What Contrast Ratio (within A Single Frame, And Dynamic) Is Useful In A High Definition Display?

[MLXXX]

It is often claimed that a high contrast ratio and/or a low black level is important for a high quality video display. Yet source material often has quite a mediocre contrast ratio, and in dark scenes quite limited 'bit depth' or subtlety in representing the actual intensity level.

A limited static contrast ratio (that is the intensity ratio between the brightest part of the screen to the dimmest part of the screen at any given instant), can be overcome to a degree by reducing the illumination of the screen lamp (LCD flat panel or LCD reflective technologies) or overall drive voltages (Plasma technology) on a 'dynamic' basis, that is if the movie changes to a twilight or night-time scene, the screen can be 'powered down'.

Whilst 'powered down' the screen is unable to display a really bright image, so this approach has its limitations. The screen cannot show both a really bright object and a really dim object at the same time. This is unlike the human eye which can see a contrast ratio of up to about 1,000,000:1 in the same field of view.

Acknowledging that current cameras and displays fall well short of the capacity of human vision, the question arises: 'How good does the contrast of a display need to be to adequately display currently available video sources?'.

A related question for those advanced displays that can vary the overall brightness by a 'powering down' process is: 'How effective are the algorithms that regulate a dynamic variation in overall screen brightness?'.

* * *

The above continues the discussion that began (somewhat off-topic!) yesterday in the thread Hd Ready,true Hd,full Hd....., WHATS THE DIFFERENCE...WHATS THE BEST!.

Here are relevant posts copied from that thread, first one of mine:-

...

A good contrast ratio is more like 20,000:1 and if real blacks are the go, a high quality CRT setup is more like 200,000:1.

...

An article by R. N. Clarke refers to the eye's ability to detect an intensity range -- in one scene -- of 1,000,000:1 as when viewing the night sky with the moon and a magnitude 3 star in the same field of view. See The Dynamic Range of the Eye.

Existing movie cameras (film or digital) cannot capture anything like a 1m:1 ratio in one scene, and need to have the lens aperture and/or exposure time per frame increased, when set up for photographing a dim scene.

Some modern LCD flat panels can vary the brightness of their backlight, in response to picture content. When displaying a twilight scene, the backlight of the display is dimmed. Some projectors can dynamically vary an iris to limit the light from the lamp (e.g. Sony SXRD rear-pro TVs).

It is relevant to mention that more than 8 bits of intensity are needed to smoothly represent an intensity ratio of 1m:1, as the eye can detect more subtly than 256 levels of intensity where a very wide intensity range is involved in one scene. Current video protocols for consumer display may attempt to hold video levels (for each of Red, Green and Blue) in the range 16 to 235, which is only 220 discrete levels if the scene involves black greys and white.

Years ago analogue television involved quite limited contrast ratios. If we attempt to require modern displays to represent a very wide range of intensity in any one scene, we need perhaps 10 bits for intensity gradations. [When we talk about 24-bit colour we refer to 8 bits of intensity for each of Red Green and Blue. Essentially the intensity is 8-bit.]

There is also the question of how to deal with a variation in overall intensity level between different scenes. There can be an efficiency in compressing the difference in overall video level between bright scenes and the dim scenes.

A twilight scene will usually be captured by the movie producer such that the peak intensity level in the final film for distribution is lower than for a midday scene, but not greatly lower. Because of limitations in film or digital camera capture, it is likely the lens aperture (or exposure time) will be at least partially adjusted so as to bring the brightness of the twilight scene up. In this way, the difference in overall light level between a midday and twilight scene can be compressed.

The technology in a modern display device can attempt to undo the compression the producer has introduced. The display can make dim scenes dimmer, and bright scenes brighter. However there can be side-effects. When a film camera pans across a midday scene and reaches shadows, the aperture and exposure time of the camera stay constant. However, the display electronics in a consumer's home display may interpret the shadowy scene taken at midday as being from a twilight scene, and dim the display inappropriately, making the shadows even darker.

As I have mentioned once before in this Forum, it could be useful to develop a video protocol that specifically encodes a brightness level adjustment factor for each frame, rather than the display device electronics relying on an algorithm that examines peak or average brightness levels of the incoming source material and attempts to infer an appropriate brightness adjustment.

