Palm Ships With 12-bit Screen, Says 16-Bit On Box

Launch was among the many readers to point out that "Palm recently announced that they made a mistake in their product description of the m130... it doesn't have the 16-bit screen they advertised. Rather then admit the mistake, Palm is using every ounce of their spinning power to mislead its less tech-savy customers into believing that the palm m130 can display 58,621 'color combinations' rather then the 'more than 65,000 colors' it had previously stated; only a 11% difference. This tricky language is meant to shade the fact that a 12-bit screen can only display 4,096 colors... that's a 93% difference." Have they not learned from the mistakes of history? On the other hand, the screen resolution is 160x160 pixels.

Comments

Palm Infocenter has complete story

[Launch]

Apparently this debate has been going on a long time... Palm info center has a good article about it... And the PIC forum where the debate first broke.

Re:Blending techniques

[roarl]

ok, graphics geeks... factor 58,621. You get 31 x 31 x 61. Looks like 5-bits, 5-bits, and 6-bits, blended. I'm wondering how they came up with that number of colors! Any ideas?

By dithering (mixing) 4 pixels in a 2x2 pattern, 16 colors can be mixed into (16-1)*4+1 = 31 colors. By dithering 2x1 pixels, 16 colors can be mixed into (16-1)*2+1 = 15 colors. So, by using a 2x2 dither pattern for green, and a 2x1 dither pattern for red and blue, 31x31x61 colors can be produced.

I do believe this is the correct explanation, but it seems so contrived that I suspect some boss ordered his engineer to invent a reason to come up with a number close to 65536. In a program, it would be much easier to do a 2x2 dither pattern for all three components, yielding 226981 colors.

For interested readers, a transition from one color to another using a 2x2 dither pattern can be as follows.

00 10 10 11 11
00 00 01 01 11

As you see, two colors turns into (2-1)*4+1 color patterns.

Some call it Temporal Modulation

[Theovon]

This "Framerate control" is called "Temporal Modulation" in some circles. It works very well with LCD displays because they have such a long decay period (change the pixel color, and it takes a while to really change on the display). If the refresh rate is, say, twice the response of the LCD display, then you can double your RGB values by doing two-frame temporal modulation. That would yield 32k colors. If you were to do four-frame temporal modulation, that would give you 64k colors.

One thing I don't know is how different shades are done on an LCD in the first place. It may be some high-rate temporal modulation in the first place, although I doubt that. One thing I know is that LCD panels have a sinusoidal gamma curve, and this is because brightness levels come from the angle of rotation of the crystals. 90 degrees gives you black, 0 degrees is white. If you were to rotate the crystal by linear angle, it would not be a sinusoidal color response.

Of course, add on top of that the fact that even a linear scale in light emission (luminance) is not a linear scale to the human eye (luma). These are why LCD displays are notorious at having poor color response, and the manufacturers don't seem to be smart enough to compensate for it, even though the math is butt easy to people like our esteemed friend Dr. Charles Poynton.

Oh, and Temporal Modulation is not a linear interpolation. Why is left as an exercise for the reader. :)