How far are we from a rolling-scan capable flat-panel?

[oldgamer]

The reason we can't just dump our CRTs is that household grade flat-panels cannot present an image in a rolling scan fashion the way a CRTs does, which is a must for a perfect sense of motion for human eyes.

Are there any household grade flat-panel displays expected to finally support rolling-scan in the near future?

Is there any manufacturer at least talking about it? Is microled the only viable technology?

I presume the reason we don't see rolling-scan commonly implemented is because flat-panels currently are not capable of enough brightness to allow for it.

So maybe microled, given how bright it can get, is the only realistically feasible technology that will finally allow us to have proper motion again. Which is bad for us, as microled, if it does comeout in reasonable sizes, is gonna be very pricy, and manufacturers may not even care about implementing rolling scan.


I myself have many CRTs at home and would love to get rid of them if I could. It was lot of work hunting for them, and storing them also is big pain in the ass. It would be great to simply buy a single flat-panel that substitutes all the CRTs and solve this problem for good.

[tongshadow]

OLED BVMs have rolling scan, but we're talking about expensive pro-grade equipment. On the consumer side things dont look promising, LG removed 120hz BFI modes from 2022 models and doesnt seem like it's making a comeback in 2023. The feature isnt even present on their 240hz OLED gaming monitor, and other brands dont seem interested in implementing BFI. I have the impression they completely gave up, we're going backwards!

MicroLED, despite its many benefits over OLED, remains a sample and hold display. The only way to improve motion blur on this type of display is to increase refresh rates, often to impractical values, or BFI.
I personally wouldnt hold my breath for MicroLED, I think it's only going to truly benefit people who would like to consume HDR content at +1000nits.

I dont know what kind of CRT you have, but if it's a pro-grade monitor or a high-end PC monitor, it's worth keeping them around to remind yourself what good motion clarity looks like.

[bigbadboaz]

As above: companies simply don't care about this at the consumer level; there has been no significant pushback from the mass market regarding the state of motion clarity on modern television displays.

It's not coming, regardless of technical feasibility.

[Guspaz]

They (consumer TVs) all technically have rolling scans, and with TVs like LG OLEDs, the BFI is technically rolling scan too. However, I understand that what you're referring to is not rolling scan, but a sort of dual rolling scan where the BFI scan line is on-screen at the same time (but delayed from) the active signal scan line. Sadly, I don't think that's going to ever happen. And the motion clarity from 60 Hz BFI is probably good enough anyway, what's not ideal is how they're done such that they add extra lag. The path the industry has chosen to higher motion clarity is clearly higher framerates. A game running at a sufficiently high framerate on a high refresh rate monitor with no BFI is going to have better motion clarity than a 60 Hz CRT. Doesn't help much for console games. Frame generation (DLSS 3) helps you hit very high framerates, thus significantly increasing motion clarity, but in that case, at the cost of even more lag. There are side-effects in the generated frame, but they tend to be quite minimized when you're feeding it sufficiently high input framerate, like going from 120 to 240 FPS.

[Josh128]

Its never going to happen. At least not by a TV manufacturer. It'd have to be up to some mad scientist/engineer like Mike Chi to make a scaler/device that would make the display do it, once we have displays fast enough to do it, kind of like where we are today. If the current MicroLED response time claims are true, that would easily be fast enough.

If pixel response were fast enough to simulate a CRT beam scan and effectively emulate the amount of phosphor illumination persistence of a CRT, it would work, but you'd lose a ton of brightness and if pixel response is that fast anyway, you just as well just sample and hold as its done now.

Emulating phosphor illumination persistence would be the holy grail, but again, you'd be sacrificing a HUGE amount of brightness to do it properly.

https://imgur.com/XkOVqTA

https://i.imgur.com/EyB1JqG.jpg

[bobrocks95]

if pixel response is that fast anyway, you just as well just sample and hold as its done now.

