PDA

View Full Version : 60Hz LCD displays...yuck!


1tribeca
10-17-2007, 10:01 PM
Went to an interesting seminar at Panasonic today. Among the most interesting demos was the side by side comparison of a 720p plasma and a 1080p LCD (although it was disguised, everyone nailed it for a Sony right away) both panels were 42". They rolled video of a simple still camera shot taken from a roadside location...probably a few feet from the curb, about 4 ft high I'd say. The point of the video was to demonstrate motion across the screen...at a good pace, but not blinding fast...traffic on a busy street. We were advised to pay close attention to the motion detail ie. license plates, rims/hubcaps etc. Keep in mind, the LCD was a 60Hz panel. The difference was astounding!! The blurriness from the aforementioned license plates and rims was brutal on the LCD...unwatchable and completely unacceptable for normal video conditions. The plasma was very good...minimal disruption in the picture.

They then compared the same video on two LCDs...one 60hz the other 120Hz...what a big improvement!!

Their point was, it's not all about 720p...1080p...blah, blah...there is so much more to consider when choosing a TV...for LCD, the 60Hz/120Hz issue is a BIG one!!

If anyone is considering an LCD for their main panel for movies, sports...anything fast moving (not sure at this point why anyone would!! Buy a plasma!!) make sure it's a 120Hz panel!!

I was always a plasma kinda guy, this just solidifies my opinion that much more.

no. 5
10-17-2007, 11:15 PM
LCD displays come in 120Hz too.

But more importantly, what was the frame rate of the video they showed?

I'm fairly certain that the difference you saw was due to processing capability, not panel type.

mtrycrafts
10-18-2007, 12:22 AM
Went to an interesting seminar at Panasonic today. Among the most interesting demos was the side by side comparison of a 720p plasma and a 1080p LCD (although it was disguised, everyone nailed it for a Sony right away) both panels were 42". They rolled video of a simple still camera shot taken from a roadside location...probably a few feet from the curb, about 4 ft high I'd say. The point of the video was to demonstrate motion across the screen...at a good pace, but not blinding fast...traffic on a busy street. We were advised to pay close attention to the motion detail ie. license plates, rims/hubcaps etc. Keep in mind, the LCD was a 60Hz panel. The difference was astounding!! The blurriness from the aforementioned license plates and rims was brutal on the LCD...unwatchable and completely unacceptable for normal video conditions. The plasma was very good...minimal disruption in the picture.

They then compared the same video on two LCDs...one 60hz the other 120Hz...what a big improvement!!

Their point was, it's not all about 720p...1080p...blah, blah...there is so much more to consider when choosing a TV...for LCD, the 60Hz/120Hz issue is a BIG one!!

If anyone is considering an LCD for their main panel for movies, sports...anything fast moving (not sure at this point why anyone would!! Buy a plasma!!) make sure it's a 120Hz panel!!

I was always a plasma kinda guy, this just solidifies my opinion that much more.

As no.5 indicated, depends on what the frame rate was when the picture was recorded. If it was 60Hz, then doubling the same picture should deliver the same picture with no additional picture info. Perhaps the original was recorded with 120Hz frame rate?
Can you find out?

1tribeca
10-18-2007, 07:11 AM
Excellent question...I'll try to find out shortly!!

Pwner_2130
10-18-2007, 12:00 PM
Went to an interesting seminar at Panasonic today. Among the most interesting demos was the side by side comparison of a 720p plasma and a 1080p LCD (although it was disguised, everyone nailed it for a Sony right away) both panels were 42". They rolled video of a simple still camera shot taken from a roadside location...probably a few feet from the curb, about 4 ft high I'd say. The point of the video was to demonstrate motion across the screen...at a good pace, but not blinding fast...traffic on a busy street. We were advised to pay close attention to the motion detail ie. license plates, rims/hubcaps etc. Keep in mind, the LCD was a 60Hz panel. The difference was astounding!! The blurriness from the aforementioned license plates and rims was brutal on the LCD...unwatchable and completely unacceptable for normal video conditions. The plasma was very good...minimal disruption in the picture.

They then compared the same video on two LCDs...one 60hz the other 120Hz...what a big improvement!!

Their point was, it's not all about 720p...1080p...blah, blah...there is so much more to consider when choosing a TV...for LCD, the 60Hz/120Hz issue is a BIG one!!

If anyone is considering an LCD for their main panel for movies, sports...anything fast moving (not sure at this point why anyone would!! Buy a plasma!!) make sure it's a 120Hz panel!!

I was always a plasma kinda guy, this just solidifies my opinion that much more.

