Show HN: Building a 42-inch E-Ink frame for generative art
eliot.blog
eliot.blog
For comparison a 42" full-color 60fps TV with remote, speakers, wifi, etc. etc. is $140.
It seems like the patent holder could be making a lot more money if they dropped the price.
Does Rakutan/Amazon/Pocketbook really pay a similar cost/size ratio for the panels on their ereaders? I hope not!
You just reminded me, a few years ago i wanted to buy a transparent screen for a DIY project. The company said they couldn't sell it to me because they're b2b.
I think it was both a volume thing and a tax thing.. and possibly a liability thing.
Because it's a lot easier and cheaper to manufacture large LCD panels at scale than E-ink film.
>It seems like the patent holder could be making a lot more money if they dropped the price.
I don't know any company or shareholders that would say no to making more money so if this would actually be true then it would, but that's not the reality.
The FUD that large e-ink displays are expensive because of some patent conspiracy needs to stop.
Large e-ink screens are expensive because manufacturing yields are very low and so are sales volumes.
Source: worked with e-ink on products with big and small displays
Edit: So I'm saying the truth and getting down voted for it? Fine, then feel free to keep believing whatever you think is the truth. No point in discussing it further.
The company that owned the e-ink patents couldn't scale properly until they were acquired around 2010: in an alternate timeline where development open enough in the early 2000s, we might have ended up with eInk display modules large enough to be assembled into incrementally larger and more profit driving devices
Even now you see that with eInk store tags for example: imagine if the profit they're driving now had arrived 20 years ago and gotten re-invested
That said e-ink price tags aren't new, them being stupidly cheap (https://www.aliexpress.us/item/3256804488587338.html) and increasingly wide spread in availability is the relatively new part.
If I knew an easy way to disable them, I’d be doing it.
Yes e-ink has high wastage, and yes demand is low, but having had a strong arming expensive patent holder didn't help the situation. The patents are actually expiring/expired but at some point the damage is already done.
A great corollarly to e-ink is 3d printers: they were very expensive, very low volume products that needed to recoup expensive development costs... but when the 200 lbs gorilla sitting on the market in the form of patents started to die down, a lot more players start to iterate, which unlocked advancements, which made things cheaper and more accessible, which improved demand, which then incentivized more advancements... and now we have $400 printers that outperform $10,000 printers from not that long ago.
Most display tech starts high wastage, low demand. Patents aren't a guaranteed death, but e-ink was just a bit too far off the path for absolutely required innovations to overcome the added resistance. In the mean time other technologies got better: sunlight performance of non-eink displays has improved dramatically for example.
Now the moment has probably passed and e-ink is doomed to stay a niche sideline product as a result.
Can you talk more about this? My phones and laptops are still as bad as ever in direct sun. Did I miss some major development...?
In reality this is one of those slow burn deals where the improvements in anti-glare coatings, bonding between digitizer and displays, and brightness come slowly enough that it's easy to take for granted just how awful old displays used to be.
We've come a long way from the days of hazy plastic suspended millimeters above the actual display, which in turn had its own coating with a different refractive index leaving two layers of reflections to deal with
I actually have a M2 MacBook Pro and a Pixel 7, both of which get decently bright, but neither of which is really usable under direct glare. And in XDR mode, the SDR colors get super washed out. HDR is a gimmick as far as I can tell.
I'd trade either for a usable transflective display or a regular matte screen from early 2000s Dell. I don't think the answer to glare is to just make the display brighter. That's liking shining a flashlight into your eyes so the sun doesn't seem as bright...
I tried a 3m anti glare coating, and that worked well for keeping out direct light, but it darkened the screen so much it was barely visible at max brightness.
The Steam Deck had some etched glass thing that seemed a little better (maybe the iMac too?) but I still wasn't able to use it in direct sun.
The only display I've used so far that was actually comfortable outside in the sun are e-ink ones, with matte screens also somewhat usable in shade on a sunny day. My M2 MacBook? Useless outside and barely usable under office florescent light. I hate the Apple glare. Makes movies look nice in the dark but useless for day to day office work, IMO.
> That's liking shining a flashlight into your eyes so the sun doesn't seem as bright...
... that's exactly the point of e-Ink: with minimal power you get brightness that rivals the sunlight you're holding the device in.
