Posts

Showing posts from October, 2020

LG 47LW6500-UA review: input lag and upscaling tested using the piLagTesterPRO

This 1080p LCD from 2012 has a very nice picture and supports 3D using passive polarized glasses. And it's even a smart TV. But it has a dirty secret: the worst input lag I've ever seen. 

Overview/Image quality

At native resolution this TV looks great. Pixel perfect (no clipping/aliasing), as you would expect. It also does well upscaling lower resolutions: 480p/i is shown with zero aliasing or scaling artifacts, at the cost of a minor amount of cropping around the edges; roughly 15 pixels are hidden on each edge. This is very good. 720p is also very good, with zero cropping.  It does introduce a slight wiggle to straight lines but it's not very noticeable, and there's no blur or other aliasing. Check out the photo at right for an example. 

It can sync to both 60hz and NTSC (59.94hz) inputs, but despite loudly claiming in their marketing material that it is a 240hz display, you cannot go a single Hz above 60. Not even 70hz pc display modes work. Shame on you LG marketing department. I think by 240Hz they mean they insert a lot of black frames, but who knows.

It has all the regular inputs: HDMI, Component, etc. I tested over HDMI.

Input Lag

This display has a game mode, which I used. It did improve input lag.  I toggled all the other display quality settings and did not see a consistent effect on lag.  I used a piLagTesterPRO (click for full details) to measure input lag. Starting from a black background it sends a white frame of video over HDMI and measures how long it takes to show it relative to when that frame started (thus, there is more lag at the bottom of the screen than at the top). 

Test Results

I report two kinds of values. 1st response measures how long it takes for the TV to start responding (I use a 5% change in display brightness). This overly optimistic value doesn't tell how long it takes to see anything useful, but matches what other reviewers call input lagfull response is a more realistic measure of lag, and requires the display to reach 80% of it's final brightness. This combines both input lag and response time, and is closer to what you would actually experience in a game.

Full results for this display

display topdisplay bottom
Resolution1st responsefull response1st responsefull responsescan outresponse time
480i136.7139.4147.0149.710.32.7
480p135.0137.7147.0149.012.02.7
720p67.369.379.181.711.82.0
1080p66.668.378.380.911.71.7

This TV is shockingly slow, and not at all what I expected for a "240Hz" display.  I had to triple check that it had game mode on. 480i/p are particularly bad, at around 135ms. Somewhat surprisingly, the lag is the same whether it's interlaced or progressive, suggesting that the TV might treat them identically (i.e. try to deinterlace 480p). Or maybe with that much lag there's plenty of time to also deinterlace. I couldn't see any evidence of deinterlacing artifacts in either mode. 

Things get better with HD content, dropping to 66ms of lag, but that's still worse than any other display I've tested in game mode. The only bright points are the scanout time, which is significantly shorter than the refresh rate of the display, meaning that once it finally starts drawing it does so faster than it receives the signal, and it's super fast response time of around 2ms. In fact I think this super fast response time is probably what ruins the TV: in order to drive the pixels to their final appearance that fast the TV needs to know what the next few frames are going to be, and that's why the lag is so bad. 

If this were an active shutter goggle 3D display, that kind of response time would be necessary in order to switch what's being shown to each eye every 16ms. But since it uses passive polarization there's no need:  every other line has an opposite polarization, allowing glasses which are permanently polarized to split the screen between eyes with zero flicker. So there's no need for such high switching speed. Perhaps LG's engineers started making a panel for active shutter goggles and switched to passive polarization later on? FYI the 3D effect this gives is very good, not that there's any new 3D content to watch anymore. 

From the marketing blurb: 
  • See sports, video games and high-speed action with virtually no motion blur and in crystal clarity with LG's TruMotion 240Hz technology

Just to be clear, TruMotion was off, as that could have explained the huge delay too. 

Just for laughs I also tried measuring the display with game mode off. The 1st response at the top of the screen was 177ms for 480i, meaning game mode is 40ms faster! On most sets that would be a significant improvement. 1080p was 75ms outside of game mode, which was a much smaller 9ms change. So game mode does something but only really for low resolution (or interlaced?) content. Obviously not enough.

Next let's compare it to other display's I've tested, just to see how bad it is.

