Introduction

Very recently we have reviewed the LG 34GK950F display. Some sections of this review for the G model contain some of the same content where the models are the same. We have also updated this review since we re-tested the 950F model and the v2 firmware in Feb 2019 (which was after this 950G review was first published)

LG are continuing to invest in their gaming display line-up through their 'UltraGear' range. The latest offering is a 34" UltraWide format display with a range of high end gaming features. The model in question is the 34GK950, and there are actually two versions of this screen available. The 34GK950F (our earlier review) features AMD FreeSync support, while the 34GK950G (this review) features NVIDIA G-sync support. There are a few different features and specs between the two, but fundamentally they are based on the same design, the same underlying panel, and mostly the same specifications.

Anyway, let's focus on the 34GK950G now that we've already tested the F model. This screen is 34" in size with a curved UltraWide format, and 21:9 aspect ratio. This size and format has become increasingly popular for day to day multi-tasking, multimedia and gaming and provides a very interesting alternative to the mass of 16:9 screens on the market. The display uses one of LG.Display's latest 'Nano' IPS technology panels (the LM340UW5 for those interested). This new IPS technology is effectively the latest generation of IPS and most notably features support for a wider colour gamut than traditional LED backlights, extending to cover 98% of the DCI-P3 colour space which is about 35% more than traditional sRGB screens according to the spec. This gives boosted colours for gaming and multimedia to help support the latest titles, HDR content and potentially some wide gamut photo/creation work. The panel offers a 3440 x 1440 resolution and a high refresh rate of up to 120Hz, via an overclocking feature that we will talk more about later. The G model supports NVIDIA G-sync for variable refresh rates as well. There's a whole load of other gaming extras and things to talk about throughout the course of this review. While there haven't been any ultrawide IPS screens with 1440p resolution, high refresh rate and FreeSync until the launch of the F model, there are quite a few alternative models with high refresh rate and G-sync support out there. This is the first model to include a wide colour gamut though which is probably the main differentiator compared to the others.

Love our work? If you would like to help support our continued work please consider making a donation no matter how small or large. Thank you.

Specifications and Features

The following table gives detailed information about the specs of the screen as advertised:

The 34GK950G offers a typical range of connectivity with 1x DisplayPort 1.2 and 1x HDMI 1.4 offered for video connections, and an additional 2 port USB 3.0 hub, with the ports located on the back of the screen. This is a bit different to the F model which features a DP 1.4 connection which can allow support for a higher refresh rate of 144Hz. Because of the need to use the ver 1 G-sync module here, only DP 1.2 is provided which means it can only support up to 120Hz maximum. You at least don't need to worry about having the latest generation of DP 1.4 graphic card to power this screen, as older generations with DP 1.2 out will be fine. This G-sync module is also limited to only HDMI v1.4, and not v2.0 like that used on the F model.

The screen has an external power supply and comes packaged with the power cable and brick you need. A headphone output connection is provided as well for audio pass-through.

Below is a summary of the features and connections of the screen:

Power Consumption

In terms of power consumption the manufacturer lists a typical usage of 75W and <0.5W in standby. We carried out our normal tests to establish its power consumption ourselves.

Out of the box the screen used 78.2W at the default 70% brightness setting. Once calibrated the screen reached 51.3W consumption, and in standby it used only 0.5W. We have plotted these results below compared with other screens we have tested. The consumption (comparing the calibrated states) is a little bit higher than some of the other screens in this size range, including the F model, although not by a massive amount. These include models like the Dell Alienware AW3418DW, Acer Predator X34 and Asus ROG Swift PG348Q.

Panel and Backlighting

Panel Part and Colour Depth

Like the F model, the LG 34GK950G features an LG.Display LM340UW5-SSA1 Nano IPS technology panel. The 'Nano' in Nano IPS is largely a marketing term, used to differentiate this latest generation of IPS technology from older panels. However, there is one significant difference and that is the extended colour gamut that this new panel offers thanks to the KSF LED backlight unit used (see below). The two new LG 34GK950 models are the first ultrawide IPS screens we have seen with a wide gamut, so that immediately differentiates them from the competition.

The colour depth on this G model is limited to 8-bit due to the G-sync v1 module that is being used. It cannot offer support for 10-bit colour depth like the F model can, and this is a limitation imposed on the screen by the G-sync module. If you specifically want to work with 10-bit content, or have support for the latest 10-bit capable games then you will be limited here on the G model.

Screen Coating

The screen coating is a light anti-glare (AG) offering. It isn't a semi-glossy coating, but it is light as seen on other modern IPS type panels. Thankfully it isn't a heavily grainy coating like some old IPS panels feature and is also lighter than modern TN Film panel coating. It retains its anti-glare properties to avoid too many unwanted reflections of a full glossy coating, but does not produce too grainy or dirty an image that some thicker AG coatings can.

Backlight Type and Colour Gamut

The screen uses a KSF LED backlight unit where a KSF Phosphor layer (K2SiF6 doped with Mn4 for the chemically minded) is added to the backlight to improve the colour gamut. This allows the screen to offer a wide 98% coverage of the DCI-P3 colour space, which is fast becoming the reference for displays now that HDR is such a big topic and focus. This coverage equates to approximately 135% of the sRGB reference space according to the manufacturers specs, so it is about 35% wider than most traditional screens. If you want to read more about colour spaces and gamut then please have a read of our detailed article.

