See, now this is what I don’t get. I can understand that the eye sees a light at one column or another, but not BETWEEN the columns by varying degrees.
Any LED is slightly ahead or behind, but by one column, not a fractional one. It doesn’t look the way you expect.
I can also understand if each row were offset by one column, so the top LED would be 7 away from the bottom one, and there would be a pronounced lean. But it doesn’t look like that. The lean is, at most, 2 columns, not 7.
Has anyone experienced this phenomenon? If you haven’t, it won’t be easy to understand what I am seeing.
Believe me, I know exactly what you are seeing, and the explanation is correct. Think about this: what happens if each “frame” (the time it takes an LED in one row to move to the left once) is one second, and the time to switch from row-to-row is 1/1000 of a second? You would agree that what you would see is a vertical line moving across the display, one second at a time, right? Now, start speeding up the scroll (frame) speed. At the point where the frame speed is equal to the row-to-row transition speed, you would see a 45° diagonal line, right? Now increase the row-to-row speed to 10x the frame speed. Now you will see a line tilted by 1/10 of the width of one LED.
Oh, and I can prove it to you: Don’t let your eyes track the LEDs (use masking tape to mask off all but one column) - the tilt effect will disappear.
I’ve noticed this on scrolling LED panels on trains here - it is because the horizontal rows of LEDs are not all being shifted at the precise same time - it’s an artifact of the same perceptual widgetry that makes us see the display as being composed of moving letters at all, and not just a bunch of lights that are turning off and on (you can often break the illusion by framing a small portion of the display with your hands - then you can see it as just a grid of flashing lights, and the text becomes very hard to recognise).
Our brains perceive the transition of LED P turning off and LED Q turning on as movement between P and Q - if this happens at the same time as LEDs V and W (stacked above), we see a short vertical line moving. If P and Q transition ever so slightly earlier than V and W, the only way our brains can make sense of this, and still see it as a connected line, is if the line is tilted slightly.
The displays do this because of the way they address and refresh the LEDs in a scanned raster-type pattern, but I’ve seen this slant effect exploited deliberately, in a display (I forget where) that obviously had absolute control over the timings of the small changes - the text could be made to ‘lean’ in either direction, or wobble like jelly.
My wife has a digital camera that is capable of shooting 1000fps video - I might see if I can snag some footage of one of the LED displays on the train - at that capture rate, it should be possible to see the raster update in progress.
In spite of the various helpful illustrations, I am having a hard time envisioning how the illusion works.
I have no doubt that masking off all but a vertical column will remove the illusion. I wonder what would happen if I masked off just two or three adjacent ones?
I do remember a gadget, ca. 1980, a hand-held wand with a single column of LEDs. When you held it vertically and waved it back and forth, a sensor would detect the motion and the direction, and turn on the LEDs at just the right time and sequence to make letters. This required persistence of vision, as the letters were being formed only in your mind. I have no problem with this concept.
But the situation here seems different. Mangetout, if you can come up with some video, would the best way to show this be to repeat a short segment, starting slowly, then speeding it up to normal? That’s what I’d do, but I cringe at the thought of asking the store owner, who doesn’t know me, if I can set up a tripod and camera right in front of his counter and busy main entrance. It might take some chutzpah on my part, so before I try that, I’d like to see what others can dig up.
You’re right - this is a different thing to floating displays that rely on persistence of vision effects.
I’ll do what I can - it might take a week or so, as my wife and kids are away visiting relatives this week. I’m not at all sure that the illusion will work at full speed in a video format, just because of the frame rate limitations, but by slowing down a 1000fps shot of it, it should at least be possible to understand the cause of the illusion.
In the meantime, I’ll see if I can sketch a description of (my understanding of) how it’s creating the effect.
I need to point out that I read a third of the way through this thread before I realized the OP meant a Point Of Sale display, not a Piece Of Shit display.
Get your mind out of the gutter and onto the counter, Sir.
Thanks, Mangetout. One way or the other I want to get to the bottom of this. I was hoping some Doper was close enough to the hardware design to speak from experience, or perhaps someone familiar with optical illusions could explain. I wish Jerry Andrus were still around – he’d be the first person I’d ask.
If I get up the balls to ask the store owner, I’ll see what I can video, too. I have a pretty good, semi-pro camera with a manual shutter speed adjustment.
I’ve already explained why this happens, and I’ve already stated that I have designed several of these signs. Do you want me to explain the electronics in detail? It’s not really a hardware phenomenon - it’s the result of the fact that your eye is moving, and the lights are flashing, not on all the time.
Sorry, I have yet to experience an “aha!” moment, and your explanation just doesn’t click with me. That isn’t to say it’s wrong, but I don’t see it.
My eye is moving? Not if I stare at the display with no eye movement, yet I still experience the illusion. Flashing lights? Why does a flashing light appear to be in locations where it’s not? Why does the sequence of off/on give the illusion of motion between the fixed lights?
OK, let me try one more time.
Say that the sign is 7x100 LEDs. Then, let the scroll speed be 2 seconds (it takes 2 seconds for a character to travel the length of the sign). This means than each LED in a row is illuminated for (2/100)/7 seconds or 2.8mS, each time it is “ON”. This also means that each LED in a vertical column is flashed on 2.8mS after the one below it. Now, since we’ve said that the signed scrolls at 50 LEDs/second, this means that on average, each column is on for 20mS. But, the LEDs in a column are illuminated sequentially, separated by 2.8mS! Since 2.8mS is (2.8/20) = 14% of a column width, each LED will appear to be 14% (or 1/7) of a column displaced from vertical.
Using My drawing for coordinates, I want to be sure we are talking about the same lighting sequence. Are the LEDS turned on from bottom to top (a column completed) then left to right, one column at a time?
Example using (Row,Col) notation:
(7,5)(6,5)(5,5)(4,5)(3,5)(2,5)(1,5) = one column at the right
(7,4)(6,4)(5,4)(4,4)(3,4)(2,4)(1,4) = next column one to the left
(7,3)(6,3)…
OK, we are on the same wavelength for the lighting sequence. Next question: is each LED lit, then extinguished simultaneously as the following one comes on? (I don’t know why that matters, but I’m just trying to understand your description.)
Or do some LEDs stay on while others are lit? And if so, how long and which ones?
I’m trying to understand if an entire column, once lit, should be viewable as a solid vertical line, even if only briefly, before the next column is touched. Or is the appearance of a vertical line something formed by the human eye/mind?
This.
If you were to slow the operation of the sign way down, you would see the lights come on and go off sequentially from top to bottom (or bottom to top, it would just cause the letters to lean in the opposite direction). No two LEDs are ever on at the same time in a given column. This is called Multiplexing.
I believe it’s supposed to be “leaning” because the display is set to an italic font. Unfortunately, italics can never look very good on a display with only 7 rows of lights… it only leads to people having discussions about LED display programming, tearing, fractions of seconds, and refresh cycles.
Not the convenience store managers intention, I’m sure.
I don’t think so. The lean is only there during horizontal scrolling (and horizontal scrolling always leans). If you look at the static pix linked earlier, you can see some dots that can be called a font enhancement, maybe even a crude serif, here and there. Otherwise I think the lean is not intentional, but just something that happens due to the electronics behind it. I doubt if the store owner cares if it leans or not.
