PocketBook Color PB633 Released

[j.p.s]

Originally Posted by JSWolf

I find the problem with the Kindle is that there needs to be in between sizes. One size is too small and the next up is too large.

That's where FONT_RAMP comes in handy.

[Nate the great]

Originally Posted by patrik

Can you mix the modes on the same screen at the same time?

I'm going to have to build an ebook to test this.

[Quoth]

Originally Posted by patrik

Can you mix the modes on the same screen at the same time?

I don't think there are modes. It's just an unusual physical filter that only filters half of each pixel. So colour is inherently 1/3rd resolution horizontally on any screen content using monochrome and colour content.

Nor can you place colour at 300 dpi. Colour is needing three pixels.

[Nate the great]

Originally Posted by patrik

Can you mix the modes on the same screen at the same time?

Originally Posted by Nate the great

I'm going to have to build an ebook to test this.

I think it does have mixed modes, yes. But I might be wrong.

[mdp]

Originally Posted by Nate the great

As for the screen resolution weirdness, Pocketbook has always said they had a screen capable of displaying 100 ppi color and 300 ppi grayscale.

Originally Posted by DuckieTigger

I think you are making a fundamental mistake in associating resolution

The following is vector text rasterized at 300DPI

the following, instead, is vector text rasterized at 100DPI

In one element of the below one are nine of the above one. I think they are unmistakable.

Instead, what we got out of photos taken on the Kaleido as implemented in the displays of the Hisense A5 Pro CC, this is the apparent difference

not the sharpest, still probably discernible and convincing. That is in no way a 300DPI to 100DPI resolution loss.

That sentence from Pocketbook seems, up to now, to be false to misleading, and a shoot in their foot.
Pocketbook must have meant the basic: that you need three filtered dots to - virtually - display one colour. I think it is important to """nitpick""" there: that means a resolution of 173DPI (300/√3 - three dots to make colour, not nine - or in the terms from Quoth, 100 in a direction but still 300 in the other one, so √(300*100)), not 100DPI.

When other commentators have mentioned such resolution downgrade, they have sort of communicated the idea that the system reduces the resolution when colour is involved. That would create the downgrade shown at the top, only tinted. Instead, it seems from the photos that the system produces an in-colour display output starting from a full resolution input. So, not the effect you are used to, but not the devastation I show above.

[DuckieTigger]

Originally Posted by Nate the great

I think it does have mixed modes, yes. But I might be wrong.

I think the easiest way to show is by using a text only ebook with different color words. E.g. some black, gray (75%, 50%, 25%), red, green, blue, purple, brown, and cyan. And then take a picture with a camera good enough to show individual 300ppi pixel. My theory is, that the general outline of each word is exactly the same, except for the color words having all red and or green and or blue on "white".

[DuckieTigger]

Originally Posted by mdp

The following is vector text rasterized at 300DPI

the following, instead, is vector text rasterized at 100DPI

In one element of the below one are nine of the above one. I think they are unmistakable.

Instead, what we got out of photos taken on the Kaleido as implemented in the displays of the Hisense A5 Pro CC, this is the apparent difference

not the sharpest, still probably discernible and convincing. That is in no way a 300DPI to 100DPI resolution loss.

That sentence from Pocketbook seems, up to now, to be false to misleading, and a shoot in their foot.
Pocketbook must have meant the basic: that you need three filtered dots to - virtually - display one colour. I think it is important to """nitpick""" there: that means a resolution of 173DPI (300/√3 - three dots to make colour, not nine - or in the terms from Quoth, 100 in a direction but still 300 in the other one, so √(300*100)), not 100DPI.

When other commentators have mentioned such resolution downgrade, they have sort of communicated the idea that the system reduces the resolution when colour is involved. That would create the downgrade shown at the top, only tinted. Instead, it seems from the photos that the system produces an in-colour display output starting from a full resolution input. So, not the effect you are used to, but not the devastation I show above.

Almost, except we need a picture that is not fuzzy. That makes a huge difference.

Edit: to further nitpick, even the fuzzy image is clearly a resolution loss of 300ppi vs 100ppi. It does no subpixel anti aliasing. There is clear big steps in the red, while the black is smoother.