With infinitely efficient OLEDs, we'd just crank the brightness on the lit pixels high enough to match the brightness of a white piece of paper in sunlight and have a better e-ink: unfortunately in the real world that'd generate more heat and take more power than mobile devices can afford right now.
At the end of the day we have both matte and glossy displays with reflective characteristics that rival old school matte displays because of improvements made across the stack... but nostalgia makes us assume things were very different than they were.
What do you mean by this? Are you saying the Macbook has a special display mode that only activates when it detects sunlight (as in lumens? or?)?
> With infinitely efficient OLEDs, we'd just crank the brightness on the lit pixels high enough to match the brightness of a white piece of paper in sunlight and have a better e-ink: unfortunately in the real world that'd generate more heat and take more power than mobile devices can afford right now.
Are you talking about the contrast ratio? I meant more that high brightness with a direct light source into the eyes can be pretty uncomfortable compared to reflected diffuse sunlight. If there's a way to get the contrast of LEDs to a similar readability of paper & e-ink without needing to blast my eyes, that'd be great. The few OLED displays I've used seem to have that more of that effect (still not as nice as e-ink), but I don't have a monitor that size either.
> At the end of the day we have both matte and glossy displays with reflective characteristics that rival old school matte displays because of improvements made across the stack... but nostalgia makes us assume things were very different than they were.
Like what? Are you saying my current Macbook is actually more readable in sunlight than old matte displays, I just don't realize it...? If so, hmm, I'm dubious but I will take it out again and try to compare it again without nostalgia, as much as I can. Or is there an objective measure of sunlight readability I can refer to?
Yes, if automatic brightness is on, the screen will exceed its standard SDR range in daylight, just like the iPhone. Otherwise you need special software to manually increase the SDR brightness into that higher range, maxing out at 1600 nits.
> I meant more that high brightness with a direct light source into the eyes can be pretty uncomfortable compared to reflected diffuse sunlight
The discomfort would come from a mismatch in brightness: a 2000 nit screen sounds incredibly painful to look at because in normal conditions your pupils are adjusted to an indoor room and you're looking at a 2,000 nit screen, which is terrible.
But ambient light on a clear day is something like 30,000 nits, so your pupils are already significantly narrowed: a "mere" 2000 nit screen will still look kind of dim if anything.
e-Ink sits around 40%-50% reflectivity and still looks fine in those conditions.
(Blue light, contrast, and the nature of the reflections themselves also play a role so it's not as simple as "more nits is better", but you need to be in a certain ballpark of brightness to even play, which is why I'm saying old devices aren't in the running here.)
_
> Like what? Are you saying my current Macbook is actually more readable in sunlight than old matte displays, I just don't realize it...? If so, hmm, I'm dubious but I will take it out again and try to compare it again without nostalgia, as much as I can. Or is there an objective measure of sunlight readability I can refer to?
It's hard to capture in a single measure because there's two parts to the equation:
- reflections from the surface
- being bright enough for your eyes to make out details at all
Back in the day the brightness would be roughly equal for both a glossy and a matte option, so matte would be strictly better.
Now brightness has advanced enough that a newer glossy panel + modern anti-glare coating + modern bonding techniques are. enough to overcome the difference in reflections vs the old screens.
You should look at even the glossy non-Retina (which was almost universally hated) vs the glossy Retina to see how dramatic the changes in anti-glare tech were at some points in the last decade: https://cdtobie.wordpress.com/2012/06/18/reduced-reflectance...
A modern matte screen would have the advantage if it had the same brightness, but afaik no one actually makes a matte screen that gets as bright as the new XDR Macbooks do: the closest are all glossy displays too.
Brightness aside, though, a huge thing for me and glossy screens is the specular reflection. Even when it's not the sun, having a tree or a sign or the person behind me reflected in my screen is very distracting.
Do they not make a high-brightness matte screen?
I imagine that the goal is to have the sunlight itself provide the lighting for the screen, not just making the existing screen brighter, no? I don't want to be staring at something that's competing with sunlight, I want it to reflect the existing sunlight that's already hitting the display. Granted this may not make much difference in practice but... that's a huge range of luminance to compete with for a theoretically low-power device.
But you still have a point. Even if a LED display has "only" a 2 or 3 day battery life, that's still pretty huge to mere hours.
How do smart watches (like the Garmins with 1-week batteries) do it?