Backlight Dimming and Flicker

We tested the screen to establish the methods used to control backlight dimming. Our in depth article talks in more details about a previously very common method used for this which is called Pulse Width Modulation (PWM). This in itself gives cause for concern to some users who have experienced eye strain, headaches and other symptoms as a result of the flickering backlight caused by this technology. We use a photosensor + oscilloscope system to measure backlight dimming control with a high level of accuracy and ease. These tests allow us to establish

1) Whether PWM is being used to control the backlight
2) The frequency and other characteristics at which this operates, if it is used
3) Whether a flicker may be introduced or potentially noticeable at certain settings

If PWM is used for backlight dimming, the higher the frequency, the less likely you are to see artefacts and flicker. The duty cycle (the time for which the backlight is on) is also important and the shorter the duty cycle, the more potential there is that you may see flicker. The other factor which can influence flicker is the amplitude of the PWM, measuring the difference in brightness output between the 'on' and 'off' states. Please remember that not every user would notice a flicker from a backlight using PWM, but it is something to be wary of. It is also a hard thing to quantify as it is very subjective when talking about whether a user may or may not experience the side effects.

100%                                                     50%                                                     0%
 
Above scale = 1 horizontal grid = 5ms

 

Brightness and Contrast

At the full brightness setting in the OSD the maximum luminance reached a very high 370 cd/m² which was only just shy of the specified maximum brightness of 400 cd/m² from the manufacturer. There was a good 286 cd/m² adjustment range in total, so at the minimum setting you could reach down to a fairly low luminance of 83 cd/m². This should be low enough for most people including those wanting to work in darkened room conditions with low ambient light, although not quite as low as some screens can reach. A setting of 12 in the OSD menu should return you a luminance of around 120 cd/m² at default settings. It should be noted that the brightness regulation is controlled without the need for Pulse Width Modulation for all brightness settings so the screen is flicker free.

    

We have plotted the luminance trend on the graph above. The screen behaves as it should in this regard, with a reduction in the luminance output of the screen controlled by the reduction in the OSD brightness setting. This is a linear relationship as you can see.

The average contrast ratio of the screen was measured at 908:1 which was good (although not amazing) for an IPS-type panel. This remains pretty stable across the brightness adjustment range as you can see from the graph. Some modern IPS panels from LG.Display have improved contrast ratios and are rated at 1300:1, but this particular panel is rated at 1000:1 only and so the results here are to be expected, if perhaps a little low.

Testing Methodology

An important thing to consider for most users is how a screen will perform out of the box and with some basic manual adjustments. Since most users won't have access to hardware colorimeter tools, it is important to understand how the screen is going to perform in terms of colour accuracy for the average user.

We restored our graphics card to default settings and disabled any previously active ICC profiles and gamma corrections. The screen was tested at default factory settings using our new X-rite i1 Pro 2 Spectrophotometer combined with LaCie's Blue Eye Pro software suite. An X-rite i1 Display Pro colorimeter was also used to verify the black point and contrast ratio since the i1 Pro 2 spectrophotometer is less reliable at the darker end.

Targets for these tests are as follows:

• CIE Diagram - confirms the colour space covered by the monitors backlighting in a 2D view, with the black triangle representing the displays gamut, and other reference colour spaces shown for comparison

• Gamma - we aim for 2.2 which is the default for computer monitors

• Colour temperature / white point - we aim for 6500k which is the temperature of daylight

• Luminance - we aim for 120 cd/m², which is the recommended luminance for LCD monitors in normal lighting conditions

• Black depth - we aim for as low as possible to maximise shadow detail and to offer us the best contrast ratio

• Contrast ratio - we aim for as high as possible. Any dynamic contrast ratio controls are turned off here if present

• dE average / maximum - we aim for as low as possible. If DeltaE >3, the color displayed is significantly different from the theoretical one, meaning that the difference will be perceptible to the viewer. If DeltaE <2, LaCie considers the calibration a success; there remains a slight difference, but it is barely undetectable. If DeltaE < 1, the color fidelity is excellent.

Default Performance and Setup

Default settings of the screen were as follows:

We measured using ChromaPure software a 132.7% sRGB gamut volume coverage which corresponds to 97.8% of the DCI-P3 reference and 70.2% of the Rec.2020 reference. This is very close to the specified 98% DCI-P3 and 135% sRGB coverage which was good news. For reference we have also provided a comparison of the screens colour gamut compared with the other common Adobe RGB reference space:

One thing we were a bit disappointed with on the G model was the lack of an sRGB mode in the OSD menu. This meant that there was no preset mode which could offer an emulation of the smaller sRGB space. You are always going to need to operate the screen in its native wide gamut. That's probably not going to be a problem for many users who just want to use it for gaming and multimedia, where they will probably prefer the more bright and vivid colours that this wide gamut offers. But if you were wanting to work with any sRGB content at all for other uses, photo editing etc then it cannot offer that. This is because of the use of the G-sync module, and lack of an additional scaler/electronics on the G model.

Default gamma was recorded at 2.1 average with a 4% overall deviance from the target which was not too bad as an average. The gamma was most off in the darker tones where it dropped down to 1.98. White point was measured at a too warm 5708k being 12% out from our target. Luminance was recorded at a very bright 289 cd/m² which is too high for prolonged general use, you will need to turn that down. The screen was set at a default 70% brightness in the OSD menu but that is easy to change of course to reach a more comfortable setting without impacting any other aspect of the setup. The black depth was 0.32 cd/m² at this default brightness setting, giving us a reasonable static contrast ratio for an IPS-type panel of 901:1. We have seen some IPS panels reach up to around 1200 - 1400:1 and the F model had reached closer to 1000:1 as well out of the box. We will see if that can be improved through our calibration process in a moment. Colour accuracy measurements here can be ignored as they are comparing the screens wide gamut output with an sRGB reference so will be skewed as a result. We will measure the sRGB preset mode in a moment where they can be considered as part of this factory calibration measurement. Testing the screen with colour gradients showed smooth gradients with only minor gradation evident in the darker tones. There was no sign of any colour banding which was good news.