[mdp]

Originally Posted by DuckieTigger

Almost, except we need a picture that is not fuzzy. That makes a huge difference

Absolutely. Nate? ; )

Originally Posted by DuckieTigger

Edit: to further nitpick, even the fuzzy image is clearly a resolution loss of 300ppi vs 100ppi. It does no subpixel anti aliasing. There is clear big steps in the red, while the black is smoother.

Eh?! You see in there a loss of 9x ?! The loss below comparable to the loss above? No, I doubt that the fuzziness of the interpolated real photos could ever manage to hide a resolution loss 300→100, which is the one I showed with the "This is a string text". Maybe an equivalence of 300→173 I could consider possible. It is true that blurriness can conceal everything, but to make 300 and 100 similar I think it would have also annihilated the tinted filter texture.

Anyway: if we got better photos there would be no need to wonder and reconstruct.

[DuckieTigger]

Originally Posted by mdp

Absolutely. Nate? ; )



Eh?! You see in there a loss of 9x ?! The loss below comparable to the loss above? No, I doubt that the fuzziness of the interpolated real photos could ever manage to hide a resolution loss 300→100, which is the one I showed with the "This is a string text". Maybe an equivalence of 300→173 I could consider possible. It is true that blurriness can conceal everything, but to make 300 and 100 similar I think it would have also annihilated the tinted filter texture.

Anyway: if we got better photos there would be no need to wonder and reconstruct.

Yes. Different sizes. Your b/w example has text much smaller than the kaleido example. Make it fuzzy, and the difference will be smaller. Zoom in on the fuzzy one. Take the "o" and compare black vs red "o" - the red "o" is rendered at a much lower resolution. The stair steps are huge, while on the black they are not. There is sharp edges on the black at the expense of color fringes (as expected). Notice on the red, there is no color fringing.

[Nate the great]

Originally Posted by DuckieTigger

I think the easiest way to show is by using a text only ebook with different color words. E.g. some black, gray (75%, 50%, 25%), red, green, blue, purple, brown, and cyan. And then take a picture with a camera good enough to show individual 300ppi pixel. My theory is, that the general outline of each word is exactly the same, except for the color words having all red and or green and or blue on "white".

I can't actually upload the file (it's too big). Here it is in Google Drive:
https://drive.google.com/file/d/1vZs...ew?usp=sharing

I didn't think to test the gray, but the short version is that the black text is very black while at the same time the colored text is significantly faded.

This image is about the best I can do, sorry.

[Nate the great]

Here's a photo showing gray text.
https://drive.google.com/file/d/1yai...ew?usp=sharing

[jgneff]

Thank you for the good pictures, Nate. E Ink seems to have come up with a solution very similar to JPEG chroma subsampling, where the color resolution (chroma) is less than the grayscale resolution (luma). It's a clever way to add color images to books without messing up the resolution of the black text.

Add to that the ability to do full-speed animation (A2 mode at 8 frames per second), but now in color, and I'm starting to think I need one of these.

[DuckieTigger]

Originally Posted by Nate the great

I can't actually upload the file (it's too big). Here it is in Google Drive:
https://drive.google.com/file/d/1vZs...ew?usp=sharing

I didn't think to test the gray, but the short version is that the black text is very black while at the same time the colored text is significantly faded.

This image is about the best I can do, sorry.

Originally Posted by Nate the great

Here's a photo showing gray text.
https://drive.google.com/file/d/1yai...ew?usp=sharing

Thank you, very nice. The display driver on the PB does a excellent job of making colored text look almost as good as black and white. Better algorithm, it appears, than on the phone. And just like any other black and white eink screen, the more more contrast the better. The lighter gray are practically unuseable.

It is too bad that you do not have a maceo lens dslr. Thank you for taking the time to do this much work.

[droopy]

So happy that there is now an color eReader in the market. This means that a color Kobo and a color Kindle are not far behind!

FWIW, I bought my very first ereader about a year ago -- a Kobo Forma.

[Nate the great]

Originally Posted by droopy

So happy that there is now an color eReader in the market. This means that a color Kobo and a color Kindle are not far behind!

FWIW, I bought my very first ereader about a year ago -- a Kobo Forma.

If Amazon really wanted to use this screen, they would have paid to be first.