Create input lag database of processors/scalers/converters?

[thetallguy24]

There seems to be just a random assortment of opinions, etc about the devices we all discuss, but there doesn't seem to be an actual list or database of how much input lag each of them create (in frames or ms). Is that a running thread we could create here (somebody have these and a 240p test suite?). This would put to rest a lot of (sometime false) preconceived notions. For example, someone on a smash board claims that the VSC 500 has too much lag to play.

Here's a few that come to mind:

31khz Console (VGA/Component 480p) -> Extron VSC 500/700 -> 15khz CRT TV
31khz Console (VGA/Component 480p) -> Extron VSC 500/700 -> Extron RGB 203 rxi -> 15khz CRT
15khz Console (240p/480i) -> Extron 4LDxi -> 31khz VGA Monitor
15khz Console (240p/480i) -> Extron Lancia xi -> 31khz VGA Monitor
15khz Console (240p/480i) -> Extron Andora -> 31khz VGA Monitor
15khz Console (240p/480i) -> Faroudja LD100/200/400 -> 31khz VGA Monitor
31khz Console (VGA/Component 480p) -> Mimo Genius II -> 15khz CRT
31khz Console (VGA/Component 480p) -> Extron Emotia/Super Emotia -> 15khz CRT
15khz Console (240p/480i) -> XRGB 1/2/2+/3(b0)/3(b1) -> 31khz VGA Monitor
15khz Console (240p/480i) -> Videon Omega One/Vigatec FX2 -> 31khz VGA Monitor

And on and on. Essentially, a database of input lag for any device from Fudoh's amazing website.

I think a member created database as a running thread would be a good thing to start building. It would eliminate a lot of threads too.

[Dochartaigh]

I think the biggest problem here will be testing procedure, especially when we're talking about microseconds or whatnot. How would you get everybody to test exactly the same?

For example, when I was doing lag tests on native 480p on a multiformat CRT vs 480p downscaled to 240p on another CRT, something as little as the orientation you hold the camera can actually skew the results...that kinda blew my mind (and they were right).

Even some SD CRT's have lag. Modern TV's have all sorts of processing and features you can turn on and off which can change the results...so how to get a baseline unless everybody is using the EXACT same everything equipment-wise.

[thetallguy24]

I think the biggest problem here will be testing procedure, especially when we're talking about microseconds or whatnot. How would you get everybody to test exactly the same?

For example, when I was doing lag tests on native 480p on a multiformat CRT vs 480p downscaled to 240p on another CRT, something as little as the orientation you hold the camera can actually skew the results...that kinda blew my mind (and they were right).

Even some SD CRT's have lag. Modern TV's have all sorts of processing and features you can turn on and off which can change the results...so how to get a baseline unless everybody is using the EXACT same everything equipment-wise.

I doubt standardization is possible, but I think who submits a result should state the setup chain (Console->cables->device->TV) and what model CRT/Monitor/LCD they are using. With LCDs in particular, they should show if a Game Mode or anything like that was on. Ultimately, it would be a database of ranges.

[Fudoh]

I will eventually (sometime next year?) test ALL the video processors I still got using some unified method. And I still got LOTS of machines

Baseline would basically be a Bodnar lag tester converted to 15khz. We just need a base measurement of the the device chain through a known-to-be-lag-free processor like the OSSC.

[thetallguy24]

I will eventually (sometime next year?) test ALL the video processors I still got using some unified method. And I still got LOTS of machines

Baseline would basically be a Bodnar lag tester converted to 15khz. We just need a base measurement of the the device chain through a known-to-be-lag-free processor like the OSSC.

You can use the Bodnar with 15khz CRTs? So Bodnar to what kind of downscaler/converter? Bodnar -> downscaler/converter -> OSSC?

[maxtherabbit]

I will eventually (sometime next year?) test ALL the video processors I still got using some unified method. And I still got LOTS of machines

Baseline would basically be a Bodnar lag tester converted to 15khz. We just need a base measurement of the the device chain through a known-to-be-lag-free processor like the OSSC.

If you're going to use the OSSC, why not just employ the "DIY latency test" mode it has? As long as you had known lag-free DACs to attach to the OSSCs HDMI output, it should allow for testing of all other scalers.

[orange808]

You wouldn't want to do any camera testing using all digital displays, anyway.

You can get a general idea of latency with a camera and two CRT's (fairly easily) if you mind the rolling shutter on your camera.

I use a consumer CRT and an old "VGA" monitor. The Hi Def NES can output the 240p Test Suite timer to both composite and HDMI. There's a small amount of latency from the Hi Def NES and the DAC, but it's not significant.