Color Temp Setting Measurements

We carried out a test of the various Color Temperature modes as well out of interest and the results are shown above. The warm mode was slightly warmer than the default 'custom' mode but not by much. Medium was a little cooler than our normal 6500k target, and cool was (as promised), cooler at 8113k. When using the 'Custom' mode you can alter the RGB channels in the OSD menu as well which will allow us to calibrate the screen more accurately and correct that white point.

Gamma Mode Measurements

We also measured the average gamma in each of the 3 modes out of interest, and because the default setting had produced a gamma curve which was a bit off our target of 2.2. On the face of it, mode 3 offered a closer average gamma at 2.2 but if you look at the more detailed table below you can see that the gamma curve fluctuates, being too low in darker tones and too high in lighter tones. This will need correcting ideally through the use of a colorimeter.

Above: Gamma mode 4 more detailed measurements

 

Calibration

We used the X-rite i1 Pro 2 Spectrophotometer combined with the LaCie Blue Eye Pro software package to achieve these results and reports. An X-rite i1 Display Pro colorimeter was used to validate the black depth and contrast ratios due to lower end limitations of the i1 Pro device.

We reverted to the 'Gamer 1' mode in the menu which used the full native gamut of the backlight. Unlike the F model there is no sRGB preset mode on the G model, and so you cannot restrict this colour space from the monitor anyway. We already had access in the OSD menu to alter the RGB channels while in the 'custom' colour temperature mode, so that we could correct the white point. Gamma had been fairly reliable out of the box in the default 'mode 2' setting, so we left that as it was. The OSD settings were adjusted as shown in the table above, as guided during the calibration process and measurements. These OSD changes allowed us to obtain an optimal hardware starting point and setup before software level changes would be made at the graphics card level. We left the  LaCie software to calibrate to "max" brightness which would just retain the luminance of whatever brightness we'd set the screen to, and would not in any way try and alter the luminance at the graphics card level, which can reduce contrast ratio. These adjustments before profiling the screen would help preserve tonal values and limit banding issues. After this we let the software carry out the LUT adjustments and create an ICC profile.

Average gamma was now corrected nicely to 2.2 average (leaving a 0% deviance) and sorting out the 4% deviance and some of the differences across grey shade gamma that we'd seen out of the box. The white point had now been corrected to 6513k which had fixed the fairly noticeable 12% deviance we'd seen before. Luminance had been improved thanks to the adjustment to the brightness control and was now being measured at a far more comfortable 120 cd/m². This left us a black depth of 0.13 cd/m² and a similar static contrast ratio of 902:1 which was ok for IPS panel technology but a bit under the 1000:1 spec. Colour accuracy of the resulting profile was excellent, with dE average of 0.4 and maximum of 1.0. LaCie would consider colour fidelity to be excellent. Testing the screen with various colour gradients showed smooth transitions with some minor gradation in darker tones and some slight banding introduced through the correction of the gamma curve. You can use our settings and try our calibrated ICC profile if you wish, which are available in our ICC profile database. Keep in mind that results will vary from one screen to another and from one computer / graphics card to another.
 

Calibration Performance Comparisons

The comparisons made in this section try to give you a better view of how each screen performs, particularly out of the box which is what is going to matter to most consumers. We have divided the table up by panel technology as well to make it easier to compare similar models. When comparing the default factory settings for each monitor it is important to take into account several measurement areas - gamma, white point and colour accuracy. There's no point having a low dE colour accuracy figure if the gamma curve is way off for instance. A good factory calibration requires all 3 to be well set up. We have deliberately not included luminance in this comparison since this is normally far too high by default on every screen. However, that is very easily controlled through the brightness setting (on most screens) and should not impact the other areas being measured anyway. It is easy enough to obtain a suitable luminance for your working conditions and individual preferences, but a reliable factory setup in gamma, white point and colour accuracy is important and some (gamma especially) are not as easy to change accurately without a calibration tool.

From these comparisons we can also compare the calibrated colour accuracy, black depth and contrast ratio. After a calibration the gamma, white point and luminance should all be at their desired targets.

Default setup of the screen out of the box was only moderate. There was a small 4% deviance in the gamma curve which was not too bad, but the white point was too warm by 12% at 5708k. That part of it is thankfully pretty easy to correct through some simple OSD changes though so not a major issue. The contract ratio was decent, but not great for an IPS panel at 902:1 after calibration. It was a bit behind the F model as well (1000:1) which could be down to the panel setup or perhaps just variances in samples of the LM340UW5 panel being used. As we said earlier, there's no sRGB emulation mode on this G model either which was a bit of a shame as you're limited to using the full native 135% sRGB gamut at all times as a result. Perhaps an issue for some colour critical / general work if you need to work within the standard sRGB colour space as it would lead to oversaturated colours.

When it comes to black depth and contrast ratio the screen performed moderately for an IPS-type panel, with a calibrated contrast ratio of 902:1. We have seen some modern IPS panels start to reach up to 1200 - 1400:1 or so as you can see above, and the F model had also reached 1000:1 as well. You can see that IPS cannot compete with VA technology panels though which typically reach up to 2000:1 - 3000:1.

Viewing Angles

Above: Viewing angles shown from front and side, and  from above and below. Click for larger image

Being based on exactly the same panel the viewing angles were the same as the F model. They were very good as you would expect from an IPS-type panel overall. Horizontally there was very little colour tone shift until wide angles past about 45°. A slight darkening of the image occurred horizontally from wider angles as you can see above as the contrast shifted slighting. Contrast shifts were slightly more noticeable in the vertical field but overall they were very good. The screen offered the wide viewing angles of IPS technology and was free from the restrictive fields of view of TN Film panels, especially in the vertical plane. It was also free of the off-centre contrast shift you see from VA panels and a lot of the quite obvious gamma and colour tone shift you see from some of the modern VA panel type offerings.