Don't worry about that. Posts that go against the general opinion always get downvoted at first. But if it is insightful and not demonstrably wrong, the upvotes will win.
There are probably many techniques and processes that would be tried and developed if it wasn't locked up by patents.
In tech, yes, 20 years is a lot, but that's more of an exception.
And imho, it's slowing down: A 20 years old computer, like an Athlon 64, today, could probably run most of the usual basic workloads (mail, text, word processing, browsing) with up to date software. It would be slow, but it would work. An 1983 years old computer in 2003 was just obsolete and completely unfit to run any of the usual 2003 software.
Without patents, there is far less incentive to seek progress. What's the point for a for-profit organization to spend resources on Research, just to see your innovation immediately copied?
And if said for-profit organization still innovate, it has a strong incentive to keep its innovation as secret as possible. Patents by contrast, incentivize sharing, on one hand it makes the innovation public knowledge, and on the other licensing agreements usually exist.
And in the end, 20 years is really short compared to human history.
Are patents an ideal solution? Hell no, this system is too easily abused (patent trolling, patent thickets, etc), and feels really wrong at times (patents on life-saving drugs for example). But so far, that's the best compromise we manage to mostly agree on in terms of worldwide IP recognition.
Even niche early adopter users at current prices don’t love the tech
For the record, the "patents" think has no links to back it up either.
But to answer your question: E-ink screens are less than $100 (depending on the size), that's already pretty cheap. They're only expensive when compared to LCDs. LCDs are produced at a rate of billions per quarter*, with ~6 billion smartphone owners, and lots of them having multiple LCD devices e.g. laptop/desktop screen, work computer's screen, supermarket kiosk, tablet, smartwatch, car screen, etc.
In contrast, e-ink screens have basically three applications: e-readers, e-notes (which is basically just an e-reader with a stylus), and supermarket pricetags. E-readers/e-notes are a luxury item that you don't need if you already have a smartphone/iPad. There's just no economy of scale for e-ink; not compared to LCDs.
Because there's no economy scale or obvious new markets to expand into, there's not much budget for R&D, so the entire field moves slowly.
If e-ink was x100 cheaper, we could use it in x1000 more use-cases
Giant E-ink screens are expensive because they're made by fusing together 4 smaller e-ink screens (4 of 10.3" or 13.3" screens IIRC), but the process is expensively done by hand because there's simply not enough demand to warrant automating the process. There's just no money in giant 1Hz monitors.
For example? What untapped use cases would be if e-ink displays which can only show B/W images and refresh at 1HZ, unable to show full motion video, and suffer from terrible ghosting, would be cheaper?
I think you're deluding yourself here. Cheaper prices won't magically fix the major limitation of this technology that hinder its expansion outside of its existing niches: e-readers and price tags. All the other display markets have been conquered by LCD, MIP and OLED already because those technologies work better in those cases and e-ink works better where it already is.
I don't know if there is a big market for any of that stuff, but certainly there's not a lack of ideas. A lot of things that aren't worth it at the current price point become viable if the display weren't so expensive.
We'd all like many things, but that doesn't mean there's a mass market for them to warrant significant investments that would guarantee a decent profit for the investors.
It seems that you have a e-ink hammer so everything looks like a nail to you, but sometimes pen and paper or a plastic white-board and a felt pen are better and cheaper than a big e-ink display that needs to be recharged.
And E-ink smart watches have been made in the past but sold much more poorly than OLED one so manufacturers dropped in favor of OLED. MIP is a thing though in some Garmins.
The core problem hasn't changed, though: LCD covers more use-cases than e-ink does; it even has better battery life than e-ink in some scenarios! (E-ink screen refreshes take a lot of energy more energy than LCDs, although they only refresh once per interaction instead of 60 times a second)
E-ink sounds cool (and is cool), but it's not practical. It's the zeppelin of electronic displays.
I've never quite understood this sort of thinking. Companies aren't perfect omniscient systems - the people in charge of various companies make poor decisions all the time, based on the same human biases we all have. So yes, it's entirely possible they could make more money that way, but executives simply hear "Lower the price" and run away. I'm not saying that's certainly what's happening, but this implication that a company could possibly only do the thing that's most beneficial to it is... strange, and very common thinking.
That is to say, it’s not that they’re charging high prices out of greed, but that there’s no competition forcing prices down.
