5.08.25 added perimeter-based step_size manipulation Now bigger ellipses will get similar accuracy to smaller ones. When you make steps size bigger, the inaccuracies will show up at the same time, regardless of size of ellipses (It's harder to see inaccuracies on bigger ellipses) htps://scratch.mit.edu/users/heldlaw/ His algorithm is very similar to mine, but intentionally or not intentionally uses quality changes based on eccentricity of the ellipse. Unfortunately it doesn't look good on low resolutions, while the one I use make them look more natural, even when they aren't accurate. My hypothesis is that ellipses with much bigger width than height have bigger perimeter, which causes inaccuracies easier to see. That's why in general big ellipses need more draws than smaller ones. https://scratch.mit.edu/users/sly_i/ https://scratch.mit.edu/users/-Jie-/ https://scratch.mit.edu/users/sukeppy/ Their implementation is perfect, looks like I copied their code, but fortunately I improved amount of steps moved and keeping the width and height constant among different resolutions https://scratch.mit.edu/users/heldlaw/ (no step size control, unintuitive variable "ratio". I'm researching his idea)
There is level of detail used, which may be disgusting to other people. Don't worry, for given width/height parameters, there is a Step_Size value which makes it indistinguishable from Step_Size approaching 0 (perfect ellipse) Feel free to zoom in the smallest and most flattened ellipses. (I want some attention to discuss about this project, sorry) #All