Above: View of an all black screen from the sides. Click for larger versions

When viewing dark content from an angle in a darkened room you will see some characteristic "IPS glow". It seems that the level of glow you will see depends on which side of the screen you are looking from. From the left the glow seemed to be a little less than some older generation IPS ultrawide screens. The glow was a little less pale and had a slight purple hue to it. On some of the older 34" ultrawide IPS screens like the Dell Alienware AW3418DW, Acer Predator X34 and Asus ROG Swift PG348Q the IPS glow was more white and more noticeable than here on the 34GK950 when viewed from that side. However, when viewing the screen from the right hand side the glow was more noticeable, having a more typical pale and white colour to it. Overall this would indicate that the IPS off-angle glow on darker content is probably going to be comparable to other Ultrawide IPS screens.

This type of glow is common on most modern IPS-type panels and can be distracting to some users. If you view dark content from a normal head-on viewing position, you can sometimes see this glow as your eyes look towards the edges of the screen. Because of the sheer horizontal size of this 34" panel, the glow towards the edges is more obvious than on small screens, where there isn't such a long distance from your central position to the edges. Some people may find this problematic if they are working with a lot of dark content or solid colour patterns, especially in lower ambient light. In normal day to day uses, office work, movies and games you couldn't really notice this unless you were viewing darker content. If you move your viewing position back, which is probably likely for movies and games, the effect reduces as you do not have such an extreme angle from your eye position to the screen edges. The glow effect was a little less than on flat ultra-wide screens as the curved nature created a smaller angle between your eyes and the edges of the screen.

Panel Uniformity

We wanted to test here how uniform the brightness was across the screen, as well as identify any leakage from the backlight in dark lighting conditions. Measurements of the luminance were taken at 35 points across the panel on a pure white background. The measurements for luminance were taken using BasICColor's calibration software package, combined with an X-rite i1 Display Pro colorimeter with a central point on the screen calibrated to 120 cd/m². The below uniformity diagram shows the difference, as a percentage, between the measurement recorded at each point on the screen, as compared with the central reference point.

It is worth noting that panel uniformity can vary from one screen to another, and can depend on manufacturing lines, screen transport and other local factors. This is only a guide of the uniformity of the sample screen we have for review.

Uniformity of Luminance

Uniformity of the screen was fairly good on this sample overall. The upper left and lower right corner regions were a little darker than central areas of the screen, and in the most extreme examples the luminance dropped down as low as 96 cd/m² (-25%). Overall though around 70% of the screen was within a 10% deviance of the centrally calibrated point which was good.

Backlight Leakage

Above: All black screen in a darkened room. Click for larger version

We also tested the screen with an all black image and in a darkened room. A camera was used to capture the result. The camera showed there was no major backlight bleed on this sample, only a small amount of light clouding in the four corners captured by the camera. This was slightly more evident on the left hand side in the corners. Remember that this is a photo to give an accurate representation of what you see in person from a couple of metres back, once the screen has been calibrated to a comfortable brightness level.

Note: if you want to test your own screen for backlight bleed and uniformity problems at any point you need to ensure you have suitable testing conditions. Set the monitor to a sensible day to day brightness level, preferably as close to 120 cd/m² as you can get it (our tests are once the screen is calibrated to this luminance). Don't just take a photo at the default brightness which is almost always far too high and not a realistic usage condition. You need to take the photo from about 1.5 - 2m back to avoid capturing viewing angle characteristics, especially on IPS-type panels where off-angle glow can come in to play easily. Photos should be taken in a darkened room at a shutter speed which captures what you see reliably and doesn't over-expose the image. A shutter speed of 1/8 second will probably be suitable for this.

General and Office Applications

One of the key selling points of ultra-wide screens like the this is it's high resolution and large screen size. The 3440 x 1440 display offers a sharp but comfortable picture. Its pixel area is about 1.8 times larger than an Ultra-Wide Full HD 21:9 monitor, and about 2.4 times larger than a Full HD 16:9 monitor. It provides an efficient environment in using Microsoft Office programs showing 47 columns and 63 rows in excel. Thankfully the high resolution is of a very comfortable size on the 34" panel, with a 0.2325mm pixel pitch is is very comparable to a 27" 2560 x 1440 monitor (0.2331mm). This means you are basically getting a wider desktop to work with, with a similar font size to a 27" model, and maintaining the same vertical resolution as well. If you're coming from a lower resolution / larger pixel pitch you may still find the fonts look quite small to start with, but like the 27" 1440p models out there you soon get used to it. Side by side multi-tasking on this screen is excellent and you really do have a nice wide area to work with. We liked the curved format of the display actually for day to day office work. It just felt a bit more comfortable than a flat screen on a model as wide as this, bringing the corners a bit nearer to you. You didn't really notice the curve in normal use but we liked the feel. Probably down to user taste, so if in doubt try and see one in person.

The light AG coating of the IPS panel doesn't produce any graininess to the image like some aggressive AG solutions can and so white office backgrounds look clean and clear. The wide viewing angles of the IPS panel technology provide stable images from different angles, meaning you can use the screen if you want for colour critical work, photos etc. It might be orientated at gamers, but it's IPS panel can deliver strong performance in other areas as well making it a good all-rounder. This panel technology still offers the widest viewing angles and so is well-suited to colour work. Some contrast shifts and IPS-glow may be evident because of the very wide size of the display, as you glance towards the edges from a centrally aligned position. That's hard to avoid on such a large desktop monitor from close up, even with IPS technology.

