Skip to main content
8 events
when toggle format what by license comment
Jun 16, 2020 at 10:15 history edited CommunityBot
Commonmark migration
May 21, 2014 at 13:22 comment added Miklas @MickLH: Vertices for such post-processing projection can be calculated offline or on load, yielding a very cheap pass-through vertex shader. For a bulge filter, a full screen quad must be rendered and rasterized as well (except if SpartanDonut has access to e.g. compute shaders), and for every pixel the offset must be calculated instead of interpolated. Therefore, I believe (depending on the available tools and implementation) both methods could be equally fast. I've updated my answer, thanks for your input.
May 21, 2014 at 13:18 history edited Miklas CC BY-SA 3.0
Corrected statement about fasted method, thanks to @MickLH.
May 20, 2014 at 22:27 comment added MickLH Both require processing the same amount of image. The projection will never be faster than doing nothing, and image processing costs the same either way. -- "Distortion" method is faster because you're skipping the vertex transform and rasterization, also you're giving the driver more freedom to iterate the pixels, where the polygon approach is more specific about what order to draw in. Finally, polygons hold less information about your intended effect, are resorting to a generic solution which stresses the texture filtering hardware and produces an approximation instead of the real filter.
May 20, 2014 at 12:40 vote accept SpartanDonut
May 20, 2014 at 12:40 vote accept SpartanDonut
May 20, 2014 at 12:40
May 18, 2014 at 12:54 comment added Tholle Fantastic answer!
May 18, 2014 at 12:13 history answered Miklas CC BY-SA 3.0