Well I was messing around with grayscale, and this happened. I have no idea what is going on here. Does the screen have some strange physical quality that is causing this? Left is supposed to be white, middle is supposed to be gray, right is supposed to be black. However, the middle appears to be cycling between dark gray and super-white. Also the center of the screen has more wild contrast changes than areas further away from the center.








EDIT: I have attached the program I used to observe this strange effect. It will only work on 15MHz calculators. Use +/- to tune the frequency, CLEAR to quit. Don't press - too much or it will probably crash! I can't easily give the source because it is part Axe, part Axiom, and part raw hex edited. But if you do look at the disassembly of it, please don't mind any unoptimized code or code that seems out of place, I never actually meant to release this.


EDIT 2: PLEASE READ THIS BEFORE DOWNLOADING

Alternating pixels between black and white at the same speed as the LCD controller alternates between positive and negative voltages means that those pixels receive only the positive pulses or only the negative pulses, effectively creating direct current just like the controller is supposed to prevent.
In summary: Don't use this effect unless you want to screw up your screen.

Perhaps I should give more information about what is going on here. The screen is split into 4 columns, each being updated at almost exactly the LCD's refresh rate (about 58Hz). It's pretty obvious what I'm doing with the left and right columns; they're being set to white and black respectively each frame. The middle two columns are each being toggled between white and black each frame and are always opposite colors of each other. From other testing I know that the fact that these two columns are always opposite colors is not the source of the strange behavior.

I find it very odd that this only happens when trying to produce perfect 3-color grayscale without dithering. When sending dithered 3-color grayscale or 4-color grayscale that is either dithered or undithered at the same update rate, you get just what you would expect, solid and correct colored columns.

Would you care to send an executable or a source for us to look at?
(I've already BSOD'd my calculator, so there's nothing worse you can do. )

I'll edit the video once it's fully processed by Youtube.

STN LCDs have to be continuously refreshed by alternating pulsed voltages of one polarity during one frame and pulses of opposite polarity during the next frame.

The T6A04 used does just that, and so do all other controllers for this kind of LCD. And for good reason:

This pin is used to periodically invert the polarity of the power supply to prevent dc charge build-up within the display.

STN displays, which require the pixel voltage polarity to be periodically reversed to prevent damage due to DC charge accumulation

The LCD must be driven by alternating current (AC). Direct current (DC) will cause electrophoresis effects in the liquid crystal and will degrade the display.

ON/OFF of Power Supply and Signal - Driving liquid crystal molecular by DC current may cause serious damages to LCD including disorder of alignment and electrical decomposition.

The contrast voltages must be symmetric about VM to avoid a DC component across the liquid crystal, which in turn would damage the LCD or shorten its life.

Alternating pixels between black and white at the same speed as the LCD controller alternates between positive and negative voltages means that those pixels receive only the positive pulses or only the negative pulses, effectively creating direct current just like the controller is supposed to prevent.
In summary: Don't use this effect unless you want to screw up your screen.