LG 34UM67 sRGB Data and Bench Tests

For color accuracy, we test before and after calibration. For calibration, we use SpectraCal CalMAN with our own custom workflow. We target 200 cd/m2 of light output with a gamma of 2.2 and the sRGB color gamut, which corresponds to a general real-world use case. We use an i1 Pro provided by X-Rite. All measurements use APL 50% patterns except for uniformity testing, which uses full field.

LG 34UM67 Pre/Post Calibration
Pre-Calibration,
200 cd/m2
Post-Calibration,
200 cd/m2
Post-Calibration,
80 cd/m2
White Level ( cd/m2) 201 198.7 79.3
Black Level ( cd/m2) 0.2056 .2153 .0977
Contrast Ratio 978:1 923:1 811:1
Gamma (Average) 2.18 2.21 2.21
Color Temperature 6558K 6548K 6482K
Grayscale dE2000 2.94 0.38 0.99
Color Checker dE2000 2.49 1.24 1.39
Saturations dE2000 2.14 1.07 1.17

Before calibration, the LG 34UM67 has a slight blue tint to the grayscale but nothing too noticeable – especially for gaming purposes. Tweaking the OSD settings to 53/50/47 RGB gives a result reasonably close to the ideal 6504K color target. The grayscale errors are all under 4.0 dE2000, which is potentially visible but not overly so, with an average error level of 2.9 dE2000. The gamma curve isn’t great, starting high and ending low but with an average of 2.18 that’s close to our 2.2 target, so things can definitely be improved. Moving to colors, there are a few larger errors of nearly 5.0, mostly in the yellows and oranges. Some of these are due to the gamut falling slightly higher than sRGB, leading to some oversaturation of green and red.

Post-calibration the gamma and RGB balance are almost perfect. The average grayscale dE2000 falls to well below 1.0, which is invisible to the naked eye. Colorchecker and saturation accuracy improves as well, though there are still colors in the 4.0 range. Again, it’s mostly shades of yellows, oranges, and some greens that cause problems, which unfortunately tend to be the worst colors to have wrong for imaging professionals. Overall it’s a good monitor, and the target audience clearly isn’t going to be imaging professionals, so with or without calibration it will do well for gaming, movie watching, and other general tasks.

Changing to 80 cd/m2, the calibration results remain pretty consistent. The dE2000 numbers are slightly higher, but if the small change in accuracy is a concern then potential buyers would have already passed on this display. Only the most finicky of regular consumers might find something to complain about.

It’s also worth quickly discussing some of the other color modes, just because certain ones can be so far off that it’s a wonder anyone would even consider using them. LG offers four picture modes (Photo, Cinema, Reader 1, and Reader 2). Photos has a strong blue tint with average grayscale dE of 6.4 and many values nearing 10.0, though colors aren’t quite so bad averaging closer to 5.0. The Cinema mode is pretty close to the Custom setting, so while it’s tinted blue the grayscale dE is 2.3 while the colors average close to 4.0, with skin tones often falling into the 6.0+ range. Reader 1 and 2 are supposed to be more like print, with the results being heavily red biased with limited blue, and minimum black levels are much higher (2.5 cd/m2). The resulting grayscale dE2000 of 10.8/8.7 and average colors of 7.5/6.0 however are not particularly useful.

And that sums up why NVIDIA didn’t bother with supporting specialized color modes on their G-SYNC module: doing one color mode properly is generally more useful than supporting multiple incorrect color modes. While some people might appreciate the ability to quickly switch between various color modes, most just set up a display for everyday use and leave it be. Most named presets other than “standard” or “custom” end up being bullet points more than anything useful.

LG 34UM67 Brightness and Contrast LG 34UM67 Display Uniformity
Comments Locked

96 Comments

View All Comments

  • xthetenth - Wednesday, April 1, 2015 - link

    For working purposes I would not consider a 16:x 4k an upgrade from 3440x1440 at all. I would be trading sufficient x space to have a third item up or a wide item and a narrow one up at the same time in return for a small amount of y space that doesn't make a meaningful difference. Past roughly 1200 pixels tall, 21:9 is by far the best aspect ratio for work. By 1200 pixels, there's plenty of y space that added information by increasing y space is facing seriously diminishing returns, while x space is starting to go from two pretty wide windows to three windows side by side, which is still giving significant returns.

    Of the current selection of monitors, I would definitely choose 3440x1440 to keep for 5 years, and spending that much tends to come with a very nice, calibrated screen. A $300-$400 2560x1440 isn't the same quality screen.
  • wweeii - Tuesday, March 31, 2015 - link

    Theoretically even if you drop below 48hz it shouldn't be all bad.
    Between 16 and 24 fps, you can just triple the refresh rate, 48-75hz would work just fine without tearing.
    Between 24 and 37 fps, you double the refresh rate, so no problem either.

    You would only have a problem between 37 to 48 FPS, which is unfortunate.
  • Soulwager - Tuesday, March 31, 2015 - link

    But AMD isn't doing that, and the VRR window is too small to do window shifting. If you want to display every frame on time you need a max frame interval needs to be greater than your frametime variance plus double the minimum frame interval.
  • Soulwager - Tuesday, March 31, 2015 - link

    You can test input lag with inexpensive hardware, for example, an arduino with native USB that emulates a mouse input and measures a subsequent brightness change with a photoresistor.
  • Ryan Smith - Tuesday, March 31, 2015 - link

    If you could, please shoot me an email.
  • Soulwager - Tuesday, March 31, 2015 - link

    Done.
  • OrphanageExplosion - Wednesday, April 1, 2015 - link

    Is there a link with an explanation for this somewhere so we can all take a look at this idea?
  • Soulwager - Thursday, April 2, 2015 - link

    Yes, here's a forum post: http://forums.blurbusters.com/viewtopic.php?f=10&a...
  • jjj - Tuesday, March 31, 2015 - link

    This gave me an idea, a Cypress PSoC board instead of Arduino could also work and maybe you could make a similar device to test touch responsiveness in phones and tabs. Cypress makes touch controllers so maybe they would help you out with some coding to enable you to test touch responsiveness. You could at least try. Guess Arduino started with Atmel chips and Atmel is also one of the major touch controller players so you could try to ask for their help too.
  • cbrownx88 - Tuesday, March 31, 2015 - link

    Yes - please email him! lol

Log in

Don't have an account? Sign up now