Finally, here are some real numbers for my LG 37LH20. I think I went overboard on the testing, but hopefully it will be a useful example of how it can be done and what things to consider.Short Version:The 37LH20's lag is 1.5 frames (25ms.)Longer Version:
Total display lag, from incoming video to fully fired pixel, is 30ms. This is the sum of 21ms input lag and 9ms pixel response time. The "practical" total lag is somewhat less, about 25ms (1.5 frames), since pixel changes are visible before they have fully settled.
Upscaling of 480i content adds one additional frame of delay, for a total of 2.5 frames.Tests Performed:
Interfaces and Resolutions:
HDMI - 1080p, 768p, 720p
VGA - 1360x768, 1280x720, 1024x768, 640x480
Component - 720p, 480i
Composite - 480i
Input Labels (blank, PC, Game, etc.)
Picture modes (Cinema, Game)
Edge Enhancement on/offTest Setup:
LGD S-IPS panel (printed on sticker visible through center rear vent)
Manufacture Date: October 2009
Firmware Revision: unknown
Core 2 Duo 3 GHz
nVidia GeForce 8600GT with dual DVI outputs and component/composite TV output
Windows 7 64-bit
Latest DirectX, Windows Updates, and nVidia drivers as of 3/1/2010
SMTT application, 64-bit version
Hitachi CM753 19" SVGA CRT
Commodore 1084 12" NTSC CRT
Nikon Coolpix 4100
(EXIF data claims 1/90th sec shutter, ISO 200)Test Results:
Results were remarkably consistent across input types, HD resolutions, and TV settings. Most measurements only varied +-1ms, with just a few outliers that were not far beyond this. As a result, I'll only post a representative sample here. If anyone really wants to see dozens of very similar pics, or a table of values, I can certainly post them somewhere.
In this pic, the highest number on the CRT is 05.640. The highest number on the TV is 05.619. It is just starting to appear on the second row. This shows an input lag of 21ms. The highest number on the TV that is fully visible is 05.610, which indicates that took an additional 9ms for the pixels to fully respond. I don't know for sure if the camera exposure settings affect this reading, so it could be lower.
Standard-def (480i) results were similarly consistent, at 48ms (2.9 frames.)
The CRT is currently at 19.364 on row 2, and the TV is at about 19.322 on row 5 (42ms input lag.) The latest fully visible time is 19.317 (5ms pixel lag.) There are fewer rows at 480i, so the individual lag and pixel times aren't quite as accurate as for the HD tests, but the combined times are consistently around 48ms.Other Tests:
I consider the SMTT results to be definitive, but I wanted to compare the accuracy to other common tests.
Using the flatpanels.dk timer, I tested three modes with 10 shots each. Individual values varied a lot (9-40ms), but the average was the same for each mode: 23ms, or 1.4 frames. This is fairly close to my "practical" figure of 1.5. Maybe this quick timer test isn't so bad after all.
I also tried a timer I found in this post http://www.avsforum.com/avs-vb/showt...4#post16445094
It averaged about 1.5 frames as well. I think this timer might be better than flatpanels.dk, but I haven't worked with it enough to know for sure. It's a little harder to interpret, but the values seem more consistent.Not Tested:
There are a few picture settings I did not test: Noise Reduction, Dynamic Contrast, Dynamic Color, and "OPC." They didn't seem relevant for game/PC use. I suspect they would not cause additional lag, anyway.
I did not test the HDMI1 input. HDMI2 was used for all tests.
I couldn't get my video card to do 1080p or 768p on the component output in combination with the CRT. It's possible that the TV doesn't support them, anyway.Additional Notes:
Picture modes can affect available options in the advanced picture menu. For example, Game forces edge enhancement to high. Cinema and Expert appear to have the most available options.
The only input label that has an effect is "PC". It disables most of the picture settings, and disables overscan for 720p sources. It does not affect input lag or general picture quality.
Even downscaling 1080p to 768p does not cause additional lag. It looks surprisingly good, too. I wouldn't want to use it with a PC full-time, but it would probably be better for Blu-Ray than using 720p upscaled to 768p.
PC display looks pretty nice at 1360x768. It may not be an ideal primary monitor, but it is fine for general HTPC use.
The VGA port does not work well. It didn't properly detect a 1360x768 signal. It reported it as 1024x768, and the image was shifted with a blank border on the right side. It wouldn't display 1280x720, but it did show 1024x768 and 640x480. (I did not test 800x600.) Image quality was fairly poor, except at 640x480. Since the VGA port doesn't affect lag, I see no good reason for using it.
Well, folks, that's it. My earlier suspicions were that the LH20 is a little slower than the Panasonic X1/S1 LCDs, but faster than most others. These tests confirm it.
I’ve had Rock Band set for 25ms lag for a while, and it feels just about right, so I'll be leaving it there. This is mostly what I cared about, but 3D games sometimes give me motion sickness if the controls aren't responsive enough, so it's certainly reassuring to know the lag is pretty low.
If you made it this far, thanks for reading!