120hz & 24p input support in LCDs: Fact or Myth?

[dmzguy]

I know this is a REALLY long, REALLY geeky post, for those that just want the point here it is: It is VERY unlikely that any of the brand new 120hz displays will get rid of "film judder"; even if they accept 24p as an input.




All of the review sites, much of the news at CES this year, and many of us LCD enthusiasts here have been hearing a lot about some of the new high-end LCD TVs supporting 120hz refresh rate (the Sharp D92U, the Toshiba LX177 (that has all of a sudden dissapeared from the Toshiba website), supposedly the Samsung 81 series when it comes out and maybe another Samsung TV that used standard CCFL backlighting).

The commonly repeated mantra is that 120hz is the PERFECT solution to all of LCD's motion-related shortcomings and that it IS the perfect solution because 120hz is an even multiple of 24hz; this allegedly means that these TVs will accept a 24p signal and will then just display every frame 5 times, (or if the TV is give a 60hz signal it will perform inverse telecine to get back to the 24hz signal and it will then display each of those frames 5 times) which would eliminate any 3:2 cadence-related "motion judder".

The problem with this theory is that LCDs are a store and hold technology, and LCD's pixels will display an image and will then change when they are sent a new signal, so if an LCD's processor is taking 24 frames every second and then sending each of those 24 frames to the monitor 5 x for a total of 120 "refreshes" every second, the problem is that there is no information sent in the second, third, fourth, or fifth frame so the image displayed on screen will not actually change until the 6th frame, the display will then receive the 7th, 8th, 9th, and 10th frame that all match the 6th frame so again nothing will change on the screen; this means that the screen would only visibly change once every 24 seconds if a TV were to accept a 24hz/p/fps signal and display every frame 5 times in an attempt to remove film judder.

It has been shown on the Sharp 92u 120hz thread that this is infact NOT what the Sharp is doing, and test for "film judder" will show that motion judder IS still present on these TVs; although I have not seen anyone able to get a straight answer from Sharp that states this.

I have been working with Toshiba Tech Support for several days asking them about their implementation of 120hz refresh rate to see if I can get a straight answer from SOMEONE. Toshiba initially told me that the LX177 series TV uses Motion Intepolation algorithms to determine what changed between two frames and to generate a new interpolated frame between the two images; we all know this; but this explaination woudl only apply if the processor is being fed a 60hz signal already. (LCDs run at a standard refresh rate; they only physically refresh at one rate; if the LX177 panel refreshes at 120hz and the motion interpolation algorithms being run in the TV's video processor is only generating one interpolated frame for every one frame we know for certain that this discusson is talking about taking a 60hz signal as an input and converting it to a 120hz signal for display (a 60hz signal would already have a 3:2 cadence applied and would therefore still have "film judder").

If the same algorithm were run against a 24hz/p input we would end up with a 48hz output; we are again talking about an LCD TV that is only capable of being driven at 120hz so the TV would not be able to display this output. So I Specifically asked the question: "This TV's manual says that it is driven at 120hz and that it can accept a 24p/24hz signal as an input; what does the video processor do with a 24p/24hz input to get it to 120hz?"

The Toshiba Customer Service Rep had to escalate this to an engineer and get back to me and the response that I received was:

"
Regarding your question on how our units process a 1080p, 24Hz signal; Our units take the 1080p 24Hz input, add the 3:2 pulldown and then apply the Clear Frame Technology to achieve 120Hz.

For further assistance, please write back or call our Customer Solutions Department at 1-800-631-3811. They are available Mon-Fri, 8AM to 7PM Central time.

[name removed]
Toshiba Customer Service
"

So, as many have guessed before, it seems quite likely that all of these 120hz sets that accept a 24hz/p input are adding back the 2:3 cadence to get to 60hz and then performing motion interpolation every other frame to get to 120hz. -This will NOT remove "film judder'; film judder is still present although changed slightly.

Without the motion interpolation running our cadence (at 60 hz after 3:2 cadence is applied) was:
Frame 1, Frame 1 (repeat), Frame 2, Frame 2 (repeat), Frame 2 (repeat), Frame 3, Frame 3 (repeat), Frame 4, Frame 4 (repeat), Frame 4 (repeat)

Remeber that an LCD will only change when it is sent new information so on a normal 60hz LCD TV frame #1 would be displayed for 2/60ths of a second, and then frame 2 for 3/60ths of a second, frame 3 for 2/60ths of a second, and frame 4 for 3/60ths of a second, this will create "judder" in motion.