But first someone has to invent it. Nobody’s gonna spend a few thousand or million dollars to invent something unless they get their money and more back, so that’s the point of patents.
And if they patent something and don’t capitalize on it enough, that’s their loss, because they had to detail their invention in their patent application and that’s public information. Patents last only 20 years and for most industries, that’s not very long.
(That said, I think software patents are a different beast. Because 20 years in software engineering is like 100 years in civil engineering, for example, since software engineering is like a few thousand years younger discipline than civil.)
How so? Nobody would bother investing in R&D if after all their labor and toil, someone else would be legally allowed to copy it for free and get all the profits without having had any of the expenses and risk that come with hardware R&D.
The (US)patent system has its flaws, but without one it would be even worse.
People say this all the time without presenting any evidence, as if it is common sense. I don’t actually think this is correct! Of course the character of investment would change. Instead of a small number of large investments, we would see a large number of smaller investments. If your competition is churning out some widget and you can make a $10,000 investment in your factory to lower your production costs below theirs and out compete them, you will make that investment.
My point is that I believe in this case, natural market mechanisms are adequate to stimulate investment. Meanwhile proponents of patents must justify the enormous cost to society caused by heavily restricting innovation with legal penalties, building a 20 year moat around every new invention. That massively hampers incremental process, and raises the cost required for investments to succeed. Without patents, smaller incremental investments will pay off, but there will be much more competition. Unless you think markets cannot function without the massive government intervention which is intellectual property restrictions.
And heck don’t take my word for it. The folks at the free market Mises Institute hate IP restrictions:
But it’s not black and white. Some people will do things for $10, others $20, others $50. We all have a different price.
With patents, it’s basically offering $50 so you get the most people doing it as possible.
If we did away with patents, it would be like offering $20. Sure, R&D will still happen… just less.
Innovators who succeed at getting a patent might get more money for their innovation. But now only one person or org can legally make incremental improvements to those ideas. Without patents, the value per innovation might be lower to an individual innovator, but there are more innovations to be made because everything is on the table.
You are not just saying "you get $50 for this innovation" you are additionally saying "this innovation is now off limits for 20 years", and there may be hundreds of potential innovators who would have invested in improvements which have been legally barred from doing so.
My classic example is the 3D printer. They were $50,000 when first sold under patent, then 13 years later they were $25,000 still under patent. Then the patents expired and within 3 years they were under $2000. Ten years after patent expiration they were $250. So the price dropped only 50% in 13 years under patent. Then with no patent it dropped 90% in three years. In ten years it dropped 99%. It takes a lot of innovation to make a $25,000 3D printer work for just $250, and absolutely none of that innovation happened with the benefit of patents. Innovation will always happen! We do not need massive government intervention to make it happen.
If you’ve ever tried to DIY something, it’s so so much harder if you can’t find someone who has done it already.
So to me, the people who may have made 3D printing cheaper aren’t exactly in the same class of innovation and risk.
It may be your favorite example but in this case it's a very poor one. You keep harping it as if it proves the point that patents are always the biggest problem holding all tech back, but in this case you're comparing apples to pickup trucks and omitting the forest from the montains.
The price of 3D printers dropped because the demand for 3D printers is huuuuge and they're also very easy to manufacture meaning in that case it actually was the patens holding the technology back and not the manufacturing challenges, but E-ink film is both tricky to manufacture and the demand for this kind of limited technology is very niche beyond e-readers and price tags meaning there's no economies of scale.
I built a 3d printer with a work colleague over 10 years ago in his garage using aluminum rails, stepper motors and an Arduino. I can't build E-ink film in a garage.
With your example you'd also think that the reason we don't have mass adoption of 3nm chips everywhere must be because of TSMC's patents and not because that sort of tech is tricky to master at scale.
Your 3d printer example just does not work in this case so please be so king as to reflect on the matter and update your viewpoint accordingly.
What makes you think there would be so much competition on driving down prices of large 1Hz B/W displays that can't display motion videos properly and suffer from terrible ghosting, if patents weren't a thing? Do you think there's large display-tech manufacturers knocking on e-inks door wishing to enter the market for such niche displays and e-ink just keeps saying 'NO'?
If that were such a lucrative untapped market then someone like Samsung or LG would have swallowed up e-ink holdings already or drowned them in lawsuits to squeeze a cross lichenizing deal out of them, but it's not.