I guess individual experiences make for individual opinions, last night at Best Buy, I compared a Sony KDL-S3000 32" Bravia (60Hz refresh) and a Panasonic TC-32LX700 32" with 120 Hz refresh and i could see no difference between picture quality in them, and they were both playing the Blu-ray demo dvd...

mtrycrafts
10-18-2007, 03:55 PM
I guess individual experiences make for individual opinions, last night at Best Buy, I compared a Sony KDL-S3000 32" Bravia (60Hz refresh) and a Panasonic TC-32LX700 32" with 120 Hz refresh and i could see no difference between picture quality in them, and they were both playing the Blu-ray demo dvd...

If that was a bluray demo, made for BD playback or movies, I bet that it is at the standard film rate of 24frames and was just processed with the 3:2 conversion process and doubled for the 120Hz. Why would there be a difference as there are only 24 frames per second of film to present?
Similar to presenting a 480 resolution and making the pixels smaller and smaller. You just end up with more pixels having the same content.

emorphien
10-20-2007, 01:13 AM
I have noticed some horrible blurring on LCDs compared to Plasmas with fast panning motion at BJs but I don't know anything about the actual signal.

What was interesting was the video at BJs was far better for comparison than everything I've seen at BB or CC (random football scenese, other useless stuff). It had all kinds of sequences designed to show how the screens performed with different kinds of signals.

snickelfritz
10-20-2007, 03:00 AM
I first discovered this phenomenon about 5-6 years ago when I started building and tweaking PCs to run games better.

We finally got the minimum framerate up to 125fps in Quake3 and decided that the higher framerate looked choppier and less smooth than lower framerates!
The culprit was the refresh rate and frame rate were not tracking properly. The framerates were actually too high!

Solution:
Setting the /r_displayrefresh to 120hz and capping the in-game framerate at 120fps resulted in sick-to-your-stomach-smoooooth in-game animation.
The catch was that you needed a display that could actually handle a 120hz refresh rate setting at a decent resolution. (many cannot refresh at 120hz above 800x600)
From that day forward, the first spec I look at in a computer display is the highest resolution that it can refresh at 120hz.

I'm more than a little surprised that it's relevant in a 30fps TV or DVD signal, though.
If the signal only carries 30 frames per second, why would the TV need to exceed that number?
At 120fps, each still frame of the video would be theoretically drawn four times. (120/30=4; yeah, I think that's correct)
Hmm... am I missing something here?

MDS
10-20-2007, 03:10 AM
At 120fps, each still frame of the video would be theoretically drawn four times. (120/30=4; yeah, I think that's correct)
Hmm... am I missing something here?

LCD TVs do not draw a 30 fps video 4 times if it is a 120 Hz display. Nor do they perform 3:2 pulldown on a 24 fps (film) to convert it to 60 Hz and then double it. They interpolate the in-between frames with the intent of smoothing motion lag caused by a fast moving object.

snickelfritz
10-20-2007, 03:23 AM
LCD TVs do not draw a 30 fps video 4 times if it is a 120 Hz display. Nor do they perform 3:2 pulldown on a 24 fps (film) to convert it to 60 Hz and then double it. They interpolate the in-between frames with the intent of smoothing motion lag caused by a fast moving object.

If I read you right, you're implying that the improvement is mostly due to advanced video processing and not purely pixel response speed.
ie: making the pixels respond faster is only one aspect of the solution that provides an opportunity to synthesize the illusion of smoother motion using the video processor.

MDS
10-20-2007, 03:37 AM
Yes, the pixel response time is inherent to the panel and only determines how fast a pixel can change state (go from one color to another). When you have a moving object that is changing position, it can be in the wrong place by the time the panel gets around to changing the state of the pixel.

The display tries to determine what the next image 'should' be in the time between when it draws the frame and the time it takes for the panel to actually change the state. By interpolating ('intelligent guessing') it seeks to minimize artifacts caused by drawing the frame at the original time point and what it would be at the next refresh cycle.

Everyone seems to think that 5:5 pulldown is the holy grail (ie simply repeating the frame 5 times for 24 fps image on a 120 Hz refresh cycle) but it is not necessarily so.

mtrycrafts
10-20-2007, 04:04 AM
Yes, the pixel response time is inherent to the panel and only determines how fast a pixel can change state (go from one color to another). When you have a moving object that is changing position, it can be in the wrong place by the time the panel gets around to changing the state of the pixel.

The display tries to determine what the next image 'should' be in the time between when it draws the frame and the time it takes for the panel to actually change the state. By interpolating ('intelligent guessing') it seeks to minimize artifacts caused by drawing the frame at the original time point and what it would be at the next refresh cycle.

Everyone seems to think that 5:5 pulldown is the holy grail (ie simply repeating the frame 5 times for 24 fps image on a 120 Hz refresh cycle) but it is not necessarily so.