Now let's look at the resulting stream on a 120hz LCD TV when the 24hz material has a 3:2 cadence applied and then motion interpolation is run between each frame to get us to 120hz: (any frames designated with -i are interpolated and their motion is the difference between the proceding and following frame)
(all I'm doing is taking the above stream with a 2:3 cadence and creating an interplated frame between each orignal frame) (Bold text - original frame, regular text=interpolated frame)

Frame 1, Frame 1-i, Frame 1 (repeat), Frame 1-i2, Frame 2, Frame 2-i, Frame 2 (repeat), Frame 2-i2, Frame 2 (repeat), Frame 2-i3, Frame 3, Frame 3-i, Frame 3 (repeat), Frame 3-i2, Frame 4, Frame 4-i, Frame 4 (repeat), Frame 4-i2, Frame (repeat), Frame 4i3, Frame 5

Now this is getting a little crazy, but frame 1-i above would be an interpolated frame that should represent half of the motion between Frame 1 and Frame 1 (repeat); of course the two original frames are identical so he interpolated frame would be identical to the original two frames.
Frame 1-i2 would actually provide us a real new frame that represents half the motion between frame 1(repeat) and Frame 2
Frame 2-i would be ANOTHER repeat of Frame 2 as would Fame 2-i2, Frame 2i-3 would again create a new frame of motion. We'd end up with an actual displayed cadence on an LCD of:

Frame 1 displayed for 3/120ths
Frame 1-i2 (a new frame that ammounts to our new "Frame 2) for 1/120th
Frame 2 displayed for 5/120ths
Frame 2-i3 (a new frame that ammounts to our new "Frame 4) for 1/120th
Frame 3 displayed for 3/120ths (and now we're back to the beginning of the cadence

In essence this would give a a whole new cadence and a whole new type of "film judder"; this could be considered a 3:1:5:1 cadence.....


Adam

[dmzguy]

I also want to add that this does not mean that 120hz displays will not end up providing some real-life motion-related benefits; the point of the above post is that the benefits are not going to come from removing film judder becaue a 24hz/p input is just going to end up having each frame displayed 5 times to come up with 120hz.

People just look at an older or lower-end LCD and see "motion issues" and spend 1 minute looking it up and come to the conclusion that all of LCD's motion problems are a result of high refresh times (12ms-16ms), or that they're a result of "film judder" or "motion judder". There are many individual, separate issues that lead to the perception of "motion issues" on LCDs.

It's kinda like when someone says to their IT group: "My email is slow", the user typically believes that there is one issue behind the behavior that they're seeing and they typically want one answer "the server doesn't have enough ram", or "your mailbox is too big", or "you're connected to an old 10mb hub",but in reality all 3 of these things could be a problem and just fixing one of them may not result in the level of improvement that the user is looking for. -It's the same with motion issues on LCDs, retinal persistence plays a part in LCD's motion issues. (You know when you stare at the american flag for a long time and then look at the wall you can still see the picture of the flag in your "mind's eye" even though you are no longer looking at the image?)

In a 60hz LCD panel (if you have a new frame sent every frame) each frame is displayed 16.7ms (1 second /60 frames in each second) and then the next frame is displayed, if you decrease how long the image is displayed for, you decrease the effect of retinal persistence.)

When you drive an LCD at 120hz (assuming all 120 frames are unique), you end up with a new frame every 8.33ms (1 second/120 frames in each second).


Adam

[chrisherbert]

Will any of the 120hz LCDs actually accept a 120hz signal? Then you could at least use an HTPC to eliminate judder.

[Spur80]

Whoa! it's a very long geeky post indeed, interesting though

I currently have the new 32" Panasonic Viera LCD TC-32LX700 with 120hz motion pro and It's really help with video games for example. I can clearly see it when I use "demo mode" (demo mode, split half of the screen to 120 Hz while the other half is 60 HZ).

It's really help to produce images more "cleans" during moving action.

This unit is 720p only so I can't comment about 1080p/24 , but I believe about what you said.

[dlhoppe]

The solution for this just seems awful easy to me using the 120Hz clock.

They could simply take the 24p source material and display each frame 4 times followed by a "clear" frame. That would be five frames per original frame (24 * 5 = 120). You'd eliminate judder by giving each frame equal time on screen, and you still get the benefit of the "clear frame" technology to eliminate "smearing" from motion. Am I missing something?

[westa6969]

Where does it state that in and of itself that 120Hz and 24fps would get rid of what appears to be motion? Please reference this as a perfect scenario statement as there are other factors at play.

Now common sense math equates that it can indeed provide Smoooother Motion and less apparent judder but the mistake is in assuming it will never be seen. Fact is you won't know unless you have a Trade Show environment where it is compared side by side - I ask how the hell are we going to do that in our homes? So Joe Consumer see's an episode of motion and assumes it's not working when in fact he cannot see the reduction since it will never be observed unless you have a 60Hz next to it. More "120Hz = Less Motion especially when matched with 24fps devices that are on the horizon.

