—- for some years one of my small project ideas to try to relearn to code has been an ascii-based graphics renderer. the fact that it does not seem very useful to me helps me have comfort around it. i had an idea of mapping the font shapes and using matrices of similarity to pick the best character for any particular area of the display and point in time. maybe libcaca does something similar, i don’t know. it was very hard for me to daydream this because my inhibitions had transferred heavily to software development, and because one of the major inhibitions is data visualization which would be aided by a terminal graphics library. so the ideas have come, had dissociative amnesia triggers developed around them, and had to redevelop differently, repeatedly. recently a concern i’ve had is representing vague shapes and vector-like imagery well. i’m imagining a naive metric of similarity based on pixel difference might work well for some things but fail, for example, to identify that a slash (/) character is similar to a diagonal line offset by a few pixels or at a slightly different slope. it would be nice if it would generate ascii art similar to what an inexperienced hobbyist might generate (here is a triangle: /_\) when it was appropriate to do so. i’m thinking, maybe a very simple approach to that could be to convolve the imagery with disparate translations, rotations, scalings, to find similarity that has spacial difference. [further thoughts include imagining a possible theory for how far a pixel or set of pixels must move to become identical, and how that motion is similar to that needed of adjacent pixels] it can be a bit of a confusing puzzle space because what a viewer would want is the greatest subjective similarity, for the human visual pathways, which is completely dependent on how humans process information (for example a picture of a stick figure is similar to a drawing of a human even if the poses are different, because we recognize it as meaning a human). that seems like a separate, maybe more modern challenge, to represent imagery in humanly clear and pleasant ways. but here i was thinking something more rote.