OLED Vs CRT input lag when gaming

[NeonPizza]

I've always assumed that CRT SDTV's had zero input lag, but now I'm hearing that they actually have 8.3ms. Where as OLED has anywhere from 10ms(60fps) to 5ms(120fps). Is this true, or are SD CRT's actually latency free/flawless?


One of my favourite CRT's I used to game on in the mid to late 2000's with Dreamcast, Gamecube & Wii. Been using them since the late 80's.

if the 8.3ms lag is true, that would mean OLED is less than 2ms slower for 60fps titles, and 120fps's 5ms is faster than CRT by around 3ms. Once you enable OLED BFI, like on the LG C1 ala' motion pro high @60fps, latency shoots up to at least 20ms which isn't ideal for most genres imo.

[jorimt]

Not sure where you're getting your very specific 8.3ms number (which also happens to be the frametime of 120 FPS, so hope you're not conflating it with MPRT), but display processing latency is entirely dependent on the model, and whether said model, be it a CRT or OLED, is doing any digital processing to the image (and how much) before scanout.

Discounting any phosphor decay on CRT and near-black overshoot on OLED, both display types inherently have near instantaneous response, so assuming the display in question isn't performing any extra processing before scanout, the scanout latency itself should be very similar between the two. I.E. ~0-2ms range.

[NeonPizza]

Not sure where you're getting your very specific 8.3ms number (which also happens to be the frametime of 120 FPS, so hope you're not conflating it with MPRT), but display processing latency is entirely dependent on the model, and whether said model, be it a CRT or OLED, is doing any digital processing to the image (and how much) before scanout.

Discounting any phosphor decay on CRT and near-black overshoot on OLED, both display types inherently have near instantaneous response, so assuming the display in question isn't performing any extra processing before scanout, the scanout latency itself should be very similar between the two. I.E. ~0-2ms range.

Here's where i read about the 8.3ms of latency. And no, I'm not taking about the 8.3ms motion persistence(Or 16ms persistence from 60fps) that you get from 120fps games, just actual latency compared to what OLED offers. OLED having 10ms input lag at 60fps, and 5ms of lag at 120fps.

https://www.resetera.com/threads/crts-h ... ncy.40628/

I'm OCD when it comes to this stuff. I want the CRT experience with current QD-OLED technology, based on how it's 1ms persistence and whatever it's latency is. But it just isn't possible yet with the high latency coming from Black frame insertion, and even with BFI, the best you can get on something like a S90C QD-OLED is an 8.3ms persistence. It's baffling, it's now 2023, and the CRT's of the late 80's still have the huge edge in motion unless you run games at 144fps, which drops the persistence down to 6.94ms, while cutting input lag down to 4.5ms from 10ms. Realistically, you obviously need an expensive PC for that to happen and that's IF the game actually allows it. With console gaming, 144fps at this point is irrelevant and it's slim pickings for 120fps titles on PS5, and most likely won't be a thing on Switch 2.

BFI on QD-OLED still has like 50% more motion blur than a CRT, and the feature is practically useless anyways since it's bumps input lag up to nearly 30ms. An external BFI device that adds 'zero' additional input lag and could get the persistence down to 4ms or less would be amazing.

[jorimt]

Here's where i read about the 8.3ms of latency. And no, I'm not taking about the 8.3ms motion persistence(Or 16ms persistence from 60fps) that you get from 120fps games, just actual latency compared to what OLED offers. OLED having 10ms input lag at 60fps, and 5ms of lag at 120fps.

https://www.resetera.com/threads/crts-h ... ncy.40628/

That person appears to be referring to the frametime of 60 FPS, which is ~16.6ms, thus they're claiming CRT has ~8.3ms mid-screen latency.

However frametime can't strictly be considered latency; all displays, regardless of panel type, technically complete half scan-in of a frame at 60Hz in ~8.3ms, so I don't get their point.

Actual display latency, regardless of frame/refresh rate, is usually in reference to any processing applied before the rendered frame begins scanout to the display, which, again, can vary based on how much that display processes the image before scanout.

Finally, higher pixel response times (more commonly referred to as "GtG") is also sometimes considered a display latency factor, even though it technically doesn't delay, but instead ghosts/smears frame information instead.

Display processing, response time, BFI, and frametime "latency" stack and can all occur at once.

BFI on QD-OLED still has like 50% more motion blur than a CRT, and the feature is practically useless anyways since it's bumps input lag up to nearly 30ms.

You and I have discussed this in several threads now; OLED isn't natively impulsive-driven like CRT (hence, in part, its extra latency) and doesn't yet have enough sustained full-field brightness to compete with CRT motion when strobed (either globally or rolling scan, hence it's extra persistence blur). Until it does, it won't, simple as that.

We're a good while off from OLED being able to sustain such excessive brightness overhead without sacrificing form factor, energy costs, yield rate, and consumer market viability.

Until then, we just have to accept what's available, or go VR and deal with the limitations of that technology; for instance, my Quest 3 is comparable to CRT resolution and motion with media, but can't compete with OLED TV resolution or contrast ratio, so it will be interesting to see how the Beyond fares in that respect.

[Chief Blur Buster]

BFI on QD-OLED still has like 50% more motion blur than a CRT

More than 50%, if we're going by a CRT=1ms baseline.

and the feature is practically useless anyways since it's bumps input lag up to nearly 30ms. An external BFI device that adds 'zero' additional input lag and could get the persistence down to 4ms or less would be amazing.

A good external BFI injector such as Retrotink 4K can a low Hz (60-120Hz) to the same persistence as max Hz.

So on a 240Hz OLED, you can have 4ms persistence at a low refresh rate.