So a novice would look and state "why this is fake BS!" if motion observed - the flaw is in believing that it will never be seen ever or expecting such a thing when strobing lights or fireworks are going on in the screen and the panel must react at lighting speeds. Other factors are at play but for God's sake most of the new panels are plenty fast and this is just another step forward for those that understand the math and understand advances come in steps that have been tremendous in this area through the past few years.

Look, I view a beautiful 57" Sharp and I have never seen motion with an HD DVD or 360 and I see it rarely in normal viewing and so little as I don't give a crap as it does so much else so great! I take the trade-off as it even blows away going to the movies now. HD Broadcasts are still in their infancy and hiccups occur and are we going to blame every little hiccup on the panel instead of the broadcaster?

Quit expecting Nirvana and accept the tradeoff's - 120/24fps is a STEP forward not the solution. The broadcasts are not 120Hz - movies are 24fps and the TV Masters I believe are done for broadcast in 24fps making the 120Hz of the panel a perfect match to = smoother motion but it doesn't mean none will ever be observed.

[dlhoppe]

I guess that all depends on how you define judder. I'm referring to the problem that seems to show most during a slow horizontal pan with unequal "frame time" on the screen (3:2 cadence). As you know this results in a "jerky" motion rather than the alternative. I won't say "smooth" here for conversation sake . I agree that even if you had equal time per frame, there is still "missing" information due to the fact that any digital film, video or broadcast is really a series of "snapshots" shown in quick sequence. The faster the motion, the further the subject of the frame will "jump" when the next frame is shown. That's a different animal from the context here. So if we can at least get equal "frame time" out of these 120Hz sets, that indeed is a step forward. Some might say a BIG step forward.

Originally Posted by westa6969

Where does it state that in and of itself that 120Hz and 24fps would get rid of what appears to be motion? Please reference this as a perfect scenario statement as there are other factors at play.

Now common sense math equates that it can indeed provide Smoooother Motion and less apparent judder but the mistake is in assuming it will never be seen. Fact is you won't know unless you have a Trade Show environment where it is compared side by side - I ask how the hell are we g0oing to do that in our home?

So a novice would look and state "why this is fake BS!" if motion observed - the flaw is in believing that it will never be seen ever or expecting such a thing when strobing lights or fireworks are going on in the screen and the panel must react at lighting speeds. Other factors are at play but for God's sake most of the new panels are plenty fast and this is just another step forward for those that understand the math and understand advances come in steps that have been tremendous in this area through the past few years.

Look, I view a beautiful 57" Sharp and I have never seen motion with an HD DVD or 360 and I see it rarely in normal viewing and so little as I don't give a crap as it does so much else so great! I take the trade-off as it even blows away going to the movies now.

Quit expecting Nirvana and accept the tradeoff's - 120/24fps is a STEP forward not the solution. The broadcasts are not 120Hz - movies are 24fps and the TV Masters I believe are done for broadcast in 24fps making the 120Hz of the panel a perfect match to = smoother motion but it doesn't mean none will ever be observed.

[chrisherbert]

I've heard that when Europeans see film-based sources at 60hz it drives them absolutely crazy. This makes me think that our western eyes have just become accustomed to judder. Who knows, maybe in 20 years when every display works in a multiple of 24hz we'll wonder how we were able to tolerate it for so long.

[dmzguy]

I feel like some sort of freak adding posts to my own thread, but I've had to dig up this information over the past 4 months from all sorts of places and it took a really long time to compile this all and I thought that there's other folks that might be interested in some Objective information on LCD's motion issues.

SOO....


Given the information in the first post what do I think that LCD manufacturers could/SHOULD do to adddress motion issues including "film judder"???


I think the BEST answer is:

1. Make an LCD that can be driven at two different rates, I'll go on a limb and even say: either 60hz and 48hz, or 120hz and 48hz

-When the TV is provided video content at 60hz it is run through a simple motion interpolation algorithm that just creates 1 new frame for every 1 physical frame to get to 60hz (as the 120hz TVs do today), if the original content is recorded at 30hz or 60hz you have no judder anyway

-When the TV is provided content at 24hz or content that was originally at 24hz and has had a 2:3 cadence applied, the user can select whether they would like a 2:3 cadence applied and they would like the contenat displayed at 120hz (it would run at 120hz but would still have "film judder", or the user can select to have the 24hz content run through the same motion interpolation process where every frame is just "doubled" to another frame and we're left with a 48hz signal that does NOT exibit "film judder", and provides "smoother" motion that 24hz would; the 48hz signal is then displyed on the TV at 48hz.

and I'm not going to post option #2 until I complete a quick patent search...


Adam

[dlhoppe]

Adam,

120Hz covers all the bases. 24, 30 and 60 all go into 120 evenly. It's simply just a matter of how you process the frames from the source material. I'm willing to bet it would be a lot cheaper to alter the chip logic for the appropriate frame processing than to make a multi-rate display. Although that's only an assumption on my part.