Here's the hard to swallow pill: The consumer market for large e-ink displays is incredibly niche that it's not worth entering by second players, and better just leaving one incumbent player deal with their niche little sand-pit.
You're looking for problems that aren't there while also grossly over-estimating the size of this super-niche market.
It's a nothing-burger that, due to lack of knowledge and understanding of the market and technology, HN loves to spin into an emotionally charged straw-man every few months outraged they can't buy TV sized e-ink displays at Wallmart for $199, without even considering the technical challenges, product feasibility, economies of scale and market situation of such a fantasy product, thinking e-ink must be evil or stupid for such a product not already existing.
The truth is, there is no conspiracy and the e-ink market has already self-regulated itself: The mass market e-ink displays like for price tags and e-readers have already gotten cheap enough, while big e-ink displays are very expensive because there's next to no market for them to warrant further optimizations of economies of scale or interest from other players to even bother competing regardless of patents.
That is a really odd assumption. There are multiple companies making electrophoretic displays and multiple different technologies, of which only one is really commercially effective so far in a very limited niche space. Why don't you think physics is the limitation? I mean, don't you think above would be the equivalent of me saying that the reason we don't have daily flights to Mars is because SpaceX is the only manufacturer of Falcon Heavy.
A surprisingly large fraction of seemingly intelligent, educated, and otherwise decent folks have no clue beyond a superficial understanding. Even though superficial appearances may suggest that they would be smart enough to figure out what they do and do not know.
Then again, if that was really it, surely there'd be a market for people who want a 42" e-ink display and are willing to accept some proportion of bad pixels in exchange for a deep discount. Which really shouldn't matter much, particularly for applications where distance-viewing is the expectation.
Try $3000 :)
> Driving boards are not included in the display module package. E Ink Salt kit is designed to support this module
> Salt Driving Board - $500.00 - https://shopkits.eink.com/en/product/detail/SaltDrivingBoard
Edit: Someone has your dream display already :) https://www.etsy.com/listing/1463146123/43-4k-dnd-tv-table
How is that a realistic feature? Who is losing power and being happy that they spent a few thousand more, nearly 40x, than a 36" photo quality print, just so the image can remain while their lights are off?
(and of course it is a lot easier to change than a printed photo)
I understand being eco friendly, but that’s an absurd, entirely negligible, justification.
I would claim the reflective display is the killer feature.
EInk doesn't have a backlight. So it's easy to see in glaring light. It's also not disruptive if so, you want to use it as a digital picture frame and you don't want light from the display bothering your eyes day and night.
EInk is a more peaceful display technology, and some people put a lot of value on that.
Transflective LCD is other option bit also pretty hard/expensive to get a big screen...
[1] https://www.groglass.com/product/artglass-ar-70/
[2] https://www.framedestination.com/prod/sh/ultravue-uv70-pictu...
Some E-ink panels can be somewhat susceptible to UV light and perform better under a filter. You will sometimes find warnings in data sheets about refresh performance in direct sunlight, and danger of long-term permanent damage. Some of E-ink's signage products have, I think, filter layers built-in.
But also if you're using a passepartout like in this project, and the frame will hang in sunlight and you're not sure how the paper will perform, it's worth springing for the UV protection to avoid yellowing over time.
ha-card::after {
content: "";
background-image: url(/local/visionect-dither.png);
background-repeat: repeat;
position: absolute;
height: 100%;
width: 100%;
display: block;
mix-blend-mode: multiply;
}
EDIT: Oh, I think I see the source of confusion:
> A matrix tuned for blue noise, such as those generated by the void-and-cluster method, produces a look closer to that of an error diffusion dither method.
If you use a dithering matrix it is by definition ordered, yes. Because it is tiling that matrix in a regular pattern. The matrix has been randomized via a blue-noise method. The test image seems to use a 16 by 16 matrix, which is very small - the seams are clearly recognizable.
However, "blue noise dithering" typically refers to generating a blue noise texture as large as the image to be dithered. This article does so, for example:
https://surma.dev/things/ditherpunk/
But yeah, I can understand the confusion now. The wording of the wiki article could be clearer.