Classic films could have these factors added in during the transfer to digital encoding, by a photography expert, so as to give a partial restoration of appropriate brightness levels for different scenes.

A post in reply by Owen:-

Nice write up mate, but you have missed the point.

What I am getting at has nothing to do with the eyes dynamic range or about the limitations of film or video, its all about black level.

Put simply, take a typical LCD with a peak light output of 450cdm2 and a native panel contrast ratio of 2000:1. Black level on this display will be 450/2000=0.225. Now this is a long way from black. If we add in dynamic black lighting, which makes the picture dimmer not brighter so peak brightness remains the same, we can get a contrast ratio of 6000:1 according the manufacturers specs. Black levels works out at 450/6000=0.0083.

This number may seem impressive, but it’s actually still very grey in a dim or dark environment.

Now take as an example my now retired Hitachi CRT RPTV.

I typically ran it with a peak brightness of 75cdm2, which I find plenty bright for normal viewing and still too bright for a dark room.

Now CRT’s have the ability too turn off completely and make absolute black in a totally dark environment, this gives them an essentially infinite contrast ratio, but let’s say 200,000:1. Black level works out at 75/200,000=0.000375 or well below the perception of humans.

To get the same black level out of the LCD would require it to have contrast ratio of 1,200,000:1 due to its extra brightness. 450/1200000=.000375

\

As you can see people the brighter the display the better the contrast ratio needs to be to maintain the same black level.

Manufacturers concentrate on making displays brighter and brighter, as bright displays attract customers like moths to a flame in retail environments, where as black levels are not noticeable in a retial environment and a lot of people (if not most) just don’t care about blacks anyway. Just look at how most people adjust their TV’s and its plain they don’t care.

If you always view in a well lit room, black level is not an issue, but for Home Theatre, great blacks are a must as they add so much to the viewing experience.

The extra brightness available with modern displays is unusable in a Home Theatre environment as the screen will be too bright for comfort, especial if it is a large screen.

I am in the process of optimising my new display for Home Theatre. This involves optimising the inherent contrast ratio of the display and discarding 50% of the displays brightness to gain 50% better blacks. This is only possible to do on projection systems.

On flat panels you have to live with what you get.

I reply to the above post at post #2 below.

And here is another relevant post from the other thread, this one by MikeAusDTV, which drew attention to some interesting points:

The sensors used in the highest quality still cameras digitise the light using a 12 bit linear convertor. This means that it can only capture a brightness ratio of 4096 to 1. For still images this is saved in an 8 bit JPG format, but because the 8 bits are non-linear, the original 4096:1 ratio can be restored.

But this depends on the display medium - photographs on paper have a maximum contrast of 100:1.

So if you have device that claims to display contrasts greater than 4000:1 it's a waste - you will need to find very specialised signal source to feed it with signals that have a brightness ration greater than 4000:1.

The resolution and contrast of video is less than for still pictures because the eye cannot detect as much detail in a rapidly changing image.

[MLXXX]

...

Manufacturers concentrate on making displays brighter and brighter, as bright displays attract customers like moths to a flame in retail environments, where as black levels are not noticeable in a retial environment and a lot of people (if not most) just don’t care about blacks anyway. Just look at how most people adjust their TV’s and its plain they don’t care.

If you always view in a well lit room, black level is not an issue, but for Home Theatre, great blacks are a must as they add so much to the viewing experience.

The extra brightness available with modern displays is unusable in a Home Theatre environment as the screen will be too bright for comfort, especial if it is a large screen.

I am in the process of optimising my new display for Home Theatre. This involves optimising the inherent contrast ratio of the display and discarding 50% of the displays brightness to gain 50% better blacks. This is only possible to do on projection systems.

On flat panels you have to live with what you get.

Thanks Owen for this explanation.

It strikes me though that in order to get to the final result of blacks looking black, the approach of using a very dark room raises the bar of difficulty considerably.

I wonder whether a very similar result is not obtainable simply by using bias lighting; that is the well known technique of a lamp behind the display unit. If the eye sees black on the screen in these circumstances, this seems equivalent to me to seeing black in a dimmer viewing environment.

I'd also mention that if the room is darkened, the human iris will dilate if the video is dark for a long time. This will lead to the need for an even darker black level!