Results Compared to other displays

Legend: Min lag is the 1st response at the top of the screen. Real lag is the full response at the bottom. 
DisplayYear madeNative Resnative min lag480i real lag480p real lag720p real lag1080p real lagresponse timescan out
Dell U2410 (game)20101080p4.062.228.326.526.56.0016.20
Dell U2410 (sRGB)20101080p20.562.445.043.143.16.1316.13
Sony 40VL130 (game)20081080p22.866.349.047.347.39.0815.43
Sony 40S20L12007720p48.490.172.973.49.6014.00
Emprex HD 32022007720p27.0126.051.050.08.5015.00
LG 42LC2D2006720p28.354.650.850.46.3015.95
Vizio VO370M20101080p2.583.049.047.024.35.4715.67
Philips 42PFL3603D/F720091080p29.084.050.050.050.05.0016.00
Samsung S27C23020141080p2.936.636.137.018.2514.97
LG 47LW6500-UA20121080p66.6149.714981.780.92.2711.83

I've never seen a slower display at native resolution. I've never seen a slower display for 480i. Hold on, I've never seen a slower display, period. Yikes.

It does have the fastest response time and scan out I've ever seen, but what's that worth when the base input lag is so slow? Nothing at all. 

Conclusion

In any case, worst gaming TV ever. By a huge margin.  Nice 3D TV if you can find some content for it. And it looks perfectly nice for non-interactive content. It does a great job of upscaling 480i/p, if you have some old movies to watch.

Other models (to avoid)

I tested the LG 47LW6500-UA model, which is the 47" size. There were two other sizes sold with exactly the same specs: 55LW6500 and 65LW6500. I would expect them to be just as awful as this one in terms of input lag. 

automatic port forwarding for the raspberry pi NAS

My pi-based NAS is working great, but I'd like it to be more robust to being moved between networks/routers. In particular I need an external port opened up for SSH. There's a nice tool for this called miniupnp which is in the raspberry pi repositories (you will have to apt-get install it).

This page describes how to automate port forwarding with upnp nicely, but since links are fragile, here's the key script:

#!/bin/bash
export LC_ALL=C

upnpc="upnpc " # extra options here
external=3333
port=22
ip=$($upnpc -l | grep "Local LAN ip address" | cut -d: -f2)

$upnpc -d $external TCP >/dev/null 2>&1
$upnpc -a $ip $port $external TCP >/dev/null 2>&1
you put this script where it will be executed any time you get a new IP from your router:
/etc/dhcp/dhclient-exit-hooks.d/
and just to be safe call it every time you reboot as well (you don't get a new IP from your router each time you reboot, so this script won't be called otherwise).
crontab -e
@reboot sleep 10 /etc/dhcp/dhclient-exit-hooks.d/port_forward
FYI not all routers support UPNP for instance ATT U-verse does not.

Samsung S27C230 review: input lag and upscaling tested using the piLagTesterPRO

 This 1080P  TN LCD from 2014 has fantastic input lag but only OK response time and does not support 480i.

480P scaling artifact
Overview/Image quality

At native resolution this LCD looks great. Pixel perfect, as you would expect for a computer monitor, with no cropping or aliasing.  The same goes for 720p. It can't do 480i at all, and 480p has bad upscaling artifacts. It does accept 1080i, which it displays using Bob deinterlacing.

Using an OSSC it's possible to feed it 480i video at 960P (aka line 4x mode) which looks pretty good, with a little aliasing, and 60 pixels cropped off the top (only).

It can sync to both 60hz and NTSC (59.94hz) inputs, and even up to 75hz at reduced resolution (1280x800). vsync is perfect with no temporal drift unlike some displays I've tested. The display update exactly matches the input frequency. 

Because it's a computer monitor the inputs are very limited: DVI and VGA. My computer had no problem driving it using my HDMI port and a HDMI 2 DVI adapter, however, it was a tight fit to get the adapter to fit in the back. The VGA input is also well handled, with no blur or ringing. 

The display stand is not adjustable and there are no VESA mounts.  There are some holes on the back but they do not line up with any standard I've ever seen. Maybe vents?

Input Lag

This display offers a "magic contrast" setting which seems to be their name for scenes (aka color/gamma defaults) and one of the options offered is game mode. It performs the same as all the other modes, save the dynamic contrast mode which looks ugly and slows the total response time up to 60ms. I toggled all the other display quality settings and did not see a consistent effect on lag.  I used a piLagTesterPRO to measure input lag. This device sends a frame of video over HDMI and measures how long it takes to display it.

I report two kinds of values. The minimum lag is the first point in time any change is detected at the top of the screen. This overly optimistic value doesn't tell you how long it takes to see anything useful, but matches what other reviewers use. I also report a more realistic measure of lag: when the display has reached 80% of full brightness at the bottom of the screen. This combines both input lag and response time, and is closer to what you would actually experience in a game.