The fast pixel response time of OLED's actually makes the effects of sample and hold even worse since you don't get any kind of blurring or smearing between frames that actually makes things look smoother to the eye. See rting's "Stutter" section for their TV reviews.

[Josh128]

if pixel response is that fast anyway, you just as well just sample and hold as its done now.

The fast pixel response time of OLED's actually makes the effects of sample and hold even worse since you don't get any kind of blurring or smearing between frames that actually makes things look smoother to the eye. See rting's "Stutter" section for their TV reviews.

I disagree with that opinion, because I have a lot less noticeable blur on my iPhone 12 when scrolling than I do on my 75Hz LED backlit LCD monitor, and thats an OLED, not a microLED. But I digress-- in any case, with a display that fast, all kinds of crazy BFI modes would be possible, such as 1:2, 2:1, 2:3, 3:2, 1:3, 3:1 etc. etc. or much more options for image to black frames per frame, or "fading" images and black frames, if it was designed with that option or a video processor enabled it. Such capabilities would render the need for a rolling scan unnecessary.

[oldgamer]

MicroLED, despite its many benefits over OLED, remains a sample and hold display. The only way to improve motion blur on this type of display is to increase refresh rates, often to impractical values, or BFI.

But I think in theory at least a microled doesn't have to be sample and handle, it should be totally possible to make display images in a rolling scan fashion. I remember reading chief blur busters saying rolling scan is easy, given it was implemented on pro OLED displays.

They (consumer TVs) all technically have rolling scans, and with TVs like LG OLEDs, the BFI is technically rolling scan too. However, I understand that what you're referring to is not rolling scan, but a sort of dual rolling scan where the BFI scan line is on-screen at the same time (but delayed from) the active signal scan line. Sadly, I don't think that's going to ever happen... The path the industry has chosen to higher motion clarity is clearly higher framerates

Not sure how the LG OLEDs do BFI, perhaps is that rolling window like sony pro OLEDs do? But I really mean just regular CRT rolling scan, one line is displayed then the next while the previous one starts fading, just like CRT phosphors.

Perhaps the reason the industry chose the high fps path is simply because current flat-panels just doesn't have enough brightness for rolling scan but could maybe switch to rolling scan again if proper tech emerged (I'm hoping this is the case ).

Maybe going back to rolling scan is even easier to achieve proper high fps like 480fps+, because, how are graphics cars going to push that much fps on any modern games?

If the current MicroLED response time claims are true, that would easily be fast enough.

If pixel response were fast enough to simulate a CRT beam scan and effectively emulate the amount of phosphor illumination persistence of a CRT, it would work, but you'd lose a ton of brightness and if pixel response is that fast anyway, you just as well just sample and hold as its done now.

Emulating phosphor illumination persistence would be the holy grail, but again, you'd be sacrificing a HUGE amount of brightness to do it properly.

Technology wise it should be possible to do just that, display a line at time then fade the previous one with similar timings as phosphors, given how bright microled is and each microled being its own source of light. BUUUUUTTTTT...

As above: companies simply don't care about this at the consumer level; there has been no significant pushback from the mass market regarding the state of motion clarity on modern television displays.

It's not coming, regardless of technical feasibility.

This is what I fear the most, that because most people don't push about this manufacturers in turn don't care as well, simply forgetting the marvels of rolling scan and good motion.

I wonder how feasible would it be to for us to raise awareness for the general public that great motion in monitors used to be a thing with CRTs and we want that back. If enough people knew perhaps we have a change.

There are probably many kids today that have never seen how good motion is on a CRT but would instantly "need" that in their flat-panel monitors if they knew.

[oldgamer]

if pixel response is that fast anyway, you just as well just sample and hold as its done now.

The fast pixel response time of OLED's actually makes the effects of sample and hold even worse since you don't get any kind of blurring or smearing between frames that actually makes things look smoother to the eye. See rting's "Stutter" section for their TV reviews.

