For one, better monitors, where it was easy enough to distinguish pictures. Basically CRTs that are better than those required for TVs. But the biggest issue w

For one, better monitors, where it was easy enough to distinguish pictures. Basically CRTs that are better than those required for TVs. But the biggest issue was display RAM. To get a decent enough resolution, you need an insane amount of memory, often 50+ kb! And then you need this for every terminal hooked up to the minicomputer/mainframe (don't even think about double-buffering). This was even an issue for the early text terminals. 72x20 characters was bad enough. Compare this to a line printer, where you just send a character and it stays there, without any backing memory. If I recall correctly, there were some experiments with persistent CRTs, and then of course terminals as we know them that had some memory of their own used to store the sent characters, and after a while even interpret quite elaborate escape sequences (compare sending one "^L" command to sending 80x24 spaces to clear a screen). Once RAM got cheaper and you had semi-decent monitors, you could build some pretty high resolution monochrome terminals. If you wanted color, too, you had to stay low-res, which is why this was mostly used for gaming, not for actual work. It took quite a while until you could build high-res color workstations, and those were pretty expensive at first. We live in enchanted times, don't we?