PS: in case your curiosity has been piqued, here are some more articles on blue noise and dithering:
https://blog.demofox.org/2018/01/30/what-the-heck-is-blue-no...
https://blog.demofox.org/2019/06/25/generating-blue-noise-te...
It would have been interesting if the author did additionally apply a low pass (Gaussian blur) filter to all the result images, not just to one homogeneous one in the beginning.
This doesn't get around all the other annoying bugs in the firmware of course.
Especially at night, I find the backlight makes it painfully obvious that it's just a TV and I'd much rather have something like e-ink which blends into the surroundings.
Even the “explore technology” link has little extra detail. I don’t even know if it’s backlit.
I have the white frame on the bezel against a light-coloured wall and when displaying art it's much less imposing on the room than an empty black screen. Obviously it uses more power than standby but it's also quite a bit less than a dynamic screensaver.
Don't know that I'd get another one; the bugs in Samsung's software are annoying (I just want it to display the art, or a picture from a single HDMI input, how hard can they possibly make it?) but perhaps that's standard for modern TVs.
You can also have it not produce images at night or when you are not around the house.
Lots of ways to save energy. The impact of an image every few hours or whatever is nanoscule compared to your transportation and heating needs.
(Also if your apartment uses electric resistive heating, fire away with your GPU, you're just producing images in the process of producing heat instead of passing it through a resistor. It's no less efficient.)
At about 300W of GPU + 120W of PC, that is 420W * 0.5hrs = 0.210kWhr. This is a rounding error on my monthly electric bill. About six cents.
I've spent more energy than that likely just sitting and reading HackerNews this week.
You should upload a massively complicated where's waldo like scene and have waldo appear in different places each hour.
Have a portrait with lots of subtle details but they change slowly over time. Ex. girl goes from earing-less to small earings to big earings. Just have lots of small details subtly change but at a rate which makes them hard to detect.
The mfg from the article has a 13.3" screen that would be fun to try a mod with: https://shopkits.eink.com/en/product/detail/13.3''ePaperDisp...
and impressive 1600 x 1200 pixels
If you're into looking at different dithering techniques, there are a few interesting ones compared in this old CodeGolf question (disclaimer: I'm the one with the Fortran answer)
https://codegolf.stackexchange.com/questions/26554/dither-a-...
I imagine the game keeping track of whose turn it is and being able to give voice commands to make moves.
Then they could play games over a longer period of time, maybe make one move per day.
the Lilygo T5 with Touch screen + a lipo battery would have all the hardware you need
just need to port chess to the platform after that https://www.hackster.io/Sergey_Urusov/arduino-mega-chess-d54...
figuring out NAT traversal to get them talking to each other from each other's homes might be tricky/fun, but there may be ways around that...
https://www.lilygo.cc/products/t5-4-7-inch-e-paper-v2-3
It would be great to be able to buy this in a size that would mount easily into common photo frame formats.
Ideally with a battery that could last months and optional power.
If I was retired it's the sort of project I'd happily waste time on.
https://shop.invisible-computers.com/products/invisible-cale...
You could render the current state of the chess game to an image and serve that image on a URL on the internet. Then you can connect the display to that image:
https://www.invisible-computers.com/invisible-calendar/image...
Admittedly that's still a lot of work from your side, and a little applet to keep running...
For that price, you can buy a 42" color printer.
All you need is a high quality RGB light sensor or three and an accurate way to dim a backlight. None of this seems very hard. Sure, it wouldn't show the shape of shadows or whatever, but it would still be highly effective. Why can't the image just go black when I turn the lights off?
It was daylight in a well lit bright room. The frame and background were just right to look like modern simple poster framing. The minute sun set, of course it stopped looking right. Thats when I realized we had a TV after all.
My partner pointed out the art had changed at least two times across daylight hours. I had just blanked that out.
At one time there apparently was this: https://mono.frm.fm/en/shop/ But the price was crazy. There's not much out there in terms of appropriately priced digital picture frames.
My wallet stopped at 13.3" though :)
Update for those who saw me post it before: I recently charged the battery for the first time since hanging it in March, which met my expectations.
Shout-out also to the people at Halbe Rahmen, the best picture frames in the world.