Also for viewers whose eyesight is slightly out of focus, the dilated human iris may cause the poor focus of the human eye to be worse and detail on the dark screen harder to resolve.

I really don't know a lot about using a darkened viewing room. Is it more satisfying watching a screen when the rest of the room is hardly visible, as there is a greater feeling of immersion?

Sorry if my questions appear very basic.

[MLXXX]

And for the record, here's a further relevant post:-

True blacks.

first point here is that I am definitely not trolling!

Here's an exercise. Look at any show on TV, especially filmed outdoor material like, say CSI or Third Watch. Blacks Blacks Blacks. Right?

Now sit in your lounge room and look around you. What blacks do you see? Well, in my case there's my black leather business shoes as I just got home from work. There's the black bezel around my computer monitor, there's my torch in case of power cuts (QLD) - erm, there's a set of dumbells in the corner with black disks on a chrome rod... that's about it.

Sit out in your yard... you probably won't see a black anywhere.

The point I'm making is that apart from coal and a few bird and animal species , the only black you will ever see in real life is some object that has been deliberately coloured black by a human.

The reason we see blacks on TV and Movies is because they are an artifact of a system of representation that really does not convey the richness of visual information that we would see if we were actually there watching the scene.

So endless arguments about 'true' blacks is as useful as arguments about the number of angels you can fit on the head of a pin. Interesting to the enthusiasts but really nothing to do with the real world.

In the case of plasmas and lcds you are really saying that one of them shows an aspect of a CRAP picture better than the other.

OOps MLXXX just noticed your post above, pl feel free to move this post. cheers.

Hi Bribie G,

One of the situations where blacks come up in the movies is looking out of the window of a spacecraft into space and seeing fighter craft attacking each other.

There can be brilliant explosions and the black of deep space called for in the same field of view.

Granted, not your everyday real life situation.

Cheers,

MLXXX

[Scalpel]

And for the record, here's a further relevant post:-

Hi Bribie G,

One of the situations where blacks come up in the movies is with sci-fi shows involving spacecraft and looking out of the window of the spacecraft into space and seeing fighter craft about to attack.

There can be high contrast and deep blacks called for for such scenes.

Granted, not your eveyday real life situation.

Cheers,

MLXXX

BribieG Today, 11:48 PM

The point I'm making is that apart from coal and a few bird and animal species , the only black you will ever see in real life is some object that has been deliberately coloured black by a human.

The reason we see blacks on TV and Movies is because they are an artifact of a system of representation that really does not convey the richness of visual information that we would see if we were actually there watching the scene.

Damm - you beat me to it - was just going to say - how dare you suggest my viewing of Battlestar galactica is an "artifact of a system of representation" sheesh

Next thing you will be suggesting that Battlestar Galactica isnt based on a real life story

Seriously ......

Blacks are far more common than suggested above - ie where there is a (relative) absence of light:

Night

Caves

etc etc

For some of us tv viewers who love science fiction where such conditions abound (Cave monsters "The descent" "Lord of the Rings", Space "Star Wars" "Battlestar Galactica" "Star Trek", .... need I go on) black levels are quite important for a realistic viewing experience. Seeing washed out greys is just annoying.

J.

[Anton-P.]

Since it's reposted here I'll do the same too

Arguments about blacks are not entirely about getting true black, also about how much white leaks into it. The same white will also leak into other colours making them washed out.

[MLXXX]

Since it's reposted here I'll do the same too

Arguments about blacks are not entirely about getting true black, also about how much white leaks into it. The same white will also leak into other colours making them washed out.

There appear to be two quite different viewing environments to consider:

1. A home theatre with very dim lighting.

• Black level of the display has to be lower than Plasma and LCD technology can currently provide
  
• A persistent low level of leakage of white can wash out the colours in dimmer parts of the picture
  

2. A living room with moderately dim lighting.

• Black levels of some Plasma and LCD displays are sufficiently low for the screen regions displaying 'black' to appear black to the viewer, under the room lighting conditions. And image content just above 'black' can be viewed satisfactorily
  
• Colours may appear more saturated, as a persistent low level of leakage of white light will have less effect because the screen is being driven to more intense levels when displaying picture content
  

I guess that I am more in camp 2 than camp 1. I find my display [a Sony 60" SXRD] quite immersive even when the rest of the room is at a moderately dim level at night.

