A moving pixel grid shimmers on some LCDs. Why? [inversion]

[dclon]

I've noticed an annoying phenomenon on some LCDs where moving a grid pattern causes the grid to shimmer.

For example, take this image: http://protie.sweb.cz/checkerboard_1920_1080.gif
View it at 100% and move it around (or zoomed out, depending on the resize algorithm and zoom level).

Does it shimmer on your LCD?

I wonder what causes it. Pixel response? LCD dithering?

[Chief Blur Buster]

Usually LCD inversion, see http://www.testufo.com/inversion

Check out the lagom/techmind links for a scientific explanation.

TL;DR: It's a rapidly-alternating DC voltages in a checkerboard pattern, to keep the LCD voltage-balanced. Prevents image retention. It sometimes creates a checkerboard or grid pattern artifact. Some displays have this less, others have this more. Usually, this is more visible on faster-response TN displays, than on slow-response IPS displays.

[dclon]

Are you talking about static images or moving ones?

I can't see much, or any, flicker in the Techmind test patterns. At least not when static.

[Chief Blur Buster]

Inversion artifacts sometimes only become visible in certain background colors, so test the different tests at www.testufo.com/inversion by changing the "Inversion Check" setting.

The artifacts you see, may also be caused by dithering (6-bit TN).

[dclon]

I think we're talking about different things then?

I'm only referring to a shimmer when you move the grid around. When static it's static.

If you look at the image linked in the first post, does it shimmer when viewed statically? Does it shimmer when you drag it around?

[spacediver]

I get a moire pattern that shimmers slightly on my CRT, when viewed statically

[Chief Blur Buster]

I think we're talking about different things then?
I'm only referring to a shimmer when you move the grid around. When static it's static.

In addition to inversion-related shimmer (the inversion patterns at lagom/techmind also shimmer when you drag the browser window; try it!) the shimmer occurs partially because of asymmetric LCD response curves. The pixel transition (GtG) curve from white-to-black is not exactly an inverse curve of the pixel transition curve (GtG) of the opposite transition, black-to-white. This asymmetry creates the shimmering.

[dclon]

That sounds like a response time issue. Do you see that on your LCDs?

I see it very clearly on a cheap laptop TN monitor.

There's also something similar on a mid-range+ VA TV, but I think it may be another problem (which was what made me wonder about the whole thing, and whether a comparison to the behavior of a poor 6-bit TN is any indication).

I also checked a CRT just for the sake of completion. I don't see any of that problem there, in all the common resolutions.

(Edit: No (s) strike-through BBCode?)

[Chief Blur Buster]

That sounds like a response time issue. Do you see that on your LCDs?

A response issue, yes.
But not necessarily a response speed issue.

I see way more shimmering on my 1ms TN than on my 8ms IPS monitor, for example.

No LCD monitors can have perfectly symmetrical response curves. Also, the problem happens on non-LCD technologies to a lesser extent, including OLED. Unless the curves are symmetrical (black and white transitioning symmetrically), your whole display's average amount of light being emitted will fluctuate if one direction of transition is faster, slower or otherwise asymmetric curve than the other direction of transition, etc. As you drag window around, black pixels are being replaced with white, and white pixels are being replaced with black. The transitions need to be symmetric in order to maintain the average amount of light being emitted. Think of (black -> dark grey -> grey -> light grey -> white) simultaneously with (white -> light grey -> grey -> dark grey -> black) ... If the transitions are asymmetrical at all, physical LCD crystal molecules moving at different speeds, you'll vary the average light, if the transition curves aren't occuring perfectly symmetrically.

It can improve a lot, with a lot of effort. Some of the better LCD panels shimmer a lot less. Response speed isn't always a good predictor of amount of shimmer, I've seen fast LCDs shimmer less and shimmer more. It has more to do with response curve asymmetry than response speed. This might change when response goes to microsecond levels, but right now even OLED pixels have slower GtG than that at the moment.

[dclon]

I'm starting to think that without seeing it directly, or without a hi-speed video camera, there's no way for us to know what kind of shimmer the other means. There could be multiple causes, and multiple outcomes could be labeled "shimmer".

I'll try more direct testing the next time I have access to other monitors.

Where do you take the asymmetric response graph idea from, BTW?