the text in the sticker I'm working on. I tend to remove the four corners from the txt2dat file, and reuse/reshape the rest, to minimize the vertices.
For sure, having a character primitive might speed things up, but only if it was already created. And then I would have to inline them to simplify the mesh, removing unnecessary vertices. The benefit of using txt2dat is that the triangulation between the letters is already done. Using inlined character prims might force a complete recreation of the background surface. OTOH sometimes you want a vertex between (or above and below) the letters to avoid thin, long triangles. But that is easy to add in LPC.