The default setup of the screen was only moderate on the G model when it comes to more general uses and any photo work. By default the screen operates with its wide gamut (around 132.7% sRGB) colour space which is fine if you're working with wide gamut content in a colour managed workflow or like the accentuated colour appearance. However there is no option on this G model to switch to any sRGB emulation mode, so if you were wanting to specifically work with sRGB standard gamut content then that might be a problem. The gamma curve was also a bit off (2.1 average) and the white point was too warm out of the box. The F model carries a better factory calibration and useful sRGB emulation mode so may be more suited if you need to use the screen for a lot of non-gaming.

The brightness range of the screen was very good, with the ability to offer a luminance between 370 and 83 cd/m². This should mean the screen is perfectly useable in a wide variety of ambient light conditions, including darkened rooms. A setting of ~12 in the OSD brightness control should return you a luminance close to 120 cd/m² out of the box. On another positive note, the brightness regulation is controlled without the need for Pulse Width Modulation (PWM), and so those who suffer from eye fatigue or headaches associated with flickering backlights need not worry.

There was no audible noise from the screen, even when conducting specific tests which can sometimes cause issues. There is a 'Reader' preset mode for office work or reading if you want which by default makes the image a fair bit warmer than our calibrated state.

The screen offers 2x USB 3.0 ports which can be useful. Both are located on the back of the screen so they aren't super-easy to access. Might have been nice to have a couple more, and maybe also offer fast charging support like a lot of other modern screens. There are no integrated speakers but there is a headphone output if you want. There are no further extras like ambient light sensors or card readers which can be useful in office environments. Remember, this is aimed at gamers really. There was a good range of ergonomic adjustments available from the stand allowing you to obtain a comfortable position for a wide variety of angles. They were mostly easy to use and it was nice to see side to side swivel included as that sometimes seems to be left off ultrawide screens like this. The VESA mounting support may also be useful to some people as well for more flexibility.

 
Responsiveness and Gaming

The 34GK950G is rated by LG as having a 5ms G2G typical response time. The screen uses overdrive / response time compensation (RTC) technology to boost pixel transitions across grey to grey changes as with nearly all modern displays. There is a user control in the OSD menu for the overdrive under the 'Response Time' setting with 4 options available as listed above. The part being used is an LG.Display LM340UW5-SSA1 Nano IPS technology panel. Have a read about response time in our specs section if you need additional information about this measurement.

We use an ETC M526 oscilloscope for these measurements along with a custom photosensor device. Have a read of our response time measurement article for a full explanation of the testing methodology and reported data.
 

Overdrive Setting

We didn't bother measuring the 'Off' mode as we don't see any reason to run with the overdrive turned off as long as there's no major overshoot with it turned on at the some of the other settings. We also didn't bother with normal mode, having realised from our review of the F model that the Fast and Faster modes returned a better performance. We started with measurements of the 'Fast' mode.

In the Fast mode the average G2G response time was measured at 7.9ms. There was no overshoot at all either which was great news and response times were fast enough to keep up with the frame rate demands of 100Hz where they need to be consistently under 10ms.

On the F model we had seen that the Faster mode delivered optimal performance. On the G model here there was some improvement in the response times, now down to 6.5ms G2G. However, some moderate levels of overshoot started to creep in, and in practice you could see some slight pale halos and trails behind moving objects. On the G model we felt that the 'Fast' mode delivered the best performance.

 

Refresh Rate and Overclocking

The screen natively supports up to 100Hz when you first connect it, but you can also enable 120Hz from within the OSD menu using the overclocking feature. This might seem a little odd because we know that the LM340UW5 panel being used can actually support 144Hz natively, and the 34GK950F (FreeSync version) model can also support that 144Hz. So why the 100 native / 120Hz overclock limit here on the G model? It's all to do with the G-sync module really.

The current v1 G-sync module can only support 100Hz natively at the 3440 x 1440 resolution of the screen. So although the underlying panel might be able to support higher, the G-sync module is holding it back a bit. LG have enabled support for 120Hz using the overclocking feature because technically, for the G-sync module, this is an overclock. From the panel point of view it should have no trouble as it's actually still below the native 144Hz it can cope with. We are limited to 120Hz maximum from the overclock here because the v1 module also only supports DisplayPort 1.2, which cannot carry the bandwidth necessary to reach 3440 x 1440 @ 144Hz. You need the more modern DisplayPort 1.4 interface for that.

There is a newer v2 G-sync module available which can support higher refresh rates natively and has DisplayPort 1.4 support, which has been used already on high end screens like the Asus ROG Swift PG27UQ and Acer Predator X27 (4K @ 144Hz). We expect that the reason why this newer module was not used was because of the cost, as we know it costs several hundred dollars more than the v1 module, and would bump up the retail price of the 950G even further. It's already an expensive screen and we suspect it would just be too high priced if they tried to use the latest v2 G-sync module as well. There may also be some technical limitations with that module at the moment as it has currently only been used for FALD HDR screens, and so it might not be available or feasible to use for a display like this currently. Regardless, to be able to offer the all-important G-sync support LG have had to make do with the v1 module, pushing it to the limit that DisplayPort 1.2 can handle and offering a 120Hz overclocking feature.

Although it is very difficult to know at this stage (only time will tell) we expect there will be fewer issues with people using the overclocking feature because we know that the underlying panel is capable of 144Hz natively. It should be more stable and hopefully have fewer issues because the panel is at least not needing to be pushed beyond it's spec. Earlier ultrawide IPS models like the Dell Alienware AW3418DW for instance are based on a 100Hz panel (LM340UW4) which is pushed to 120Hz along with the G-sync module. This 950G screen on the other hand is a 144Hz panel which doesn't need overclocking, in fact it's being under-clocked, and so it's only the G-sync module which is being pushed. It is unclear whether pushing the G-sync module with this overclock could cause any problems like flickering or artefacts over time, but we would expect them to be less likely than if you're pushing both the G-sync module and the panel at the same time. LG warranty and support should cover you as well, and only time will tell as users push the screen using the overclocking feature for extended periods of time. Certainly beyond anything we can establish with our short time with the screen for this review. We had no trouble enabling or using the feature without any artefacts or flickering being evident.

