Fascinating. I found a bug in the panel driver algorithm of my monitor (!)

A certain glitch pattern makes it smear horizontally. But then a white dot on black grid pattern cancels it out. Even just one or two rows of it. ??????

I have 3 similar monitors (27" LG IPS 4K) and it only happens on the oldest one (mirrored HDMI feed to all 3). I tried changing the settings in the OSD menu but nothing made a difference.

I think what's going on is there's some kind of optimization (power saving??) where by default it uses some kind of drive mode that isn't suitable for certain pixel patterns (high horizontal frequency?). And then it detects that pattern in the image to switch to a different drive mode. But the glitch image does not trigger the heuristic, perhaps due to its alternating colors, so it can't display it properly. Maybe the heuristic only considers luminance instead of R,G,B individually???

If I just show the white dot pattern without the glitch image, I still see a subtle change in color for the whole screen when it appears (only needs one or two rows of pixels, but there's a large minimum width, 100px or so), so there's definitely a global mode change it triggers...