Color E-Paper not without problems

[ProDigit]

If 8 bits per pixel results in 256 grey tints.

Then in color there are 4 pixels per cluster (white/black, white/red, white/green and white/blue; or whatever magenta color it will be).
so if every pixel had 2 modes, 'on' or 'off' (off being white, on being color), you'd have 16 colors.

If every pixel had 4 positions (Off(white), light, dark, on(color)) you'd end up having 256 colors. Depending on if it's possible for each pixel to have sub colors or not.
If it's not possible, you'd have to 'borrow' other pixels to form a cluster of pixels.
This way one could like on a TV eg: have 4 pixel rows merging as one major pixel. That's (probably) called rendering.
If colors are displayed red green blue white in this order on the screen:
RGBWRGBWRGBW
RGBWRGBWRGBW
RGBWRGBWRGBW
RGBWRGBWRGBW
They probably will create major pixels like this:
=====
|RGBW| \
|RGBW| .| FIRST LINE, FIRST PIXEL GROUP
|RGBW| '|
|RGBW| /
=====
|RGBW| \
|RGBW| .| NEXT LINE, FIRST PIXEL GROUP
|RGBW| '|
|RGBW| /
=====
..... } .....etc.....



===== ==== =====
|RGBW|RGBW|RGBW| \
|RGBW|RGBW|RGBW| .| On screen Pixels are
|RGBW|RGBW|RGBW| '| grouped like this
|RGBW|RGBW|RGBW| .|
===== ==== =====
|RGBW|RGBW|RGBW| '|
|RGBW|RGBW|RGBW| .|
|RGBW|RGBW|RGBW| '|
|RGBW|RGBW|RGBW| .|
===== ==== =====
|RGBW|RGBW|RGBW| '|
|RGBW|RGBW|RGBW| /
...........

Each pixelgroup having 4 lines of 4 colors assigned to them; that's 16 effective pixels that get lost in what supposed to be one pixel.

In other words with current 800x600 e-ink screens, we might end up with a device which has a meager 200x150 pixels to get the full 256 colors. 200x150 pix is good enough for thumbnail pictures or avatars on a forum.

To get more detail they need to use some sort of rasterization, or whatever the name is (like the old bootscreen of Windows98 and NT which has only 256 colors, but due to the ... effect used ... (the name of which I forgot) it looks like it actually has more than just 256 colors).
Large surfaces of plain colors are drawn as close to the original color by borrowing other pixels to complete the color, while fine details will lose some colors over displaying details (pixels are released of a group to get a sharper detail)...A nice compromise between the resolution and colors.

I only wished I remembered the name of that style of rendering.

[delphidb96]

Originally Posted by ProDigit

To get more detail they need to use some sort of rasterization, or whatever the name is (like the old bootscreen of Windows98 and NT which has only 256 colors, but due to the ... effect used ... (the name of which I forgot) it looks like it actually has more than just 256 colors).
Large surfaces of plain colors are drawn as close to the original color by borrowing other pixels to complete the color, while fine details will lose some colors over displaying details (pixels are released of a group to get a sharper detail)...A nice compromise between the resolution and colors.

I only wished I remembered the name of that style of rendering.

Dithering.



Derek

[ProDigit]

Right! It was 2AM when I wrote that, and all the time I just couldn't find the word! lol

Thanks!

[DaleDe]

Originally Posted by ProDigit

Right! It was 2AM when I wrote that, and all the time I just couldn't find the word! lol

Thanks!

You arrange the main color tree as:

WR|GB|WR
GB|WR|GB or something similar. The resultant resolution for a 600x800 screen is 300x400 which is bad but not as bad as you indicated. Dithering can help as you say to make a more effective use of color. You can certainly get 256 colors if you use RGB color scheme but as you say using a four pixel array reduces that somewhat.

If you consider a 6 bit array of RGB you get 64 colors 2**6. Adding whilte provided additional shades of those 64 colors which is an additional color to the extent that White lightens the color. So if we consider the 4 values of white with the other two bits we end up with 4*64 which is still the full 256 colors. Using 2 bits per pixel takes advantage of the sub-pixel capability of a display as shown at: http://www.eink.com/technology/howitworks.html

Dale

[ProDigit]

Ow they use the pixel sorting of the OLPC XO; diagonal instead of horizontal?

The following respond I originally wrote with the thought in mind that e-ink pixels have only 2 positions: On, or off. With that in mind I wrote the following:


As for the white color pixel, I expect it to be black, since each pixel is either white or RGB (RYB)
It would make no sense to have a white pixel white in on and off position.
Also, the right color mix for passive colors (like on a paper) is Red, Yellow and Blue; as for emitted light like on a tv goes Red Green and blue.

