GoatFoam wrote:That's pretty interesting. Just curious about how you are doing the simulation? Converting RGB colours to a spectrum, removing bits of it and converting back to a renderable colour space, or are there shortcuts?
RGB->
XYZ, then the channels are mashed up accordingly to the defect (which is pretty messy), and finally it's converted back to RGB again.
For those two kinds of achromatopsia it's just a simple vector multiplication, however. Right now the first one isn't all that accurate (as I found out yesterday). With complete color blindness you also get 5-10% less visual acuity (could be simulated with some blur) and increased sensitivity to lightness (could be simulated with aggressive bloom). Well, with those conditions there isn't really much you can do to aid them. Except of really bold shapes and strong contrasts. I'd guess that most of them don't play any games or browse the web with usual tools/settings. Well, it's a bit hard to find any information on that topic, because only about 1 in 100,000 are affected by this. So, I'll probably leave this part of the simulation as it is for now, since it's a good indicator for the contrast levels.
--
Made lots of progress these days. Lock free multithreading is there. Unsurprisingly it scales pretty linearly with the number of cores, which is truly great since about 80% of those Macs (and more and more PCs) got 2 or more cores. Automating the Mac packaging also went pretty well (used the
JarBundler Ant task). Still need to do the same for Windows (won't be as easy I'm afraid). Also added i18n support (will support en_US, en_UK, and de_DE out of the box) and dodged
some weird Mac i18n pitfall.
Mac still needs a bit more attention here n there, but it looks like I'll get there soonish. And there are a few other things I need to take care of. I didn't really expect it to be so much work... oh well... the devil is in the details.