I disagree with that opinion, because I have a lot less noticeable blur on my iPhone 12 when scrolling than I do on my 75Hz LED backlit LCD monitor, and thats an OLED, not a microLED. But I digress-- in any case, with a display that fast, all kinds of crazy BFI modes would be possible, such as 1:2, 2:1, 2:3, 3:2, 1:3, 3:1 etc. etc. or much more options for image to black frames per frame, or "fading" images and black frames, if it was designed with that option or a video processor enabled it. Such capabilities would render the need for a rolling scan unnecessary.

But wouldn't it be just easier to go with proper rolling scan than implementing all those BFI modes?

I have used software BFI with retroarch on a 120HZ flat-panel, 1 frame of content, one black frame type of thing. It improves motion a lot, but not fully as good as a good ole CRT tv at 60Hz. Maybe it would work better in a 1 frame of content and 2 or 3 of blackness, but then brightness would be punished even more.

[orange808]

if pixel response is that fast anyway, you just as well just sample and hold as its done now.

The fast pixel response time of OLED's actually makes the effects of sample and hold even worse since you don't get any kind of blurring or smearing between frames that actually makes things look smoother to the eye. See rting's "Stutter" section for their TV reviews.

That's probably related to the display's built-in 24Hz film "pulldown" frame rate conversion and interpolation video processing. Our games shouldn't be subject to frame rate conversion. (Although, I did briefly own a Samsung years ago that performed frame rate conversion on NES and SNES inputs and the scrolling wasn't smooth. Why? I couldn't say. I don't remember the frame rate range that the display worked properly. I used the DVDO judder test pattern to verify it and sent that piece of shit Samsung back. You never know what you're going to get.)

With video games, the OLED is definitely to my eye. (I'm talking ~50-60Hz refresh rates with no "BFI" tricks here. Just sample and hold. Please don't move the goal posts around. We all know what refresh rates our legacy consoles and PCB's output.)

OLED has better clarity. The truly instant response time of a DLP projector produces even clearer motion. All you have to do is turn on NES Megaman. Those games have moments where the gameplay stops and the screen performs a scroll transition to the next area. You get a moment to really take it all in--and the blur is always obvious. The smearing on any LCD is especially gross.

I have no idea what film content looks like using the built in processing on an LG OLED. I don't use one of those for that use case. If that's a problem, it's not the response time, it's LG's video processing.

[Guspaz]

The limitation is the display controllers, not the panel tech. Yes, if you had OLED subpixels with a 0.1ms response time, a display controller could update the pixel brightness ~167 times per frame and fully simulate CRT raster rendering. But that's an absurd amount of throughput from the display controller. Like, approaching the terabits per second range for 4K.

However, I don't think we need to do that to trick the eye. Refresh rates on monitors is being cranked higher and higher. Today, 540 Hz is the fastest. I imagine they're going to continue increasing to 1000 Hz, both because that's what the enthusiasts have been asking for, but also because 1000 Hz is a nice number for marketing. So let's say we get that. We run it at 960 Hz. That gives us 16x the framerate of our 60 Hz CRT. Now imagine that you simulate the CRT raster in 1/16th of a frame. That is, you basically draw a CRT-style raster beam (with a fadeoff) 1/16th down the frame, then 2/16th down the frame, then 3/16th down the frame, and so on. Is your eye really going to be able to tell the difference between that CRT-style raster updating in (taking a 240p example) one scanline at a time versus 15 scanlines at a time? No, I think that part would be identical to your eye.

I don't think the brightness is there to do that yet, but I think it would otherwise look identical to your eye.

[orange808]

Frame generation (DLSS 3) helps you hit very high framerates, thus significantly increasing motion clarity, but in that case, at the cost of even more lag. There are side-effects in the generated frame, but they tend to be quite minimized when you're feeding it sufficiently high input framerate, like going from 120 to 240 FPS.

I'm not getting those frame rates in new games using DLSS 3 with detail settings at ultra. Look below. If I'm not getting them, who is?