I just tested it, and have this color palette:

1- Red
2- Red+Yellow=orange
3- Yellow
4- Yellow+blue=green
5- Blue
6- Blue+red=purple
7- Red+Yellow+Blue=grey
8- nothing = white

Each pixelgroup can be paired with an additional pixel, being either white or black, resulting in a lighter or darker overall color of the pixel. So that's 16 colors. It is possible to get white, grey and a mix of colors, but impossible to get a deep black color there, unless every RYB pixels is overlayed with a polarization filter and a LCD like screen but with only 2 modes; black or transparent.

So you'll need 2 screens, lose transparency, and contrast, or will have to do with a color ebook that can only display grey as it's closest color to black.

So I guess pixels will be:
R(ed) Y(ellow) B(lue) & W(black) (W for black, to differentiate between B from blue).

So take 2:

RYBWRYBWRYBW
BWRYBWRYBWRY } These row can be used to correct the color of the scanline on the top and/or on the bottom.
RYBWRYBWRYBW
BWRYBWRYBWRY } " " " " " " " " " " " " " " " " "
RYBWRYBWRYBW
BWRYBWRYBWRY } " " " " " " " " " " " " " " " "

The resultant resolution for a 600x800 screen is 300x400 which is bad but not as bad as you indicated.

You are right, IF RYBW pixels each can only have an 'ON' or 'OFF' state; meaning IF the total color of the screen is limited to 16 colors.
*Edit: If you want 256 colors either the resolution will be further cut in halve (to 200x300 pix, and I hereby correct my previous mistake of 200x150 pix), or a pixel needs more than 2 states (pref. 4).

If you consider a 6 bit array of RGB you get 64 colors 2**6. Adding whilte provided additional shades of those 64 colors which is an additional color to the extent that White lightens the color. So if we consider the 4 values of white with the other two bits we end up with 4*64 which is still the full 256 colors. Using 2 bits per pixel takes advantage of the sub-pixel capability of a display as shown at: http://www.eink.com/technology/howitworks.html

I have to yet find out what you mean with "6 bit array of pixels".
You mean that the technology has 3 bits per pixel instead of 2?

About the link you send:
I thought the technology on the screen of the Sony PRS-505 reader is not this technology, because the screen stays uninfluenced by strong magnetic fields. Yet it does seem to get influenced by sunlight, making it grey.
UV light seems to influence the screen, not really heat or magnetic field; something that wouldn't make sense when the e-ink technology used in the Sony reader would have been the technology displayed on the link you send.

I don't know if the link explains how e-ink works on all devices, but I suspect it is not this technology at work at the sony reader.


So, my question would be if the upcoming e-ink technology will have 3 or 2 bits per pixel. (Or perhaps more?)

2 bits would effectively set the resolution of current 6" 800x600pix models back to 400x300 effective resolution @ 16 colors, and halve that resolution even again for 8bit 256 colors.

3Bit pp. could up the colors to a 81 color, 400x300 effective res, or 243 colors on a 200x300 effective res.
3bit would mean each pixel would have:
1- 0% color (white)
2- 50% color
3- 100%color (RYB, or black)

4bit would bring 256 colors on a 400x300 effective res screen, and give:
0%, 33%, 67%, 100% of color. Each pixel need to have 4 states.




But like said, they need some sort of Calculator polarization filter with 800x600 pix. to overlay and darken the color pixels below, and create the Ebook resolution solely with RYB (leave the fourth B/W pixel).
you'll effectively need 2 layers, that need to be layered perfectly on top of eachother, which would make it harder to create flexible color e-ink screens.
Within such layers one could also create a third touchscreen layer.

I wonder when we're going to get rid of these separate layers, with touchscreen being 1mm away from the LCD, due to a bunchload of other filters, and empty space being inbetween the touch screen and the LCD glass substrate...
Why don't manuffacturers just simply not get rid of all those filters, and empty space, and just get a touchscreen glued on top of the glass substrate? Then there's no need to put an anti-glare layer on top, neither some plastic protection layers or UV/IR filters for the eye, since a touchscreen already has most of these functions built in itself (and is not even necessary for e-ink, only for CRT/LCD screen).

[DaleDe]

The Sony uses exactly the technology of E Ink. The only difference from the site is that the site description of sub-pixel is simplified by showing on 1/2 white and 1/2 black in the pixel. In fact the sub-pixel can be divided in much smaller granules than 1/2. For example to do 4 gray scale you really need 1/3 granularity. The choices then become all black, 2/3 black 1/3 white, 1/3 black 2/3 white, and white. To do 8 levels you will need even more subpixel division and to do 16 is currently the limit. The sub-granules mix of black and white are interpreted as gray by the human eye.

In a similar fashion you can use yellow/black, or blue/black, etc. to make various shades of any color. Using subpixel shades of each color represented by 2 bits means that each color will have 4 shades and 3 colors will need 6 bits or 64 different colors for the triad. that is how I got to 64 colors.

Dale