Switching out the DAC and monitor for direct HDMI into digital displays produces the expected lag the Leo Bodnar predicts for each digital display.

So, that's the way I test.

It's not in milliseconds, but it doesn't need to be.

[orange808]

Also, FYI, the Extron scan converters have ~1 frame of lag.

That's playable.

You'll need some very precise gear and a lot of time on your hands to find the latency of an Extron RGB. That's a very very small number.

[thetallguy24]

Also, FYI, the Extron scan converters have ~1 frame of lag.

That's playable.

You'll need some very precise gear and a lot of time on your hands to find the latency of an Extron RGB. That's a very very small number.

I remember Fudoh once saying that the Extron scan converters have no lag.

Do the Extron (Andora/Lancia/4LDxi) or Faroudja line doublers have any lag?

[orange808]

I remember Fudoh once saying that the Extron scan converters have no lag.

Maybe I'll give it a second look sometime. Although, with frame rate conversion, at least a frame makes sense.

Do the Extron (Andora/Lancia/4LDxi) or Faroudja line doublers have any lag?

Sorry. Don't have those machines.

[Xer Xian]

Regarding the VSC 500, the Leo Bodnar tester (plus Hdfury) picks up between between 4-5 and 17-18ms of lag on my setup.

I can't perceive a latency of 1-1.5 of a frame in any way - only when it gets to somewhere around 3 frames I start to feel iffy and uneasy, and this is on skill-based games on which I've logged hundreds of hours. Honestly, I think that whoever 'feels' a latency of less than 2 frames is under some kind of nocebo effect..

[maxtherabbit]

Regarding the VSC 500, the Leo Bodnar tester (plus Hdfury) picks up between between 4-5 and 17-18ms of lag on my setup.

I can't perceive a latency of 1-1.5 of a frame in any way - only when it gets to somewhere around 3 frames I start to feel iffy and uneasy, and this is on skill-based games on which I've logged hundreds of hours. Honestly, I think that whoever 'feels' a latency of less than 2 frames is under some kind of nocebo effect..

Problem is that most displays add a frame or two as well, so getting up to 3 frames when stacking on a scaler is easy to do

[Xer Xian]

And that is why you should always be gaming on good ol' CRTs

[Unseen]

You'll need some very precise gear and a lot of time on your hands to find the latency of an Extron RGB. That's a very very small number.

A signal generator and a fast oscilloscope should be enough. I'm not sure if it's already in a region where you need to make sure that the cable lengths for the reference and measured signal paths are (sufficiently close to) identical.

Alternatively, you could use a Time Interval Counter instead of an oscilloscope, but that is slightly more obscure hardware.

[thetallguy24]

And that is why you should always be gaming on good ol' CRTs

Sometimes people want ways to play an NES lag free on a 31khz CRT monitor.

[orange808]

Sometimes people want ways to play an NES lag free on a 31khz CRT monitor.

Yep.

But, remember that the Extron scan converters output 15Hz and there's very few reasons why anyone would add any other scalers to the chain--between the console and the 15Hz CRT.

So, it's just the worst case scenario of one frame--and that's not enough to worry about.

[thetallguy24]

Sometimes people want ways to play an NES lag free on a 31khz CRT monitor.

Yep.

But, remember that the Extron scan converters output 15Hz and there's very few reasons why anyone would add any other scalers to the chain--between the console and the 15Hz CRT.

So, it's just the worst case scenario of one frame--and that's not enough to worry about.

Yes, but again, there are plenty of modern consoles people want to play on VGA CRT Monitors and PVMs/BVMS

[orange808]

Sometimes people want ways to play an NES lag free on a 31khz CRT monitor.

Yep.

But, remember that the Extron scan converters output 15Hz and there's very few reasons why anyone would add any other scalers to the chain--between the console and the 15Hz CRT.

So, it's just the worst case scenario of one frame--and that's not enough to worry about.

Yes, but again, there are plenty of modern consoles people want to play on VGA CRT Monitors and PVMs/BVMS

We probably need multiple threads to start breaking out each individual use case.

Sorry I couldn't help with latency numbers for the other hardware on your list.

[Xyga]

Problem is that most displays add a frame or two as well, so getting up to 3 frames when stacking on a scaler is easy to do

And that is why you should always be gaming on good ol' CRTs

Come on, 'most' applies for people who don't take a moment to inform themselves and just buy whatever at the nearest store or on the internet wherever there's plenty of happy comments with full stars customer ratings.