Having established the optimal mode was the 'Fast' response time setting we also measured the response times at other refresh rates.

You can see that the response time behaviour does vary a bit as you change the refresh rate of the screen. At 60Hz in this 'Fast' mode we measured an average 9.5ms G2G response time. This improved quite nicely down to 7.9ms at 100Hz, and a little bit better at 120Hz where it reached 7.7ms G2G. At all refresh rates there was no visible overshoot which was great news in this response time setting. If you are using G-sync to variably control the refresh rate then the response times will be controlled dynamically by the G-sync module, but in every case they are fast enough to keep up with the frame rate demands thankfully.

Refresh Rate Impact on Motion Clarity, G-sync and Blur Reduction Mode

As you increase the refresh rate of the screen there is an additional benefit related to motion clarity due to the way the human eye perceives blur from LCD displays. Not only are the pixel response times improving to help keep up with the frame rate, but there is a direct relationship between refresh rate and perceived motion blur levels. There is certainly a lot less blurring at the higher refresh rates, and it is easier to track moving objects across the screen. Although there might not be a major difference in response times between 100Hz and 120Hz (7.9ms vs 7.7ms average) there is still an improvement in motion clarity thanks to the extra 20Hz of refresh rate. We would recommend running the screen at the overclocked refresh rate if you are able to push the screen to these levels from your graphics card and system. That might vary by game, and over time it will become easier to run 3440 x 1440 @ 120Hz as graphics cards improve too.

One important feature of this screen is the support for NVIDIA G-sync which offers support for variable refresh rates, helping to avoid tearing and stuttering in games without introducing the lag associated with older Vsync options. The supported range is 30 - 120Hz assuming you've enabled the overclock to push it to the maximum supported refresh rate. It's a very useful technology for when your frame rates fluctuate, especially considering it will take a powerful system to run the screen at its native 3440 x 1440 resolution @ 120Hz. This is only supported from compatible NVIDIA graphics cards, with AMD/other users not being able to make use of the variable refresh rate sadly.

Unfortunately unlike the F model there is no added blur reduction mode (MBR feature) on the G model. This was added separately to the F model to allow support for a strobed backlight system, which can help improve motion clarity in practice. We have written a detailed article about such blur reduction backlights so we would encourage you to read that if you are unfamiliar with how these operate and the benefits they can produce. Some people aren't bothered by these strobed backlights and would rather game with G-sync and a flicker free experience anyway, so for some people it won't be missed. Others like to use them and may be a bit disappointed that it has been left off here.

 

Detailed Response Time Measurements
Response Time mode = Fast
Refresh Rate = 120Hz

 

Having determined that the 'Fast' mode delivered the optimal response times, and that the best experience was at 120Hz, we measured a wider range of grey to grey transitions to give a more complete picture of the performance. Overall the average G2G response time was more accurately measured at 7.6ms. Some were a bit faster, reaching down to 5.5ms and being just about in line with the manufacturers spec of 5ms G2G. A few were a bit slower up to around 13.5ms, where the shade was changing from black (0) to light grey/white (200/255). Overall the response times were fast enough to keep up with the frame rate demands of 120Hz which require a new frame to be sent to the screen every 8.33ms. The other great thing was that there was no overshoot at all in this 'Fast' mode which was excellent.

The 'Faster' mode can push the response times a little more and you may still want to experiment with that. We saw from our measurements that it started to introduce a moderate level of overshoot though, and you could see some slight pale halos and trails on moving content in practice. We preferred the 'Fast' mode but feel free to experiment.

 

Gaming Comparisons

We have provided a comparison of the 34GK950F against many other screens we have reviewed. The overall responsiveness was comparable to most of the other Ultrawide IPS screens we have tested. With an average response time of 7.6ms G2G measured, it was close to the Acer Predator X34 (7.9ms, an overclocked 100Hz G-sync screen) and Asus ROG Swift PG348Q (7.8ms, overclocked 100Hz G-sync) for instance. These response times were achieved without any overshoot becoming evident, even at the 'Fast' response time setting on the LG.

Those Acer and Asus models are based on an older LM340UW2 panel which offered a 60Hz refresh rate, and was overclocked to 100Hz thanks to the G-sync module. We had seen some improvements more recently from the Dell Alienware AW3418DW which featured a newer LM340UW4 panel with a native 100Hz refresh rate, and this time overclocked to 120Hz thanks to the G-sync module. That panel had produced some faster response times at 6.9ms G2G average. Here on the LG 34GK950G we have the latest generation of LM340UW5 panel with a native 144Hz refresh rate, which is limited down to 100Hz native / 120Hz overclocked because of the G-sync module. It isn't quite as fast as the Dell when it comes to pixel response times which had reached 6.9ms G2G but it's not far behind, and not really noticeable.

If we compare the G model with the F model you can see that the G has slightly better response times, with 7.6ms vs 8.2ms G2G average. There is also a small amount of overshoot on the F model, while there was none here on the G model. So overall the G model has the slight edge when it comes to response times and pixel transitions.

Additional Gaming Features

• Aspect Ratio Control - the screen has 3 options for hardware level aspect ratio control options which were full wide, original and 1:1. It is useful to see an 'original' mode available to maintain the source aspect ratio if it is different to the screens 21:9 format. Especially considering a lot of sources operate at the more common 16:9. There is also a 1:1 pixel mapping mode if you want to directly match the resolution to the screen without any scaling at all.

• Preset Modes - There are quite a lot of preset modes available in the 'Game Mode' menu. This includes presets for FPS and RTS games, as well as two customisable 'Gamer 1' and 'Gamer 2' modes. You should be able to set up different modes for different gaming uses.