Then, where does that 3:2 pulldown come into play? A combination of both calculations?
5:5 pulldown still uses the same pixel locations or are they calculating inbetween frame locations, interpolating? If so, then it is the 5x interpolation that would be displayed and wow, it could really mess things up. Not much time to crunch the numbers.

MDS
10-20-2007, 04:18 AM
As far as I know no display does true 5:5 pulldown to convert 24 fps film to 120 Hz. They do 3:2 pulldown to convert 24 fps to 60 Hz and then interpolate the in-between frames to get to 120 Hz refresh rate.

videobruce
10-20-2007, 12:05 PM
(not sure at this point why anyone would!! Buy a plasma!!) make sure it's a 120Hz panel!!Because they want less of a power hog, no burn, less weight and less if any reflections (to name a few) for starters.

mtrycrafts
10-20-2007, 05:33 PM
As far as I know no display does true 5:5 pulldown to convert 24 fps film to 120 Hz. They do 3:2 pulldown to convert 24 fps to 60 Hz and then interpolate the in-between frames to get to 120 Hz refresh rate.

Great, thanks. So then there is no interpolation involved when 3:2 pulldown is accomplished, just a straight pulldown.
With 24 fps material the only real benefit would be to have a higher original fps to start with for a smooth fast motion sweep, right?

1tribeca
10-22-2007, 05:06 PM
Ok...phew...I'm back! The film used for the demo was 60Hz

no. 5
10-22-2007, 06:43 PM
As far as I know no display does true 5:5 pulldown to convert 24 fps film to 120 Hz. They do 3:2 pulldown to convert 24 fps to 60 Hz and then interpolate the in-between frames to get to 120 Hz refresh rate.

That is very interesting, I had thought sets with a 120Hz refresh rate were using a 5:5 pulldown... ah, what marketing would have us believe. :p

majorloser
10-22-2007, 07:29 PM
Great, thanks. So then there is no interpolation involved when 3:2 pulldown is accomplished, just a straight pulldown.
With 24 fps material the only real benefit would be to have a higher original fps to start with for a smooth fast motion sweep, right?

I was at a ISF seminar a couple weeks ago. It has to do with 120 Hz is evenly divided by 24 fps where as 60 Hz is not. The 60 Hz picture will create some nasty image problems known as "jutter".

I'll see if I can dig up an easy description. :o

EDIT: here's a little something
http://www.hometheatermag.com/gearworks/607gear/

mtrycrafts
10-24-2007, 03:02 AM
I was at a ISF seminar a couple weeks ago. It has to do with 120 Hz is evenly divided by 24 fps where as 60 Hz is not. The 60 Hz picture will create some nasty image problems known as "jutter".

I'll see if I can dig up an easy description. :o

EDIT: here's a little something
http://www.hometheatermag.com/gearworks/607gear/

Thanks for the link. It showed some deinterlacing issues. Is that what jutter looks like too?

BMXTRIX
10-24-2007, 12:41 PM
As far as I know no display does true 5:5 pulldown to convert 24 fps film to 120 Hz. They do 3:2 pulldown to convert 24 fps to 60 Hz and then interpolate the in-between frames to get to 120 Hz refresh rate.
This is not correct as far as I know and really is very unlikely if you are to analize how film actually looks. There are a few VERY new displays that support frame interpolation at 120hz, which is a feature that can be turned off, and has been considered to look poor.

Let me explain:

Original film based material is shot a 24fps (hz) - that's a given.
BUT - and this is the most overlooked, and very important point...
Film is not shot a 1/1000th of a second per frame, or 1/500 of a second - it uses a slower shutter speed, so that if you pull one frame of film that has motion in it, you will see blurring within the image.

If you put two frames next to each other, both of which have motion, the second frame will pick up the blurring where the first frame left off. This means that shutter speeds run very close to 1/24th of a second and you get that much blurring in every frame.

Durtion motion sequences, on video, this helps to make the motion look smooth - otherwise you would end up with motion that appeared jerky, like under a strobe light. Older video games that didn't blur motion often had complaints about this issue and newer games address this... or speed up motion significantly.

But, for video, to perform tween frame interpolation, the original frame must first be deblurred to compensate for the new frame that will be inserted, which also must contain blur. This requires that multiple frames be pulled into a frame buffer, and they must be compared to frames around them to determine the proper way to de-blur all the motion - which may not all be in the same direction.

So, with 1080p, that's an interpolation of 2,000,000 pixels at 24fps to determine tween frames by deblurring existing frames, and properly blurring and inserting new frames into the video path.

The complexity of the math - and the raw horsepower to accomplish such a feat is staggerring. Good photo editing programs have issues doing this well on a single frame - and it often takes a half second or more to accomplish.

So, you begin to see that the more likely way to interpolate tween frames is to try to find a mid-point. The problem is, this will actually make the video look worse since it doesn't truly maintain the motion flow that was in the original filming and may lead to additional blurring of the video playback.

