Excellent video. I thoroughly enjoyed all the detail you went into here! It’s so neat to see these techniques used to push the limits of these old LCD panels.
Since we have the whole video in advance (we can look forward in time) and we can also measure the response function of the pixels, it would be possible to preprocess the whole video to make it look even better (less ghosting).
Any screen is not fully digital. The PCM (multibit) signal is completely not understandable for human beings and it always must be converted into analog as the sum of the bits with the different value for each bit. Most of LCD screen have error correction chip which also simplifies the values of the input signal beside removing of most unwanted types of signal and this chip must be removed if we want gain more exact scale of lumination levels than the predefined in the chip one (e.g. 15, 31, 63 or 255 levels). Converting the binary PCM signal into the "unary" PWM signal (which is the most native form of the LCD signal and does not require any extra chipa beside AM signal) is a very nice way to increase the signal precision of any LCD screen without making (or buying) any extra chips. The PWM signals, while displayed on the screen look almost the same as the AM signals, because LCD matrixes very strongly tend to minimize the differences of signal inside the pixels if the input signal resolution is not euqal to the matrix resolution. Difference between PWM and AM on a CRT screen would be much more visible than on the LCD one, because the differences minimizing in each pixel is not as strong as in LCD screens, and the borders between the extreme values would much more visible on a CRT screen (but acceptable if the picture would be seen enough far from screen). In CRT screen, where the resolution of signal was often not equal to the matrix resolution, there were strongly visible borders between lines even inside the pixels. In LCD screens, the differences of values inside every pixel are almost completely removed and all surface of every pixel displays almost the same value as the average, but the differences still exist. Even in the LCD screen, the differences in signal level in every pixel could be almost perfectly minimized, but they will never disappear. In this case, the signal horizontal resolution is 255 times bigger than the matrix resolution, because the value of every matrix pixel is encoded not as the amplitude value which can differ, but as a proportion of length of the zeros and the ones, which must be equal to the intended value of lumination.
Holy shit though, the 2nd order noise shaper looks insane. I have to say, I think I prefer the 120Hz over 240 - 240 seems way darker, and loose a lot of detail in darker areas. But still, incredible what can be achieved by a 1 bit display!!
Is it possible to output composite video to a display like this? I watched a video on 90s LCD pocket TVs and one of them had a screen like this but without the back part so light could get through so you could watch it and they had the simpsons playing on it.
now add another script on it to calculate motion vectors and run another script to compensate in order to get rid of motion tearing. there also may be some calculations done on a sperate emulation of the video that is about to be played ahead of time to deliver the correction in real time. (ofc this would need a better processor but could possibly be implemented). Also not sure if the motion blur is coming from the fact that im not viewing in person. Anyways, the staring at dust for 12 minutes was fun to watch on a mono screen
This is very awesome! I often wondered how those worked, and thought I've seen a little flickering which would indicate temporal dithering. I think plasma televisions used it too. Another characteristic of these LCDs was their slow speed to go from one density to another, and you can see some ghosting caused by that in your movie playback. In one way, this characteristic helps the temporal dithering, but with moving objects it also hurts. I wonder if you could characterize the rate of density change versus pixel input and use that to push pixels to your desired state faster by holding the 0 or 1 value longer (in other words, use a higher or lower PDM value for a bit)? It would be kind of like a PID controller, but instead of for position, for pixel density.
I know that STN displays like that are pretty slow... I'm wondering if you could use the trick that modern IPS panels use, where if you change between gray levels, for a short while you 'overdrive' the panel, as in, you go past the point you were supposed to go; effectively you add a portion of the differential between input pixels to the output pixels. So say you go from 100 to 200 between 2 24-bit frames, your 120fps output would go 100-240-200-200-200.
Not sure I grasp exactly what you're describing (I'm still awestruck by what was already accomplished, here), but it sounds a bit like what I was thinking in seeing ghosting between sharp changes in brightness... Since we're already adding extra buffers for error, etc. maybe when, say, a pixel goes quickly from black to gray, it could be first shown white for a bit to overcompensate? (Is that what you mean by overdrive?) This is really quite impressive, either way. I'd've never expected this much from STN.
I think there are indeed monitors implementing the overdrive mechanism on modern TN panels to reduce greyscale response time, and there is a setting in the on-screen manual to turn it ON/OFF. Not exactly sure which monitor has it, but I have seen that on several LG and Asus monitors.
Can this be applied to an SBC for output of its desktop interface? I've been trying to find a way to drive monochrome/grayscale LCDs to create a low power portable project, but I am only ever able to find color TFTs...
Yes and no. Yes I have done this before and you can check my GameBoy Advance LCD downgrade mod video which is basically what you described: converting from a standard RGB display interface designed for TFT-LCDs to interface with grayscale STN-LCDs. But also no because this won't actually save you much power: First of all your SBC is probably the biggest consumer. Like those small 1.x inch screens usually only consumes about 0.1W of power. But SBCs usually consumes way more than that (1W-20W) so LCD shouldn't be your biggest concern. Second, assuming you get rid of the SBC or able to let it stay in sleep most of the time, when 0.1W is becoming a lot, chances are my grayscale LCDs driving mechanism also won't pay the bill. It's just too power hungry to do all these fancy modulation in software to extract last bit of performance out of those poor LCDs. You would better off use existing monochrome/ grayscale LCDs with built-in controllers (like ST7565, UC1698u) screens, and they use way less power. Or in some extreme you may also want to consider SHARP memory LCDs which use even lower power.
Learn and enjoy ! I do it all the time just not in this scale (mostly arduino, esp32, little displays etc) 🫣 Just like other people have their passions like Cars, photography, boats, and spend their money and time…
I want to know what the phone is at the start of the video. Is it one that will work in the USA? What are its specifications? It's been a long time since a smartphone with a real keyboard was available here, and the telcos got rid of them by claiming "nobody wants them" - by deliberately crippling the specs on any model with a keyboard. Less RAM, less storage, lower resolution, worse cameras, slower CPUs etc.
It is a Fujitsu F-07C, the probably only cellphone that runs Windows 7. It uses Intel Atom Z600 processor and it was released in 2011. TBH that's mostly just collector's item today.
I aways thought that why game boy can display nice graphics with this mind of Black white display, why can a casio watch do that........ Game dispay has 4 shades of gray.....