Spoiler

[tongshadow]

But I think in theory at least a microled doesn't have to be sample and handle, it should be totally possible to make display images in a rolling scan fashion. I remember reading chief blur busters saying rolling scan is easy, given it was implemented on pro OLED displays.

Rolling scan is a "hacky" way to overcome the issues with sample and hold, and while it improves clarity it increases flickering, so it could be uncomfortable for most viewers. According to Vincent from HDTV, it increase motion resolution from 300 lines to 700 lines:

https://youtu.be/ESzWY0hW85Y?t=1001

Not a bad improvement and it doesnt cause a drop in Luminance. The thing is that a proper BFI implementation isnt hard at all, look how the Viewsonic XG2431 looks at 60hz!
https://www.rtings.com/assets/pages/q2a ... -large.jpg

Guess our hopes are on Viewsonic making OLED monitors

[Guspaz]

Rolling scan does decrease luminance dramatically, that BVM that Vincent is reviewing is compensating by cranking the brightness up by the proportional amount.

[oldgamer]

The limitation is the display controllers, not the panel tech. Yes, if you had OLED subpixels with a 0.1ms response time, a display controller could update the pixel brightness ~167 times per frame and fully simulate CRT raster rendering. But that's an absurd amount of throughput from the display controller. Like, approaching the terabits per second range for 4K.

However, I don't think we need to do that to trick the eye. Refresh rates on monitors is being cranked higher and higher. Today, 540 Hz is the fastest. I imagine they're going to continue increasing to 1000 Hz, both because that's what the enthusiasts have been asking for, but also because 1000 Hz is a nice number for marketing. So let's say we get that. We run it at 960 Hz. That gives us 16x the framerate of our 60 Hz CRT. Now imagine that you simulate the CRT raster in 1/16th of a frame. That is, you basically draw a CRT-style raster beam (with a fadeoff) 1/16th down the frame, then 2/16th down the frame, then 3/16th down the frame, and so on. Is your eye really going to be able to tell the difference between that CRT-style raster updating in (taking a 240p example) one scanline at a time versus 15 scanlines at a time? No, I think that part would be identical to your eye.

I don't think the brightness is there to do that yet, but I think it would otherwise look identical to your eye.

Took me a little while to understand your idea of how to simulate a sort of rolling scan behavior.

Now I think I got it: We have a 960Hz monitor to display 60HZ content, so that would be 16 monitor frames to display a single frame of content, so we divide that frame of content in 16 parts, each being 15 lines tall.

15 lines of a frame are displayed in one monitor frame, i.e 1/16 of of a 240p frame in 1/16 of 60Hz frame, i.e one of those 960 frames a second.

Then in the next monitor frame, the next 15 lines are displayed after the first 15 while the previous 15 start to "fade". Then in the next monitor frame, the third 15 lines are displayed while the previous two are fading still.

An so on. That is the idea right?


That seems like it could work, for emulation at least.

Current flat-panels are not bright enough, but Microled likely is. The bad part is we're still far away from 4K/8K at 1000Hz. HDMI/displayport specs don't even have these numbers in sight.

Oh sigh, no true solution seems realistic in a short time frame.

[Guspaz]

That's the basic gist of it. The falloff could be smooth too, it's just that the simulated CRT-style scanline will have moved 15 lines down every monitor frame. You'd never get this in a display controller, it'd need to be in software or a scaler.

As far as bandwidth is concerned, you can do 4K960 4:4:4 with DisplayPort 2.0/2.1 and DSC. Current DisplayPort 1.4 should be able to do 1440p960 with DSC and 4:2:2. I don't think bandwidth is a limiting factor.

[strayan]

Rolling scan is a "hacky" way to overcome the issues with sample and hold, and while it improves clarity it increases flickering, so it could be uncomfortable for most viewers. According to Vincent from HDTV, it increase motion resolution from 300 lines to 700 lines

I use a 25 inch Sony OLED with rolling scan as my primary gaming monitor. There is definitely flicker but I don’t notice it while gaming. As soon as you go back to the desktop though it’s unbearable.