This is why, tweening of frames is a function that can be disabled in the newest displays that can accomplish it. Instead, you do get a 5:5 sequence where 24fps original material is displayed 5 times in a row, then the next frame is brought up. It is the most accurate representation of the original film material and delivers the theatrical experience to the home without the added influence of a bunch of faked digital frames which don't enhance the original look.

Duffinator
10-24-2007, 01:44 PM
I have nothing to add to this thread other than it's an excellent disucssion. Keep it coming. :D

MDS
10-24-2007, 04:02 PM
I have seen no evidence that any TV simply quintuples each frame. I read a response from a Toshiba engineer somewhere that explained why it is not done. They do the 3:2 pulldown to get to 60 Hz and apply a frame interpolation algorithm to calculate the in-between frames for display at 120 Hz. Toshiba calls their algorithm 'Motion Vector Frame Interpolation' and other manufacturers have similar processing with their own trademarked name.

It may require a lot of complex math and a lot of processing power to do motion interpolation 'correctly' and the result may not yet be as good as we would like, but that seems to be the way that all current gen LCD handle 24 fps film.

BMXTRIX
10-24-2007, 04:09 PM
To explain things a bit further (with a picture!)...

Look at the photo below:
http://www.ics.uci.edu/~eppstein/pix/ivy/sluggers2/MotionTrails-m.jpg

This was shot at 1/40th of a second - well within the bounds of motion picture cameras.

Now, think about the two obvious points of motion that are going on.

1 - The bat is moving left to right across the frame in an arc.
2 - The ball is moving downward diagonally.

In normal motion pictures, the next frame would have very similar blurring with the start of the blur beginning right where the old blur left off.

With improper tweening, there would merely be a new frame inserted between the two which would contain half the blur from the first frame and half the blur from the second frame... GREAT! Except, for every frame seen on screen the motion would appear jerky because the ball and bat would actually move halfway BACKWARDS through motion that it has already covered. While not something someone may easily identify, it will definitely be noticable, and may be part of the complaints about the new 120hz displays that perform some sort of tweening process.

The proper way to create a tween frame would be to take the photo above and actually de-blur it by 50%. This would require restoration of the background and proper determination of the motion of the different objects in the frame. That is, the ball going one direction would have to be shown over a period of 1/80th of a second (or less) and the bat would have to be shown, as well, over a period of 1/80th of a second (or less). Then the new frame would need to occur duing the next 1/80th of a second and contain the blurring that was removed from the first frame - along with, perhaps, some white space to account for the shutter closing during the film.

Full, and proper implemenation of 120hz frame tweening from a 24fps (hz) source, would actually require about a 5 fold reduction in blurring and then 4 new frames be created with the proper blur on each and every frame.

...

or - just repeat the same frame five times and call it a day with proper 24hz playback.

MDS
10-24-2007, 04:21 PM
The proper way to create a tween frame would be to take the photo above and actually de-blur it by 50%. This would require restoration of the background and proper determination of the motion of the different objects in the frame.

That is exactly what motion vector interpolation tries to accomplish. Whether that process is better or worse than simply displaying a single frame 5 times, I don't know. I think the argument against simply duping the frame 5 times is due to 'retinal persistence' (google it).

BMXTRIX
10-24-2007, 04:45 PM
...They do the 3:2 pulldown to get to 60 Hz and apply a frame interpolation algorithm to calculate the in-between frames for display at 120 Hz. Toshiba calls their algorithm 'Motion Vector Frame Interpolation' and other manufacturers have similar processing with their own trademarked name.
This is actually a published spec of motion tweening and is similar to the incorrect way of tweening that I described above. It, more or less, introduces judder into the image as it shifts objects backward in time and brings down image quality.

Realistically, I have no idea why a company would do things this way instead of simply repeating a frame 5 times to accurately represent the original film content.

Seriously: Why would a manufacturer intentionally mess up film? They actually seem to work hard to TRY to mess up the film content! I don't mind the concept of tweening, but I would think that no matter what 'cool' feature is added, that a simple click or two of the remote can put it back to the way it is supposed to look. Skip the 3:2 pulldown please.

Buckeyefan 1
10-25-2007, 08:57 PM
Because they want less of a power hog, no burn, less weight and less if any reflections (to name a few) for starters.

My 37" LCD display is excellent for Discovery HD, as well as HD prime time shows. When the noise reduction is turned off, the picture is awesome. If I'm watching football Of faced paced movies/sports) in HD, the pixelation is a bit too obvious when the noise reduction is turned off. There are three settings for noise reduction - low, medium and high. The low and medium settings do a great job with sports, without affecting the sharpness of the picture. On high, the picture becomes too soft. The slight loss in sharpness is a small price to pay for an anti-glare, non burn-in screen.