One problem is external BFI injectors have to buffer a slow-scan signal before fast-scan output. To bypass this, external BFI injectors can become zero-latency if you use QFT on the original signal. (e.g. configuring a PC to output 60Hz refresh cycles in 1/240sec). Then BFI goes lagless on a Retrotink 4K. And to preemptively answer, you can't use QFT on a console currently (PS or XBox), so you've still got the prerequisite buffering lag.

[Chief Blur Buster]

I've always assumed that CRT SDTV's had zero input lag, but now I'm hearing that they actually have 8.3ms.

Think "frame of reference" concept from Einstein's Theory of Relativity.

Not all pixels refresh at the same time!

1. Processing latency (0ms for CRT) -- this is a tape delay style latency behavior
2. Pixel to pixel latency (0ms for CRT) -- this is a tape delay style latency behavior
3. Scanout latency (at 60Hz it is 0ms for top edge, 16.7ms for bottom edge, 8.3ms for center) -- this is a latency difference between different pixels

The art of serialized refresh, means the upper-left corner refreshes at zero lag, but many lag testers (photodiodes / oscilloscopes) are put at screen center or crosshairs.

Also, sync technology can affect things, since it affects position of latency for a new frame. VSYNC OFF frameslices are subdivided latency gradients of [0...frametime] from top edge to bottom edge of frameslice between two tearlines.

So yes, a CRT has both 0ms latency and 8.3ms latency simultaneously -- depending on your frame of reference!

FRAME OF REFERENCE
FRAME OF REFERENCE
FRAME OF REFERENCE

(Or rather pixel of reference, since not all pixels refresh at the same time)

Proof:
- High speed videos of OLED and LCD www.blurbusters.com/scanout
- High speed videos of CRT www.google.com/search?q=high+speed+videos+of+crt

Not all pixels refresh at the same time = different lag numbers for different pixels.
Not all sync technology creates same lag-differentials for different pixels (so scanout latency behaviors can change from TOP<CENTER<BOTTOM to TOP=CENTER=BOTTOM or even TOP>CENTER>BOTTOM). So top edge can even become laggier than bottom edge, or bottom edge becomes laggier than top edge, depending on your display settings (sync technology setting, strobe settings, etc).

When one website tests latency of screen center, they're not always testing latency of top edge or bottom edge. And their settings matters; VSYNC ON versus VSYNC OFF.

A faster scanout (e.g. 240Hz OLED) can overcome the scanout latency problem of refresh rate (e.g. 60Hz CRT), so that's why even with extra tape-delay latency (e.g. +4ms), the 1/240sec scanout is 12.9ms faster than 1/60sec scanout. So brute refresh rate can overcome a latency bottleneck. There are no 240Hz 1080p or 1440p CRT tubes, so game over, OLED wins in latency contest against CRT when it comes to scanout latency, win win, despite being laggier in tape delay latency. Remember, latency is not a single number. It's like the tortise and hare race -- the CRT is zero tapedelay latency, but the 240Hz-ness is the rocket-speed scanout guaranteeing that last pixel never refreshes more than 1/240sec after first pixel. On a CRT 60Hz with fighting games (VSYNC ON), you're forced to wait 1/60sec=16.7ms between first and last pixel, due to scanout latency (different from absolute latency).

Now if you spew 1000fps VSYNC OFF, the 60Hz CRT may beat the 240Hz OLED, because of lack of tapedelay latency, and since VSYNC OFF latency is always [0...frametime] and 1000fps = 1ms frametime, you've kind of bypassed scanout latency with extra tearlines (over 10 tearlines per refresh cycle) splicing new frames continually mid-scanout. So you can get much lower latency with ultrahigh framerate VSYNC OFF, but you still only have 60 refresh cycle opportunities per second, which can still mean other pixels on the screen are missed (= laggier than 240Hz).

EXAMPLE:
60Hz WITH ZERO LAG: [T+0ms] [T+16ms]
240Hz WITH 2ms LAG: [T+2ms] [T+6ms] [T+10ms] [T+14ms] [T+18ms]

So see? Three extra refresh cycles that paints the full frame sooner, despite having 2ms more tapedelay latency.

Brute Hz can overcome some tapedelay latency. And guess what, 480Hz CRT's don't exist. So even with a bit of lag, upcoming 480Hz OLEDs of 2024-2025 with 1 framebuffer lag (1/480sec) = +2ms = still less laggy than a 0ms CRT if you're consider "first any pixel" reaction tests (high speed camera lag tests), instead of "single pixel latency" (single-point photodiode lag tests).

Big rabbit hole!

Lag is never a single number.

It's not Kindergarten 2+2.

It's a "University Calculus/Algebra" sized rabbit hole.

CRTs are great, but simplistic single number Luddite LagThink gotta go.

[tong]

It's also important mentioning that VRR@high refresh rates on 60fps content will produce the same latency as that high refresh.
ie, 240hz VRR@60fps will have the 240hz latency, that is ~4.4ms.

[thatoneguy]

The 8.3ms figure is some stupidity from resetera years ago where they measured from the middle of a 60hz crt.
CRTs have a few added microseconds of lag at best when measured at the top left corner of the screen.

60hz CRTs have 16.67ms lag at the end of the frame. This is called "scanout latency" which every display has and depends on refresh rate.

[jorimt]

It's also important mentioning that VRR@high refresh rates on 60fps content will produce the same latency as that high refresh.
ie, 240hz VRR@60fps will have the 240hz latency, that is ~4.4ms.

Yes, this is more commonly known as QFT (Quick Frame Transport); VRR does it dynamically, and fixed refresh can do it statically when manually configured.

60hz CRTs have 16.67ms lag at the end of the frame. This is called "scanout latency" which every display has and depends on refresh rate.

Yup.