mouse@shamash.mcrcim.mcgill.EDU (der Mouse) (06/27/90)
>> My point is that "the center of the pixel" is not a precisely >> specified notion. > The protocol description seems to implicitly define that pixel > centres are on integral coordinates (I have a feeling there is an > explicit statement somewhere, but I cannot find it). rws posted something saying it's in the Glossary. As I noted in a followup, I have an incomplete copy of the protocol document; my copy has no glossary. (Soon to be remedied, I hope.) >> Since line endpoints are always on integer coordinates, the most >> reasonable jaggies happen if pixel centres have coordinates of the > ^^^^^^^^^^^^^^^^^^ >> form (integer+.5,integer+.5), but [...]. > Why do you say this? Because I hadn't thought about it enough. What I meant to refer to was the differences among O............. OOOOOOO....... OOOOOOOOOOOOO. .OOOOOOOOOOOOO .......OOOOOOO .............O for a line drawn between the upper left and lower right corners of a 2x14 rectangle. As the pixel grid is shifted with respect to the coordinate grid, each of the above pixelizations (and several others) is possible. der Mouse old: mcgill-vision!mouse new: mouse@larry.mcrcim.mcgill.edu