There's enough decent-to-great TVs, and tons of monitors, with barely any lag that are available in all price ranges around the globe these days.
Lagless monitors (0~4ms) have become so common it's practically the standard, and for TVs if you look at Rtings it's crazy how so many available are almost as responsive as monitors now.
Remember in their test they still give you middle-screen measurements which is not 'input lag period' but input lag+1/2frame@60, so sustract 8.3ms
For instance say you want a TV for displaying a device or two that you know will produce/cost 2 frames, but don't want your input lag chain over 2.5f (42ms); if you pick a model in their list they measured under 16ms @1080p, you're safe.
Even though we're at the end of 2018, a lot of 2017 models are still relatively easy to find, and considering the many sizes, all combined with 2018's thats quite a large choice.
Apply this to a less laggy chain and the number of eligibe models increases significantly, even up to the LG OLEDs that are just under 1 frame.

I mean it's fine and all to go all-CRT if you're really worried about lag, but it's not difficult to pick a flat panel that's as responsive or adding so little delay that it'll not be a problem.

[maxtherabbit]

Come on, 'most' applies for people who don't take a moment to inform themselves and just buy whatever at the nearest store or on the internet wherever there's plenty of happy comments with full stars customer ratings.

There's enough decent-to-great TVs, and tons of monitors, with barely any lag that are available in all price ranges around the globe these days.
Lagless monitors (0~4ms) have become so common it's practically the standard, and for TVs if you look at Rtings it's crazy how so many available are almost as responsive as monitors now.
Remember in their test they still give you middle-screen measurements which is not 'input lag period' but input lag+1/2frame@60, so sustract 8.3ms
For instance say you want a TV for displaying a device or two that you know will produce/cost 2 frames, but don't want your input lag chain over 2.5f (42ms); if you pick a model in their list they measured under 16ms @1080p, you're safe.
Even though we're at the end of 2018, a lot of 2017 models are still relatively easy to find, and considering the many sizes, all combined with 2018's thats quite a large choice.
Apply this to a less laggy chain and the number of eligibe models increases significantly, even up to the LG OLEDs that are just under 1 frame.

I mean it's fine and all to go all-CRT if you're really worried about lag, but it's not difficult to pick a flat panel that's as responsive or adding so little delay that it'll not be a problem.

Most people also already have a flat panel of some sort and probably aren't into the same replacement cycles that you seem to be

[Xyga]

Most people also already have a flat panel of some sort and probably aren't into the same replacement cycles that you seem to be

What makes you believe that? same TV for about six years, and since then bought four large 'entry+' (27"/32") level monitors @+/-200 each, sold two. Rest are CRTs I got for free (relatively easy in Europe). Because I've tried a lot of displays over the years doesn't mean I've bought them all! Being around gaming communities all those years I've met enough people who buy new big TVs like every couple of years (mostly americans probably bc prices and availability are the best), I'm far from that.
What I mean is despite all that's been written and tested since well over a decade now that this little world started getting interested in flat panels performance, we still regularily read generalizations I'm only saying aren't true anymore. In fact for a lot of people in many countries it's become easier and cheaper to procure a lagless lcd than a crt.
And anyway think about it; wouldn't it be odd to seek precise lag information for processors/scalers/converters for building a setup, yet stop at a misconception that technically cuts one's options on the display side of the matter by like 90%? I'm not saying that to bother you and Xer Xian, just shoving in facts to consider for the topic discussed here.

Rather it's those processors/scalers/converters that pose more issues in terms of spendings rationality, IMHO a lot of processors in particular have become obsolete. A database will surely highlight that, I think.

[Fudoh]

IMHO a lot of processors in particular have become obsolete

absolutely. Even the vast majority of them. And even more, if you limit your reqirements to gaming purposes.

In terms of down converting the TVOnes are the obvious choice now and it's hard to find niche applications to justify any of the other scan converters.

For upscaling I would still divide the field into processors suited for upscaling 240p content, those nice for 480i and those great for 480p. For 240p there are the Retrotink, the OSSC and the FM. I don't think anything else makes sense on a modern setup. 480i feel at home on a Framemeister or a good DVDO machine and 480p is best processed on an Extron DSC unit (or a CII if you have the money).

[Xer Xian]

@Xyga: Flat panels may now be at a point where lag is not an issue, but are they as good as CRTs at keeping females away from your man cave? We now how important it is to have an uninterrupted focus while playing shmups and retrogames in general, and avoiding female disturbances is key. Most human females tend to be repelled by the simple sight of a CRT, while they are indifferent, or in some case even attracted by, modern flat panels TV.

Therefore, we can conclude, with mathematical precision, that the all-important gal-rejection to lag-injection ratio is a neighbourhood of zero for flat panels, but diverges to infinite for CRTs, which makes them absolute winners today and forever.

Ball is on your court

[Xyga]

Maybe so but are you sure the repulsion ratio is the same for flat and curvy gals?