<div dir="ltr">All that is stored in the pixel from the first drawing is what percentage of the pixel was covered by the shape. It does not remember exactly what parts of the pixel are covered.<div> </div><div>When drawing the new shape, for any pixels that are not fully covered, it cannot tell how much of the old shape was covered. It could be all of it (if it was all under the new shape) or none of it (if it was all in the area of the pixel not covered by the new shape). So Cairo does a reasonable guess, which is that the old shape covers the pixel evenly (you could think of it as an infinitely fine pattern that covers exactly that percentage of the pixel). Say the old percentage of cover is X. The new shape covers Y percent of the pixel, leaving (1-Y)*X of the old shape visible. And this is the color it produces.</div><div> </div><div>The only real solution is to draw a much higher resolution with coverage rounded to exactly zero or one for each pixel. This aliasing will not be visible to the user as the pixels are too small (or a number of them are averaged together to make the real on-screen pixel). Contrary to what many people think this is not possible yet, aliasing artifacts are still visible on many graphics and will be unless resolution is raised huge magnitudes. Therefore antialiasing is still applied to lines and characters at least.</div></div><div class="gmail_extra"> <div class="gmail_quote">On Fri, Sep 16, 2016 at 5:55 AM, Josh Sanford <<a href="mailto:xrjoshs@gmail.com" target="_blank">xrjoshs@gmail.com</a>> wrote: <blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr">[I previously posted this on stackoverflow, but that appears not to be the best forum for the question.]While using Cairo 1.14.6 for display purposes, I found that overwriting the very same path with another color does not necessarily overwrite all pixels, and leaves undesirable artifacts behind.As evidence of my claim I offer this output from a short self-contained example, the source for which follows further below.<a href="http://i.stack.imgur.com/QBXjh.png" rel="nofollow" style="margin:0px;padding:0px;border:0px;color:rgb(0,89,153);text-decoration:none" target="_blank"><img src="http://i.stack.imgur.com/QBXjh.png" alt="enter image description here" style="margin:0px;padding:0px;border:0px;max-width:100%"></a>An explanation of the six parts of the image, from left to right:<ol style="margin:0px 0px 1em 30px;padding:0px;border:0px;font-size:15px;color:rgb(36,39,41);font-family:arial,"helvetica neue",helvetica,sans-serif;line-height:19.5px"><li style="margin:0px 0px 0.5em;padding:0px;border:0px;word-wrap:break-word">Original shape stroked in blue.</li><li style="margin:0px 0px 0.5em;padding:0px;border:0px;word-wrap:break-word">Original shape overwritten in RGBA white.</li><li style="margin:0px 0px 0.5em;padding:0px;border:0px;word-wrap:break-word">Original shape overwritten in RGB white.</li><li style="margin:0px 0px 0.5em;padding:0px;border:0px;word-wrap:break-word">Original shape overwritten in RGBA white with CAIRO_OPERATOR_SOURCE mode.</li><li style="margin:0px 0px 0.5em;padding:0px;border:0px;word-wrap:break-word">Original shape overwritten in RGBA white with CAIRO_OPERATOR_SOURCE mode and CAIRO_ANTIALIAS_NONE.</li><li style="margin:0px;padding:0px;border:0px;word-wrap:break-word">Original shape overwritten in RGBA white with CAIRO_OPERATOR_SOURCE mode and CAIRO_ANTIALIAS_BEST.</li></ol>The image was generated from the following code:<pre style="margin-top:0px;margin-bottom:1em;padding:5px;border:0px;font-size:13px;width:auto;max-height:600px;overflow:auto;font-family:consolas,menlo,monaco,"lucida console","liberation mono","dejavu sans mono","bitstream vera sans mono","courier new",monospace,sans-serif;color:rgb(57,51,24);word-wrap:normal;background-color:rgb(239,240,241)"><code style="margin:0px;padding:0px;border:0px;font-family:consolas,menlo,monaco,"lucida console","liberation mono","dejavu sans mono","bitstream vera sans mono","courier new",monospace,sans-serif;white-space:inherit">#include "cairo/cairo.h" #define M_PI 3.14159265358979323846 void draw_shape(cairo_t* cr, int x, int y) { cairo_arc(cr, 50 + x, 50 + y, 48, -M_PI, -M_PI / 2); cairo_stroke(cr); cairo_move_to(cr, x + 2, y + 2); cairo_line_to(cr, x + 48, y + 48); cairo_stroke(cr); } int main(int argc, char** argv) { int x = 0; int y = 0; cairo_surface_t* surface = cairo_image_surface_create(CAIRO_FORMAT_ARGB32, 300, 50); cairo_t* cr = cairo_create(surface); /* Draw a white background and a few shapes to overwrite */ cairo_set_source_rgba(cr, 1.0, 1.0, 1.0, 1.0); cairo_paint(cr); cairo_set_source_rgba(cr, 0.0, 0.0, 1.0, 1.0); draw_shape(cr, x, y); x += 50; draw_shape(cr, x, y); x += 50; draw_shape(cr, x, y); x += 50; draw_shape(cr, x, y); x += 50; draw_shape(cr, x, y); x += 50; draw_shape(cr, x, y); x += 50; x = 50; /* Leftmost shape is left unchanged for reference */ /* Stroke in RGBA opaque white */ cairo_set_source_rgba(cr, 1.0, 1.0, 1.0, 1.0); draw_shape(cr, x, y); x += 50; /* Stroke in RGB white */ cairo_set_source_rgb(cr, 1.0, 1.0, 1.0); draw_shape(cr, x + 0, y); x += 50; /* Stroke in opaque white without blending */ cairo_set_source_rgba(cr, 1.0, 1.0, 1.0, 1.0); cairo_set_operator(cr, CAIRO_OPERATOR_SOURCE); draw_shape(cr, x, y); x += 50; /* Stroke in opaque white without blending, with no antialiasing */ cairo_set_source_rgba(cr, 1.0, 1.0, 1.0, 1.0); cairo_set_operator(cr, CAIRO_OPERATOR_SOURCE); cairo_set_antialias(cr, CAIRO_ANTIALIAS_NONE); draw_shape(cr, x, y); x += 50; /* Stroke in opaque white without blending, with best antialiasing */ cairo_set_source_rgba(cr, 1.0, 1.0, 1.0, 1.0); cairo_set_operator(cr, CAIRO_OPERATOR_SOURCE); cairo_set_antialias(cr, CAIRO_ANTIALIAS_BEST); draw_shape(cr, x, y); x += 50; /* Write the results to a file */ cairo_surface_write_to_png(surface, "output.png"); return 0; }</code></pre>It doesn't make intuitive sense to me that overwriting the very same shape would not overwrite all of its pixels, especially if I force it into non-blending CAIRO_OPERATOR_SOURCE mode. The results are the same on the framebuffer that constitutes my actual surface, so this is not an issue with the backend.Cairo is usually so good at what it does that I'm very surprised at this. Is there no way to overwrite an anti-aliased shape exactly in Cairo?Josh</div> -- cairo mailing list <a href="mailto:cairo@cairographics.org">cairo@cairographics.org</a> <a href="https://lists.cairographics.org/mailman/listinfo/cairo" rel="noreferrer" target="_blank">https://lists.cairographics.org/mailman/listinfo/cairo</a> </blockquote></div> </div>