Input Latency: HD CRT vs HD LCD

Background

A while back my trusty CRT, a Sony KV-32XBR450, started having some weird sync issues where the image would sometimes be too low, then it would start from the middle, then jump around a bunch… it wasn’t great. At first it only happened for a few minutes when it was first turned on, but as time went on, it took longer and longer for the TV to “warm up”. Eventually, it was like this all the time, so it was time to start looking for something new.

While browsing Craigslist for a 1080p flat panel I stumbled across something interesting: an NEC LC5220AV. Knowing that NEC makes some dank shit, I did some research to see if this would be a good display. I was really excited about the inputs: native RGB H/V, VGA, DVI, HDMI, component, composite, s-video; it seemed to go on forever. $200 later it was in my living room and it. was. fucking. awesome. The picture-in-picture options on this 52-inch panel meant that my roommates and I could play multiple consoles on the same TV at the same time (8-player Mario Kart, anyone?). I always wondered what the input latency on this panel was, especially compared to my CRT I had given up for this much larger, more modern display.

Thanks to the 240p Test Suite, I’m now able to answer that question.

Testing

 

The first test I performed involved hooking my Dreamcast up to each display and running the Manual Lag Test five times, using the median as my final result. I did this with the CRT, then the LCD using no scaling (so it was rather small on the screen), scaled to fit the display, then again using the Framemeister XRGB-Mini to upscale the 240p image and output it to the LCD over HDMI.

For the second test, I connected the Dreamcast to the CRT, then used the CRT’s “monitor out” to send the video to the LCD, so the image is displayed on both screens. Using the Lag Test and my DSLR’s video recording feature, set to 60 frames per second, I recorded the difference between the two screens. Afterward, I connected the Framemeister between the CRT and LCD to see if that added any additional latency.

Results

Out of the five input latency tests, I observed the follow results:
CRT: 1.6 frames (26.67 ms)
LCD (native): 2.1 frames (35 ms)
LCD (scaled): 2.1 frames (35 ms)
LCD w/ Framemeister: 1.4 frames (23.33 ms)

I was happily surprised that the LCD’s internal scaler only added about half a frame of delay (compared to the 2-3 I often hear being added on consumer LCD panels), but I was astonished that the results with the Framemeister were better than the CRT’s results. I should point out that the KV-32XBR450 is a high-definition CRT with image processing that cannot be disabled, meaning that it, like modern displays, is going to add some amount of latency compared to a more traditional consumer CRT or PVM/BVM.

The results from the second test were also surprising. Since the signal being output from the CRT’s “monitor out” shouldn’t be subject to any image processing, this would present a good way to run a “drag race” between the two panels to directly compare display latency.

The image on the left is with the composite signal going directly into the LCD panel, which adds one frame of input latency. the image on the right us using the Framemeister to upscale the image to the LCD’s native 1080p, which results in no additional input latency. This isn’t to say that the Framemeister doesn’t add any input latency at all. What it means is that the internal scaler and image processing of the Sony CRT add the same amount of latency as the XRGB-Mini. If my manual input latency testing is accurate, they both add o frame and a half of latency; something only the most elite fighting-game players would notice.

Closing

The purpose of this testing wasn’t to determine which display is better, but to test specific differences between the two. Using the 240p Test Suite’s various tools, I was able to see frame stutters in the LCD panel that aren’t present on the CRT. The scaling of 240p content on both displays was between mediocre and above average. Geometry was accurately represented on the LCD, but rather poorly on the CRT due to lack of calibration, age, whatever lingering issues that had caused the sync issues mentioned earlier. All that said, it would appear that I’ve lost nothing by upgrading from this particular CRT to this particular LCD. Which is pretty cool.