Originally Posted by dmzguy

I feel like some sort of freak adding posts to my own thread, but I've had to dig up this information over the past 4 months from all sorts of places and it took a really long time to compile this all and I thought that there's other folks that might be interested in some Objective information on LCD's motion issues.

SOO....


Given the information in the first post what do I think that LCD manufacturers could/SHOULD do to adddress motion issues including "film judder"???


I think the BEST answer is:

1. Make an LCD that can be driven at two different rates, I'll go on a limb and even say: either 60hz and 48hz, or 120hz and 48hz

-When the TV is provided video content at 60hz it is run through a simple motion interpolation algorithm that just creates 1 new frame for every 1 physical frame to get to 60hz (as the 120hz TVs do today), if the original content is recorded at 30hz or 60hz you have no judder anyway

-When the TV is provided content at 24hz or content that was originally at 24hz and has had a 2:3 cadence applied, the user can select whether they would like a 2:3 cadence applied and they would like the contenat displayed at 120hz (it would run at 120hz but would still have "film judder", or the user can select to have the 24hz content run through the same motion interpolation process where every frame is just "doubled" to another frame and we're left with a 48hz signal that does NOT exibit "film judder", and provides "smoother" motion that 24hz would; the 48hz signal is then displyed on the TV at 48hz.

and I'm not going to post option #2 until I complete a quick patent search...


Adam

[MUGEN]

why not 180hz
http://www.watch.impress.co.jp/av/do...1/display2.htm

[chrisherbert]

Originally Posted by OreoJoe

Does film look like video on a 120Hz display? 1080p48 with flicker might be more like the cinema, but I have not yet compared a 120Hz display.

I don't think film would look at all like video on a 120hz display, unless it's creating extra frames by interpolation. I'm seen film content on a CRT display at 120hz and it certainly doesn't look like video.

Even at 60hz it's not like LCDs have noticeable "flicker." That's not what 120hz is trying to fix.

I have an old Philips video processor that interpolates extra frames for film content. It's an incredibly strange effect and looks terribe in my opinion.

[My00t8]

Is there really anything wrong with just showing the frame five times without interpolation? Wouldn't that be how the film appeared in theaters? Isn't that how the film's director envisioned it and wanted it conveyed?

Honestly, I don't see the need for interpolation, but I do see the need for a refresh rate that is a perfect multiple of the various source material frame rates the end user can be expected to come across. I hope that, once there IS a 120Hz LCD that actually accepts a 24p signal, the manufacturers will either not do interpolation or at least give the end user the choice of turning it on or off.

[chrisherbert]

Originally Posted by My00t8

Is there really anything wrong with just showing the frame five times without interpolation? Wouldn't that be how the film appeared in theaters? Isn't that how the film's director envisioned it and wanted it conveyed?

Honestly, I don't see the need for interpolation, but I do see the need for a refresh rate that is a perfect multiple of the various source material frame rates the end user can be expected to come across. I hope that, once there IS a 120Hz LCD that actually accepts a 24p signal, the manufacturers will either not do interpolation or at least give the end user the choice of turning it on or off.

I think we'll definitely see 120hz displays that actually display a 24 fps source properly. These are first generation issues.

[irkuck]

Sorry, this thread seems to me aberrated.

Fact: 120 Hz IS NOT, REPEAT NOT, equivalent with ability to properly display 24p. 120 Hz is related to the INTERNAL frame display only, it does not say ANYTHING about the INPUT frame rates accepted. Thus, there are 120 Hz displays which accept only 30p/60i input from sources producing it.

For proper handling of 24p there has to be a source producing it AND display prepared for 24p. New displays coming later this year will have 24p support. 120 Hz will help in this respect.

[dlhoppe]

I don't believe anyone in this thread stated that the 120Hz feature was designed to display 24p material. We're just discussing the fact that it could be tweaked to provide for that functionality.

I think it's pretty well known that the 120Hz feature is there to help eliminate "smearing/blurring" artifacts associated with fast motion.

Originally Posted by irkuck

Sorry, this thread seems to me aberrated.

Fact: 120 Hz IS NOT, REPEAT NOT, equivalent with ability to properly display 24p. 120 Hz is related to the INTERNAL frame display only, it does not say ANYTHING about the INPUT frame rates accepted. Thus, there are 120 Hz displays which accept only 30p/60i input from sources producing it.

For proper handling of 24p there has to be a source producing it AND display prepared for 24p. New displays coming later this year will have 24p support. 120 Hz will help in this respect.

[x2000]

Will 120hz a big improvement on football games at 1080i?

[mike_j_johnson]

So...do we have any TVs today with 5:5 true 24p output to the screen?