This display has an interesting property I've never seen before: if the video mode is for a TV (CEA) the
response time is slower than if it's for a computer monitor (DMT). The lag (aka when the response starts) is the same, but how long it takes to finish is hugely different! Here's an example for 1080p; the left side is DMT mode 82, the right is the same resolution but using a CEA mode. Apologies for just using a screen capture; this means that the plots are on their side with time increasing as you go down, with gray lines every 10ms.


The monitor responds 18ms after input is sent to it in either case, but on the left it quickly reaches full response in about 5ms. But on the right it starts to slow down mid-transition, waiting 10ms before resuming full speed. Very strange. Now you might notice the scales are not identical, which is because in the CEA mode the LCD uses a limited brightness range of 15-240. I've recently added an option to the piLagTester to specify brightness for the background and target rectangles so I could emulate this reduced brightness range while in a DMT mode. Conclusion: it's the limited brightness range, not the mode. If I use near-black background of 15, and a target rectangle of 240, I get identical results to the CEA mode. 

This smells of specsmanship: they've carefully optimized the display to produce a fast brightness change when going from black to white, the transition most likely to be tested by reviewers. But they haven't similarly optimized smaller transitions, so that paradoxically a smaller change actually takes longer than a full black to white transition. Yuk. And of course guess what RT the official specs report? 5ms. 

This highlights why it's so important to measure both input lag and response time. The numbers reported above were measured at the bottom of the screen, meaning that the display is really twice as slow as the input lag would suggest. And keep in mind the display's input lag at the top is just 4ms; you'd never guess that it takes almost 10x that to completely draw the image.

Test results:

480i is not supported; here I report values for the CEA modes since that gives a more realistic measure of response time.

DisplayYear madeNative Resnative min lag480i real lag480p real lag720p real lag1080p real lagResp. Timescan out
Dell U2410 (game)20101080p4.062.228.326.526.56.0016.20
Dell U2410 (sRGB)20101080p20.562.445.043.143.16.1316.13
Sony 40VL130 (game)20081080p22.866.349.047.347.39.0815.43
Sony 40S20L12007720p48.490.172.973.49.6014.00
Emprex HD 32022007720p27.0126.051.050.08.5015.00
LG 42LC2D2006720p28.354.650.850.46.3015.95
Vizio VO370M20101080p2.583.049.047.024.35.4715.67
Philips 42PFL3603D/F720091080p29.084.050.050.050.05.0016.00
Samsung S27C23020141080p3.036.636.13718.2514.73

And the raw results:

480p top 1st480p top full480p bot 1st480p bot full720p top 1st720p top full720p bot 1st720p bot full1080p top 1st1080p top full1080p bot 1st1080p bot full
4.223.118.636.6320.918.336.12.920.918.137

Basically the values are all the same: 4ms to start responding at the top of the screen and 36ms to finish at the bottom. I measured the same values for 1280x960 as well.

The monitor supports 75hz at lower resolutions and I thought it would be interesting to compare to 60hz. Note these are DMT modes with black=0 and white=255, giving overly optimistic response times.

mode top 1st (lag) top full bottom 1st (lag) bottom full
1280x800x60 3.1ms 8.1ms 17.7ms 22.7ms
1280x8000x75 2.5ms 7.5ms 15ms 19ms

The result, unsurprisingly, is that everything is faster in 75hz, with the total lag reduced by about 3ms per frame. This matches what you would calculate in abstract for a 60 vs 75hz signal. Another reality check passed by the piLagTesterPRO.

Conclusion

This is a fine computer monitor for everyday use. It's not bad for gaming, either. While its lag is really good, its response time is very slow, the slowest I've ever seen. But the combination is still competitive, better than most TVs I've tested. Of course, computer monitors tend to have lower lag, and this is a TN display, which also tend to be faster. Off-angle viewing is not nearly as good on this display as an IPS monitor. But it's not so bad as to be distracting; if you sit in the sweet spot you won't even know it. 

Personally, I think that 27" displays are too large for 1080p resolution; there's no advantage to having a screen this size unless your desk forces you to it fairly far way.

Other models

I tested the S27C230B model. The S24C230BL appears to be identical in terms of specs and features but smaller, and so it's likely to preform the same. 

Meanwhile, although similarly numbered,  models S24C230JL, S24C230JY and S27C230J appear to be different based on the addition of a "magic upscale" configuration option and the removal of the VGA input. While likely to perform similarly, they may not be identical to the  S27C230B. 

Email me

Name

Email *

Message *