Curious to see whether more recent dithering approaches would produce better results. They don't even have to be more resource hungry than the classic Bayer or Floyd-Steinberg dithers!
Interleaved Gradient Noise[0][1][2] comes to mind as an alternative to Bayer, and it can even be approximated quite well with just 8-bit operations! Basically, use the following function to determine your threshold based on pixel position: (x, y) => (142 * x + 79 * y) & 255
And for error diffusion, there are the Ostromoukhov and Zhou-Fang dithering algorithms[3][4], which I also have an JavaScript implementation of on Observable, so you can try it live[5].
For those interested, their readme file has a really great explanation of how e-ink displays work, the ideas behind them, and how they compare to LCDs at various tasks: https://github.com/Modos-Labs/Glider
I'd love to have one, I have glare issues at work and home where this would hopefully help, being unwilling to simply black out the window.
I like having both a sky, and somewhat changing digital artifacts in my field of view.
A minor concern is that there are so few supply chains for the underlying technology. This is a bit like "that screen really came from Samsung like all the others" or the more real world of where disk platters come from (Thailand, for almost everyone, at one point if not continuously)
I have my screen placed on a table that sits right next to a window and use motorized blinds that have slats which can be tilted for adjustable light control. These let me have both a great view and some protection from the direct sunlight.
I know the Crowd Supply part isn't live yet but I wish this had some kind of ballpark cost, at least to an order of magnitude. Are we talking $300 or $3000?
For the complete monitor, a 13.3 e-ink screen alone has a retail cost of around 350-400 USD depending on where you look. There's also the FPGA and other components.
We have commercial displays of similar size available around 800 usd. So, adding up the manufacturing, development and other overheads my guess is it could be around 1000 USD.
The dev kit should cost less if we can source the monitor & enclosure ourselves.
I didn't go through the complete list of supported hardware but if we can get the dev kit work with old kindle or kobo 6" displays it should be an interesting diy to work on.
Given that a 13in e-ink display is 450$ by itself[1], and the FPGA costs about 75$[2], I'd guess north of 600$. I've no thoughts on an upper bound though.
To give an indication, similar e-ink panels (the likely most expensive single component) to what they seem to be using can be found on AliExpress for a bit below 200usd.
13.3" but the github page lists some alternative, larger displays. Supports some interesting interfaces. Seems to need USB PD power so it sounds power hungry. Regular 5 volt USB would be nice, for this to be a good companion display to a Raspberry Pi or the like. Otoh it sounds like the complete monitor will cost around $1K based on comments below.
I'm still mostly interested in an Inkplate 10 though 13.3" is a better size for an e-reader. For a desktop monitor I really want 20" or larger. For E-ink that would be super expensive.
Author here. It does not need USB PD for power, it uses regular 5V USB and consumes less than 1A (usually less than 0.5A). The reason it supports USB PD is for the DisplayPort Alt Mode to allow power + video on a single cable. DisplayPort Alt Mode is part of the USB PD, to do DP over USB it has to support USB PD and communicate to the host through the PD that it's a monitor. Inkplate is a different class of device (has it's own processor, targeting lower frame rate, etc.) and we are not trying to compete with them.
I guess it's not exactly that it needs USB PD, but that it's pretty convenient.
Also, it seems like large panels could be relatively power hungry, at least on peaks, to be powered by a classic 5V USB brick, when the standard is limited from 55mA to 1.5A (rarely) [0].
The CrowdSupply page says it includes an "E-paper power supply with up to 1 A peak current on +/-15 V rail supporting large panels sizes".
How have they managed to get such smooth video/scrolling on an e-ink screen? All the other things I've seen have this jarring black flash every refresh.
The flash kind of "resets" each pixel. You can update without the flash (and some e-readers support this) but artifacts slowly accumulate.
In the video they show a mode where it is B&W while scrolling, but after stopping the scroll it flashes and renders in gray-scale. The narrator mentions several possible modes; noting that 16-shade gray-scale is too slow for any interactive use (including typing), but 4-shade can allow typing and B&W can allow fast scrolling &c.
Patents were never the issue, there was one crazy comment on HN a few years back claiming that it was and somehow it became a mindworm that kept getting repeated. You can prove this to yourself by asking anyone that repeats the claim of patents being the issue, what patent specifically they mean. Invariably they can't answer or say "all of them". The real issue is volume and scale of production and the physics of the technology not being suitable for a broad enough range of solutions to hit the volume needed to get to a higher scale.
Absolute massive fan of E-Ink and quite disappointed it's been on the back burner for so many years. I imagine it being one of the perfect devices for a software developer.
Interleaved Gradient Noise[0][1][2] comes to mind as an alternative to Bayer, and it can even be approximated quite well with just 8-bit operations! Basically, use the following function to determine your threshold based on pixel position: (x, y) => (142 * x + 79 * y) & 255
And for error diffusion, there are the Ostromoukhov and Zhou-Fang dithering algorithms[3][4], which I also have an JavaScript implementation of on Observable, so you can try it live[5].
[0] http://www.iryoku.com/next-generation-post-processing-in-cal...
[1] https://bartwronski.com/2016/10/30/dithering-part-three-real...
[2] https://blog.demofox.org/2022/01/01/interleaved-gradient-noi...
[3] https://perso.liris.cnrs.fr/victor.ostromoukhov/publications...
[4] https://dl.acm.org/doi/abs/10.1145/1201775.882289
[5] https://observablehq.com/@jobleonard/variable-coefficient-di...