View Single Post
Old 04-20-2012, 01:26 AM   #108
geekmaster
Carpe diem, c'est la vie.
geekmaster ought to be getting tired of karma fortunes by now.geekmaster ought to be getting tired of karma fortunes by now.geekmaster ought to be getting tired of karma fortunes by now.geekmaster ought to be getting tired of karma fortunes by now.geekmaster ought to be getting tired of karma fortunes by now.geekmaster ought to be getting tired of karma fortunes by now.geekmaster ought to be getting tired of karma fortunes by now.geekmaster ought to be getting tired of karma fortunes by now.geekmaster ought to be getting tired of karma fortunes by now.geekmaster ought to be getting tired of karma fortunes by now.geekmaster ought to be getting tired of karma fortunes by now.
 
geekmaster's Avatar
 
Posts: 6,433
Karma: 10773668
Join Date: Nov 2011
Location: Multiverse 6627A
Device: K1 to PW3
Quote:
Originally Posted by jmseight View Post
I think a better way is to scale first, so the brightness values goes instead of 0~63 to -3~67. Then use the same dither table. This will make a smooth transition from all black to linear to all white.
Agreed, but a proper S-curve might be better than linear rescaling. I rescaled mine to 0-255, so I can display 8-bit images without rescaling pixel values. Unfortunately, I discovered that on the K5(touch), 8-bit values are truncated to 16 colors (without rounding). So we need to dither 256 colors down to the actual 16 colors that it DOES display, similar to how I am dithering 65 colors down to 2 colors (black and white). Or, we could just use eips to display a PNG file and let it do the dithering for us.

It appears that kindle apps that display images must be dithering them from 256 color down to 16 colors when shown on eink, so we need to do the same, and I can adapt my "geekmaster formula 42" to support this. It is just a matter of time, and there is never enough of that.

EDIT: I just realized that my formula as published relied on bit 8 being a sign bit, so I had to change 128 to 256 in four places in the formula, and now it works correctly with my range 0-255 8x8 dither table. That will mean no pixel conversion step going from 8-bit gray to dithered black and white. Nice!


Last edited by geekmaster; 04-20-2012 at 04:47 AM.
geekmaster is offline   Reply With Quote