[fernan1234]

The rolling scan in those Sony pro OLEDs works well (for video and gaming content) because they are industrial grade and have lots of passive and active cooling to let the panels get bright enough to compensate the brightness (as well as top notch process that allows this compensation to not affect colors and gamma). Plus the panel is small so heat is less of a problem than on big TVs. You're not gonna find those cooling fans and bulky frames on a commercial display, not even a gaming monitor.

So OLED is out, but I don't think you need to wait for microLED though. miniLED with good dimming zone count and algorithm can be good enough, it'll give you the nits needed in HDR mode, but the solution for BFI/rolling scan will need to be external. Hopefully the next Retrotink scaler as someone said.

[orange808]

"Frames per second" don't really matter.

OLED panels have very fast response and essentially random access to every "pixel". The display controller only needs the ability to display a rolling scan. A string of lights on a marquee that flickers quickly to create a flashy pattern doesn't have any frames per second. It's a circuit. It's just a circuit.

Obviously, an OLED display controller isn't a simple circuit, but the example is still useful. I don't need the ability to handle ridiculously high frame rate inputs from a wired external source. I don't need the ability to unpack and process a high frame rate input from external devices. It has little to do with the display's ability to handle external sources. It's a matter of how the display controller handles the panel. RAM is sufficiently fast to update whatever functional "display surface buffer" there is and the OLED panels already have fast response.

All we're doing is activating pixels on a fast panel using a display controller--and we can manipulate information from the current frame very quickly. I don't understand why brute forcing a rolling scan over the wire using an external device is even a suggestion. That's a ridiculously expensive and inefficient idea. The display controller has direct access to the current frame information and the panel. The RAM and processing power are already fast enough. It's not a really a string of lights and a simple circuit, but the same principle applies.

What you need is a display controller that will display and fade pixels on the panel in a specific way. If a hardware manufacturer wanted to, we have multiple existing avenues to design a display controller that would active "pixels" in a specific way. We already have the ability to perform the processing at suitable speeds.

I'm not saying other roadblocks don't exist. I just don't understand talk about sending it down the wire.

[fernan1234]

I'm not saying other roadblocks don't exist. I just don't understand talk about sending it down the wire.

The reason is simple: no manufacturer will ever care about this (making legacy <60fps content look nearly as smooth as it did on CRTs).

[tongshadow]

Seems like the perfect opportunity to mention that only because a display technology becomes more accessible, doesnt mean we will be getting the best implementation of it. True RGB OLEDs, for example, still only exist in the professional market while WRGB became the more widely available option for consumers due to cost.
So even if there are benefits to the microLED technology, what ends up in the hands of consumers might be a less than ideal implementation of it. So it's good to keep expectations in check.

[Josh128]

I'm not saying other roadblocks don't exist. I just don't understand talk about sending it down the wire.

The reason is simple: no manufacturer will ever care about this (making legacy <60fps content look nearly as smooth as it did on CRTs).

As stated on the 3rd and 5th posts on this thread and repeated numerous times!

[orange808]

Well, sure. Current generation console gamers don't need rolling scan.

With so much console software only invalidating and updating the viewport 30 or 40 times per second (and repeating frames to offer higher "refresh rates"), there's really nothing we can do. The most popular gaming platforms don't offer motion resolution by design (and necessity).

Regardless, there are display reviews out there that explain motion resolution to consumers. We'll see if people start to care and demand better performance. Right now, neither display manufacturers or software houses care very much about motion resolution. Truthfully, they don't care enough about latency, either.

[fernan1234]

Regardless, there are display reviews out there that explain motion resolution to consumers. We'll see if people start to care and demand better performance.

They won't. People have had shitty motion resolution for how many years now, around 20? And when reviews talk about it they mainly refer to motion interpolation for film content or video. When BFI/flicker gets brought up, the usual advice is to not use it because of dimming and flicker, giving manufacturers even less incentive to improve it or even add it as a feature.

