Conversation

> > Does there not exist some ridiculously parallel equivalent of Bresenham or Wu?

> Sure—it's trivial. Draw some bounding geometry and calculate the Euclidean distance to the line. Shade accordingly. The trick is finding that minimal bounding geometry.

Line drawing. Euclidean distance. OH! You just project each fragment on the line, measure the distance and use clamp() and mix() for the alpha. That's so damned clever. I wish I was smart enough to have thought of it.