Item — HN Reader

I wonder if it's 2560x1440 (as claimed in the upper section), or 1280x720 (as claimed lower). Given the lower resolution is 1/4 of the larger, and the product description diagram shows quarters.. perhaps it's 4 displays merged into the final project (perhaps it can be further embiggened).

4096 colours seems limited though.. the 13.3 linked in January [discussion](https://news.ycombinator.com/item?id=21996326) looks much more impressive - although the NDA and lack of definitive numbers makes it hard to compare.

4 comments

It is probably 2560x1440 with four different colour subpixels per addressable pixel, giving you an image resolution of 1280x720.

Forgive me if I'm misremembering details, but I think the way color e-ink displays work is they have the black/white "ink" capsules that are pushed up/down by magnetic charges, just like a regular black/white e-ink display, except for in the colored ones, each "pixel" also has a red or green or blue color filter over it (one of each in a cluster). But that means the pixels per inch drops because now 3 black/white "pixels" have to work together to be one single color "pixel"

I'm not really sure how they measure resolution with this, that could still just be a mistake. But I do know e-ink displays sacrifice resolution for that colour with current technology.

So, assuming that’s what it is, subpixel resolution is now being advertised as the “true” resolution? Because when I buy a 1080p display, I’m expecting 1920x1080 addressable pixels. Whether that’s implemented as 3840x2160 (2x2 subpixel) or 5760x1080 (3x1 subpixel) is not my concern. And if I bought a display advertised as 3840x2160, and my computer told me it was 1920x1080 (because it’s a 2x2 subpixel layout), I’d be upset.

In Apple ][ hires graphics, the 280 pixel-wide display area was really 140 pixels with the extra resolution coming from sub-addressing the RGB pixels on the display. On a black and white display you got the full 280 pixels but in color you would get weird color fringing happening. Even more fun was that one byte of memory was mapped to 7 "pixels" with the high bit indicating how the RGB sub-pixels would be split into the left and right halves of the pixel. Technologically it was rather cool, although it was a challenge to program against.

Wow, I never understood until reading this comment that this is why Apple II graphics often had that fringing effect. Those visual artifacts I remember from The Oregon Trail finally make sense.

Unfortunately, there's plenty of precedent around this, especially with Pentile displays and similar.

Bad precedent, to be honest... Pentile was infamous for its inferior image quality on things like text and fine lineart.

It reminds me of the LCDs on digital cameras, some of which use the same trick to reduce costs --- natural photos don't look any different, which is why it gets a pass, but text has a noticeable graininess to it as a result.

> natural photos don't look any different

They can! Not frequently, I'll grant. Once I was using Lightroom and was very confused by a pixelated looking edge of a yellow flower against a blue sky, the edge of which moved diagonally across the pixels. Eventually I realized it was the mosaicing! I actually ended up switching to RawTherapee for that picture, the AMaZE demosaicing algorithm handled it much better in that particular case. Lightroom in the last few years added a high quality demosaicing mode that you can selectively enable, and it handles the picture correctly. It's pretty slow though.

In any case, reporting resolution using the single-color "pixels" is probably more justified for an image sensor (with its array of square light detecting sensors), than it is for an LCD screen, which is usually a square array of non-square sub-pixels.

Pentile is more complicated, since for most of them the advertised number of pixels matches the number of green elements, but not the number of red or blue elements.

People buy digital cameras this way, have for years.

Somewhat true, but the splitting of pixels / bayer filter work is something that gets corrected later in either post/RAW processing or otherwise.

When we apply that work to screens, our eyes can't really post-process and correct for these errors.

There are several ways to do color, with wildly different implications.

But calling subpixels pixels would be borderline fraudulent.

At least one HN story in the last month was an eInk type product, either them or a competitor, able to position the tiny cells into more than one pair of modes, the current ones is up or down this one had some colour space like rgb as well. So it wasn't 1/3 the pixel density for a colour pixel.

>4096 colours seems limited though

That's what the original Amiga chipset had. 4bit/RGBchannel definitely is not as good as 12bit/ch but it isn't bad. DPI can make up for it, that's how print works.

But at 31.2" and those resolutions, the DPI isn't great.

If you click on "Specification" in the lower section, it says 2560 (H) × 1440(V). And if you download the manual it also says 2560x1440. So maybe just a typo?