It's not like CRTs had excellent motion resolution by design either. It was a happy byproduct of the way they happen to work to display a picture.

[orange808]

It's a complicated space.

More people are seeing legitimate real high refresh rates in action on their mobiles. That trend will continue. I don't know where things will go. Consumers have high refresh displays that they use often, plenty of scrolling to showcase it, and there's a lot of marketing from manufacturers.

Hard to say what people will demand in the future. You could be right. For now, consoles are getting away with lies. They upscale to 4k and they repeat frames to 120Hz. People don't seem to notice or care.

[tongshadow]

What your average consumer only cares about in a display is: Brightness. They wont be satisfied until they have a Sun in their living rooms!
All improvements of OLED technology and the whole LCD/LED vs OLED debate stems from the demand of greater brightness levels, even though the improvements are small enough to not be of any practical use or make difference on real content.

HDR is another interesting case. Before, we had SDR. So content was mastered in a way that would look correct on most displays. But with HDR it's a completely different story, some content is mastered to look correct with a display that dont you even have! It's almost as if... they're trying to force you to buy newer and brighter displays, how convenient is that!? Now you can easily make LCD/LED remain relevant, even though OLED is superior in every conceivable way, AND make OLED users buy new OLED models, even though their brand new TV is still fine.

Meanwhile, this is how a $1000+ monitor displays 60hz content, imagine playing retrogames on this:
https://i.rtings.com/assets/products/9Z ... -large.jpg

And without a strobed display technology in the horizon, I guess it's safe to say we're screwed.

[orange808]

Meanwhile, this is how a $1000+ monitor displays 60hz content

That's nothing. Imagine the blur with a "120Hz" console game that actually pushes new refreshes at 40fps and repeats each frame three times. It's really 40Hz.

I guess if people are dumb enough to believe it looks super clear when it obviously does not, it's hopeless. Although, people should be able to see it's not right. Their mobiles actually refresh at high frame rates in most cases. If people can't tell they're getting lousy motion resolution, then they won't demand anything better. We'll see how it goes.

[Fudoh]

Is there any scientific background to why the usual sample and hold motion resolutions defaults to something like 300 to 350 lines ? I know the test patterns and the numbers seem to make sense.

I'm also wondering how higher fps directly translate into higher perceived resolution (like Vincent's mentioning of 700 lines of motion resolution with BFI on the LGs compared to 10+ year old ratings of 1080 lines of motion resolution for 200+ Hz TVs with combined interpolation, black frame insertion and strobing backlight). There really doesn't seem to too much reliable background info out there...

[Josh128]

Is there any scientific background to why the usual sample and hold motion resolutions defaults to something like 300 to 350 lines ? I know the test patterns and the numbers seem to make sense.

I'm also wondering how higher fps directly translate into higher perceived resolution (like Vincent's mentioning of 700 lines of motion resolution with BFI on the LGs compared to 10+ year old ratings of 1080 lines of motion resolution for 200+ Hz TVs with combined interpolation, black frame insertion and strobing backlight). There really doesn't seem to too much reliable background info out there...

Is that sample and hold figure ~ universal regardless of pixel response time? If so its almost certainly due to limitations / workings of the human eye and brain. It seems logical that strobing + BFI might be superior to BFI alone.

[tongshadow]

Is there any scientific background to why the usual sample and hold motion resolutions defaults to something like 300 to 350 lines ? I know the test patterns and the numbers seem to make sense.

I'm also wondering how higher fps directly translate into higher perceived resolution (like Vincent's mentioning of 700 lines of motion resolution with BFI on the LGs compared to 10+ year old ratings of 1080 lines of motion resolution for 200+ Hz TVs with combined interpolation, black frame insertion and strobing backlight). There really doesn't seem to too much reliable background info out there...

Blurbusters goes into detail regarding old test patterns for motion resolution:
https://forums.blurbusters.com/viewtopic.php?t=4759
Apparently it's an outdated way to measure motion resolution and their MPRT/GTG method gives results more in line with reality.