• Black Stabilizer - there is an OSD menu setting which allows you to alter the gamma detail in dark content to make it easier in darker games

• Crosshair - an on-screen crosshair can be enabled via the OSD menu as well which can help accuracy in some games

 

Lag

We have written an in depth article about input lag and the various measurement techniques which are used to evaluate this aspect of a display. It's important to first of all understand the different methods available and also what this lag means to you as an end-user.

Input Lag vs. Display Lag vs. Signal Processing

To avoid confusion with different terminology we will refer to this section of our reviews as just "lag" from now on, as there are a few different aspects to consider, and different interpretations of the term "input lag". We will consider the following points here as much as possible. The overall "display lag" is the first, that being the delay between the image being shown on the TFT display and that being shown on a CRT. This is what many people will know as input lag and originally was the measure made to explain why the image is a little behind when using a CRT. The older stopwatch based methods were the common way to measure this in the past, but through advanced studies have been shown to be quite inaccurate. As a result, more advanced tools like SMTT provide a method to measure that delay between a TFT and CRT while removing the inaccuracies of older stopwatch methods.

In reality that lag / delay is caused by a combination of two things - the signal processing delay caused by the TFT electronics / scaler, and the response time of the pixels themselves. Most "input lag" measurements over the years have always been based on the overall display lag (signal processing + response time) and indeed the SMTT tool is based on this visual difference between a CRT and TFT and so measures the overall display lag. In practice the signal processing is the element which gives the feel of lag to the user, and the response time of course can impact blurring, and overall image quality in moving scenes. As people become more aware of lag as a possible issue, we are of course keen to try and understand the split between the two as much as possible to give a complete picture.

The signal processing element within that is quite hard to identify without extremely high end equipment and very complicated methods. In fact the studies by Thomas Thiemann which really kicked this whole thing off were based on equipment worth >100,1000 Euro, requiring extremely high bandwidths and very complicated methods to trigger the correct behaviour and accurately measure the signal processing on its own. Other techniques which are being used since are not conducted by Thomas (he is a freelance writer) or based on this equipment or technique, and may also be subject to other errors or inaccuracies based on our conversations with him since. It's very hard as a result to produce a technique which will measure just the signal processing on its own unfortunately. Many measurement techniques are also not explained and so it is important to try and get a picture from various sources if possible to make an informed judgement about a display overall.

For our tests we will continue to use the SMTT tool to measure the overall "display lag". From there we can use our oscilloscope system to measure the response time across a wide range of grey to grey (G2G) transitions as recorded in our response time tests. Since SMTT will not include the full response time within its measurements, after speaking with Thomas further about the situation we will subtract half of the average G2G response time from the total display lag. This should allow us to give a good estimation of how much of the overall lag is attributable to the signal processing element on its own.

Lag Classification

To help in this section we will also introduce a broader classification system for these results to help categorise each screen as one of the following levels:

• Class 1) Less than 8.33ms / 1 frame lag at 120Hz - should be fine for gamers, even at high levels

• Class 2) A lag of 8.33 - 16.67ms / One to two frames at 120Hz - moderate lag but should be fine for many gamers. Caution advised for serious gaming

• Class 3) A lag of more than 16.67ms / more than 2 frames at 120Hz - Some noticeable lag in daily usage, not suitable for high end gaming

For the full reviews of the models compared here and the dates they were written (and when screens were approximately released to the market), please see our full reviews index.

We have provided a comparison above against other models we have tested to give an indication between screens. The screens tested are split into two measurements which are based on our overall display lag tests (using SMTT) and half the average G2G response time, as measured by the oscilloscope. The response time is split from the overall display lag and shown on the graph as the green bar. From there, the signal processing (red bar) can be provided as a good estimation.

We measured a total display lag of only 3.3ms. We estimate that pretty much all of that will be a result of the pixel response times and so can infer that there is basically zero processing lag at all. This isn't an exact science but provides a good indication of what the signal processing lag is. We have seen similar behaviour from some other G-sync screens such as the Dell Alienware AW3418DW and Asus ROG Swift PG279Q for instance which had basically no signal processing lag and would be directly comparable in this area.

Note that at the time we originally tested both the F and G screens, the G model was quite a lot faster than the F model which had shown an estimated 9.70ms of signal processing lag. While the G-sync module might be holding the screen back in some places, it does at least allow for a super-low lag. LG later updated the firmware for the F model which massively improved the lag, down to an estimated 0.2ms of signal processing lag and removing that difference between the G and F models.

Movies and Video

The following summarises the screens performance for videos and movie viewing:

34GK950G vs 950F Comparison Updated

 

When we originally reviewed these screens in October 2018 we looked at the F model first, and then at the G model. We included this comparison section in this G model review to compare the two screens and try to help readers decide which model might be right for them. Later on we updated the F review with tests from the new firmware in February 2019. As a result, some of our views have changed, so we have now updated this section as well.

There are obviously quite a lot of similarities between the F and the G model, with both having basically the same fundamental design (except the Sphere lighting on the G model) and both being based on the same underlying IPS panel. We have provided a table above summarising some of the key differences in performance and features between the two screens though to help you decide.

