G-Sync Errata and TCON Artifacts

G-Sync Errata

Early in evaluating the XB270HU, we attempted to recreate the issues we have noted with prior G-Sync panels. The ASUS ROG Swift and all other TN G-Sync panels we have tested exhibited a perceptible flicker when a game engine stalls. This occurs while game content is dynamically loaded or during level load screens. Repeating our worst case scenario tests on this new IPS G-Sync panel resulted in zero perceptible flicker. We even hooked up the oscilloscope and confirmed zero variance in output during stall events. It appears that the G-Sync module firmware has been updated to correct the issue in this display, as the slower response time of IPS alone should not be sufficient to completely eliminate the issue on its own. Either way it's good news for the XB270HU.

TCON (Timing CONtroller) Related Artifacts

Sub-pixel dot inversion related

We confirmed that the XB270HU exhibits the same dot inversion related artifact that we have seen in other recent panels. Patterns that are the same color alternated every other pixel may cause visual banding or color change across that entire row of pixels. The test pages over at Techmind.org can be used to test your own display. Bear in mind that this test is supposed to only produce a flicker, not a complete change in color as the window is stretched horizontally wider across the display. Observe:

The checkered pattern should appear grey in the above image, but here we can see that it has turned green where the pattern has been drawn on a wider portion of the screen. Here is a worse-case example, where I overlaid two test pages in such a way that the patterns which should appear as 50% grey actually appear at green and purple. If you look closely, you can even see that the effect spills across horizontally, imprinting its checkered pattern onto the portion of the desktop icon that falls on the same rows as the miscolored pattern:

It should be noted that this is a TCON (timing controller) issue. It is not specific to G-Sync and seems to be the most apparent in the TCON design that drives 2560×1440 LCD panels. We noted this issue presents identically in the following displays:

  • Acer XB270HU (1440P G-Sync)
  • ASUS ROG Swift (1440P G-Sync)
  • BenQ XL2730Z (1440P FreeSync)
  • Acer HG270HU (1440P FreeSync)

This is not an all inclusive list, as we only sampled a few displays from around the office, but many panels have a hard time with this test. Even the high end ASUS PQ321Q 31.5" 4K LCD exhibits it to some degree (though in a different manner than the 1440P panels listed above).

While this may seem to be a significant issue, it is limited to that specific type of pattern being drawn. We never saw this effect present itself in typical desktop or gaming usage.

Overdrive sub-pixel dot inversion banding

Some ASUS ROG Swift owners have reported an odd artifact related to overdrive. Where there is fast motion in a scene to the point where a range of pixels must change brightness dramatically from one frame to the next, the pixel sections experiencing those rapid changes also contain an artifact in the form of colored bands. One easily noticeable scenario is the muzzle flashes in an FPS. Since the muzzle flash was not present in the prior frame, pixels that are rapidly transitioning to white will tend to do so with alternating red-blue-green striping. Here is an example using the AMD Windmill Demo on an ROG Swift:

The blade is moving from left to right in the above scene. Contrast has been increased in post to make the effect more visible. Starting at the top, note the alternating brightness of the red pixel columns. As the content of these pixels is 'new' to this frame, every other pixel row is brighter. This is an abnormality, as all red pixel columns should be lit evenly. Moving down to what should be the white part of the blade shows a red-blue-green pattern. LCD pixels are normally red-green-blue, but the reason for the reverse coloring here is that each color column is seeing the same alternating bright / dim striping seen with the red-only portion above, only this time all three colors are doing it in tandem. Looking to the left of the blade, another related artifact can be seen, in the form of green stripes within a slight negitive ghost (where the blade was in the previous frame). If you look closely, you will see that these green stripes are on opposite rows as compared to the bright green stripes in the currently drawn blade, indicating that the same sort of striping is being applied while portions of the image are being driven back towards black.

This issue also appears to be TCON and/or TN related, as it could also be seen in the BenQ FreeSync display in this ghosting test image.

I highlighted this issue first so that I could demonstrate that it is *not* present in the Acer XB270HU. Here is the same scene crop, with the same exposure and post processing applied:

As you can see, the red and white portions of the windmill blade are consistently solid, with every pixel column consistently lit in this newly drawn blade / frame.

Here are another couple of shots taken from a different location. In this shot the blade is moving from right to left, with an overlap of ~2/3 of the blade from frame to frame. This means that the left third of the blade will see the striped pattern on the ROG Swift and BenQ FreeSync displays:

…and those same artifacts are not present on the XB270HU:

It appears that as G-Sync monitors continue to make it to market, slight improvements are made that improve the fringe details from older models. This banding correction, whether it be caused by the panel itself or changes to the G-Sync module, along with the removal of zero-frame-rate flicker, is just another reason the Acer XB270HU becomes one of our favorite displays.

« PreviousNext »