Mine is still wired but I'm thinking of using ESP32 + deep sleep.
esp32 + deep sleep + external RTC is what this is using. I posted some of my experiences with getting it as low as I could here: https://news.ycombinator.com/item?id=37663913
Here's a few more pics: https://imgur.com/a/NoTr8XX
I found the mention of Blue Noise Dithering in this project quite interesting! I've not put a lot of effort into picking the nicest possible dithering method (just a basic Floyd-Steinberg) for the embedded article photos, partly because the whole newspaper look didn't seem to warrant it, but it might be worth taking stock of what the latest on dithering is.
Also the album link in the other comment.
I suspect people expect everything to be magic now.
When displayed on the frame you can barely see the dithering. Interestingly you see the jpeg compression artifacts a lot more
Some devices actually clear by outputting the invert of the image. Maybe this will be faster
LED panels got cheap because we built factories that pump out a lot of them.
Here's the data sheet for the ITE IT8951, a frequently-used ASIC for this purpose that supports basic partial updates as well: https://www.waveshare.net/w/upload/1/18/IT8951_D_V0.2.4.3_20...
Here's a sourcing price: https://www.win-source.net/products/detail/ite/it8951e-64-dx...
It's basically a fairly simple SPI interface (or here a SPI wrapper around an internal Z80 protocol). The rest of the board is power supply handling and a DRAM framebuffer chip, external to the driver ASIC on SPI as well. I wrote a custom driver for this in Rust for my project, and it only took about a day despite a few leaky-abstractions oddities in how it communicates over SPI.
The chip itself is basically buffer handling, image processing (features like JPEG decode, resampling and some LUT mechanisms) and a waveform generator. Others will have some IP blocks for, say, a HDMI frontend. Any decent chip company can crank this out pretty quick. There's easily 10+ product lines on the market.
For simple things you may not even need one and can drive the panel from an MCU directly.
There are more expensive, more advanced drivers that implement more complicated and higher-performance (say, refresh rate) update schemes or I/O though.
The waveforms, and the algorithms that create waveforms, used to drive eInk displays at reasonable speeds and produce high quality images are highly proprietary and very difficult to develop. In theory it’s easy to make an eInk display an image, but doing that in a reasonable period of time, when transitioning from potentially any starting state, and handling the crosstalk between pixels, makes creating good images on an eInk display pretty hard to get right.
https://dheera.net/projects/einkframe/
I really want to upgrade to at least 31" but it's rather expensive.
Separately in the next irritation of this project I'm hoping to use an ESP32 instead of an RPi Zero to power it, and use the deep sleep mode so it can periodically update but sleep through the rest of the time, allowing a few months of battery life with no wire sticking out.
Even more ideal would be putting a thin strip of solar panel on the top edge of the frame to keep it charged, and use some kind of supercapacitor to power the ESP32, though I don't know if the parts for that exist, especially solar panel that is 1cm wide.
The best option I found so far is a timer shield that can wake up the Raspberry periodically: https://www.pishop.us/product/sleepy-pi-2-power-management-s...
https://framelabs.eu/en/product/artframe-31/
I use the 31.2" panel because it is 40% cheaper and has higher DPI, but the 42" panel is also possible.
There's some projects that forego a driver ASIC entirely and drive the waveforms directly from a MCU, although I guess this large panel has a different voltage domain.
Or maybe you could hide a Qi charger into the wall?
Look to what hobbyists are hacking to find commercial opportunity.
er, what?
AI art is not generative art, it is reductive "art". It requires samples of the art style you wish to cop, and it always produces something artificial and less than the sum of its parts.
On the other hand, generative art is procedural, and represents the thought process of a human being.
Better luck making meaning.
But aren't the prompts used also representative of a thought process? The chosen network architecture and the choices of images used to train also represent thought.
Walking into a convenience store and buying milk also requires a thought process, but it's not creating art. Walking into a home store and selecting from various linens to assemble a complementary set requires a creative thought process, but it's not art. The closest analog is commissioning a piece from an artist-- that requires a creative thought process about art but that doesn't make the commissioner the artist.
Most commercial art directors have more granular control of their project's outcome than people generating their AI art have over theirs. While they are credited for their direction and curation of those multiple pieces to create a unified project, they are absolutely not credited for the individual pieces-- the artists that created those pieces are.
Maybe deliberately curating a collection or collaging AI-generated art that, through juxtaposition or some other method, elicited a response or communicated something emotionally that each piece individually could not. But that's not what's happening here. People are commissioning pieces of art from a computer which makes them out of other people's art.