Originally when we compared the two models there was a fairly clear target market for each, and a strong reason to buy one of the other. That choice really boiled down to your graphics card choice, with the F model being sensible (and clearly the right option) for AMD FreeSync users, and the G model being preferable for NVIDIA users in our opinion. These variable refresh rate technologies provide some significant benefits in gaming and should not be overlooked. Some NVIDIA users still decided to buy the F model for its other benefits like a higher refresh rate and sRGB emulation mode, and live without the support for VRR in gaming. While also saving themselves a bit of money in the process as the F model was a bit cheaper despite offering more features. With the arrival of the new NVIDIA drivers in January 2019, and the support for VRR on FreeSync screens from compatible NVIDIA graphics cards, it is a bit of a game changer! No longer are you restricted to buying the G model if you want VRR support, as long as you have a high end NVIDIA card you can get the VRR experience from the F model now too. There may be some people who don't have a high end GTX 10-series or RTX 20-series card, and they would still benefit from the G model we feel, especially as with a less powerful card they are likely to experience lower and more varied frame rates - exactly where G-sync is most useful. However, if you've got a high end card, we really don't think there's any major reason to buy the G model over the F model now.

The F model has support for a higher refresh rate, being able to handle the full 144Hz of the panel and giving a 24Hz boost over the G model. However, don't forget you're going to need a very modern DisplayPort 1.4 graphics card to reach that and so that is an important consideration. The F model also has the added blur reduction (MBR) mode which some people like for their gaming. Having said that though, the G model does have the edge when it comes to response times, but only by a little. The pixel response times are a little faster and also 100% free from any overshoot. Originally the lag of the screens was also a fairly important differentiator for most buyers, with the G version having basically no lag at all thanks to the added G-sync module, and the F model having around 9.7ms of signal processing lag. With the arrival of the v2 firmware in early 2019 for the F model, the lag has been eliminated basically completely and so this is no longer a difference between the two models.

Away from gaming the F model has the advantage. It has a factory calibration that the G model does not, and so provides a more reliable default setup. There is also the useful sRGB emulation mode which might be useful for colour critical work or viewing content in the more common sRGB colour space. On the G model you are restricted to always using the wider gamut of the backlight. Probably fine for gaming and multimedia, but it doesn't give you the flexibility to run in sRGB if you wanted to. There is also support for 10-bit colour depth if you have an appropriate graphics card and workflow, or want to use it for the latest games. Finally there is a better static contrast ratio with 1000:1 measured, compared with 902:1 on the G model. That may vary from unit to unit, but those were our findings.

The G model does have the added Sphere Lighting system which is pretty attractive, and may be something some people like. It does have the older DP 1.2 connection because of the use of the ver 1 G-sync module and that's the reason it cannot reach higher than 120Hz here. The HDMI version is also limited to 1.4 because of the G-sync module and so cannot support beyond 50Hz at the full resolution unlike the F model.

The G model is still the right choice in our opinion if you've got a lower end NVIDIA card (lower than GTX 10-series and RTX 20-series) and won't be able to support either DisplayPort 1.4, or VRR on the F model. The G-sync support is going to be particularly useful for gaming on such a high resolution and refresh rate if you've got a lower range NVIDIA card. Given the F model can now support VRR from top end NVIDIA cards, and the lag has been eliminated, it is the obvious choice over the G model for those with either a high end NVIDIA card, or an AMD card of course.

 

Conclusion

Love our work? If you would like to help support our continued work please consider making a donation no matter how small or large. Thank you.

It's been very interesting to be able to test and compare both versions of the 34GK950. Ultimately the FreeSync vs G-sync option is what is going to drive a purchasing decision between the two models here. The G screen in this review performed very well when it came to the primary use of gaming, offering fast response times, zero overshoot, high 120Hz refresh rate, and zero lag. While the G-sync module holds the screen back a bit in some areas as we discuss below, it has at least allowed for a strong gaming performance and zero lag which is great news.

We were a little disappointed in some regards because of the sacrifices that had to be made to provide G-sync support on this screen. It's a great technology for gaming, and overall we still feel that the benefits it brings out-weigh the sacrifices that have had to be made in a few areas. It does feel a bit like the panel and technology is a step ahead of the G-sync ver 1 module that LG have had to use here, and is not being used quite to its full potential. We can understand the likely reasons why they've had to use this ver 1 module given the increased cost and possible technical considerations of the very recent ver 2 module, but when you have the 34GK950F model offering more features and making full use of the new LM340UW5 panel it is a little bit of a shame. The 24Hz worth of extra refresh rate isn't a major deal but felt a little bit of a shame here, but on balance it does make the screen more accessible to more users because you only need DP 1.2 and not the latest generation of DP 1.4 capable graphics cards to run it at its full capability. Being limited to HDMI 1.4 was a little bit of a shame as you can't push the higher refresh rates from your PC using that connection, but there is DP instead which should be fine anyway. The MBR mode was a nice addition to the F model and we missed it a bit here on the G model, presumably because of the refresh rate restrictions or again to help keep costs down. Perhaps the thing we most missed was the sRGB emulation mode really, as that can be useful when you don't want to run at the full native gamut. Having seen it on the F model, we were sad not to see it included here on the G model. Like we've said though, the variable refresh rate support for NVIDIA cards that the G-sync module brings, and the incredibly low lag probably outweigh these limitations.

We would have liked to have seen a bit more focus on a factory calibration here from LG, like they'd done with the F model. Perhaps again that was left off to avoid adding to the already higher retail cost of the G model compared with the F model. Contrast ratio was a little lower too than the F sample we'd reviewed, but overall the screen still offers the solid all-round performance, reliable picture quality and wide viewing angles you'd expect from an IPS panel. It's the first and only currently available ultrawide IPS screen with G-sync and wide gamut, so if you want boost colours for multimedia and gaming then it's a great feature. Overall the G model felt like it was more firmly a gaming screen than the F model which was perhaps a little behind for really fast gaming, but a bit more rounded for other uses. There's a fair few ultrawide G-sync monitors out there to choose from which are a bit older and run up to 100Hz maximum, but we felt this was an interesting alternative to the Dell Alienware AW3418DW (one of the few other 120Hz capable IPS ultrawide screens) and with the wide gamut support and some different features it is certainly worth considering.