[BribieG]

Apart from the awful plasmas you see in pubs and clubs I have never actually had a good look at a HD Plasma - I need to go to WOW or JB to get a prezzie for someone in the next week or so and might just have a really good look at the TVs on display to see what the fuss is about. At WOW (QLD franchise) they have really nice viewing lounge set up.

Popped into JB for a DVD last week and the plasmas on display there didn't look particularly good either in colours or blacks but JB have their places lit up like 7 - 11 stores.

[Owen]

I find that I need quite a bit of light in the room (enough to read by) for me not to be able to see that blacks are not black on my 70” SXRD, even after adjusting the iris in the service menu to improve black level 30% below standard and completely blacking out the interior of the cabinet.

With the Window Logo screen saver displayed (black screen with logo) it is clear that the blacks are not black, especially when I compare it to my Hitachi CRT RPTV which is running in parallel of the HTPC in the same room.

When I look at the Hitachi, the Windows logo appears to be floating in a black hole and looks almost like a hologram, no way the Sony can do that.

Now when you consider that my modified SXRD has a much lower black level then any other current digital display, it goes to show how much digital display technology has got to improve to approach CRT black performance.

I have one ace up my sleave and that is a neutral density filter.

I was hopping it would not be needed, but it is clear to me now that it will be.

The camera shop that I ordered the filter from rang me on Friday to say that the Oz distributor had no stock and I will have to wait another month or more. Not happy Jan.

Anyway, the ND2 filter will cut light output in half and consequentially lower the black level equally. This still will not get me the same blacks as the Hitachi, but in anything other then a dark room it should be close enough not to annoy.

The fully tweaked SXRD will have a black level 70% lower then standard but will also have its peak light output reduced by 50%, making it unsuitable for bright rooms, even with the lamp in normal (bright) mode. My viewing environment is never bright so that’s not a problem.

For the SXRD to maintain its original brightness while at the same time maintaining the black level of the modified set would require a contrast ratio of about 30,000:1.

Even the much touted next generation Pioneer Plasma’s will not be in that territory, and it makes displays with 6000:1 contrast ratio’s woefully inadequate.

SED technology was the great hope for people who appreciate blacks, but it seems that SED is a long way from production, if ever.

P.S. As an interesting aside, while in the process of moding my SXRD I installed a shield for the light engine. My first attempt blocked the light path to the bottom corners of the screen. This made it obvious that blacks where not black as a black projected image could clearly be seen as lighter then the shielded black lower corners.

[drsmith]

Owen,

Other than the service menu adjustment to the minimum iris setting which you have all ready done, will you be performing any other modifications to your SXRD prior to receipt of the ND filter ?

[Owen]

Owen,

Other than the service menu adjustment to the minimum iris setting which you have all ready done, will you be performing any other modifications to your SXRD prior to receipt of the ND filter ?

As I said above, I have already fully blacked out the interior of the SXRD cabinet.

This does not help black levels on a fully or mostly black screen but does help a little with ANSI contrast (black levels when bright objects are on screen).

It’s hard to quantify how much improvement has been made by this mod, but its done now so I know I will have the best performance possible.

[drsmith]

Have you performed any other modifications and if so, what has been their impact ?

[Owen]

The only remaining mod is the ND filter.

[MLXXX]

As no-one has provided a quantitative answer to the question raised in the topic title for this thread, I will provide a simple qualitative answer to the first part of the question. Current LCD flatscreen panels provide insufficient contrast within a single frame.

I base that on simple observation - the picture of typical LCD flatscreens has noticeably insufficient contrast at the dark end of the intensity range when displaying a bright scene. Even a bright scene can include rather dark picture content, but such content tends to look a lightish grey, rather than a dark grey approaching black.

The Panasonic Viera TH-65PV600A 65 inch 1920x1080 pixel Plasma TV is specified as having a 5000:1 contrast ratio. To my eyes, just viewing the display in a showroom, fed an HD movie, its contrast performance was very good, noticeably better than the other, non-projection, displays. The performance seemed to be in the ballpark of what is needed.

However the 5000:1 spec is reportedly a dynamic ratio, which raises some doubt as to whether the display would have sufficient contrast to display frames that contain both very bright and very dark content.

____________________________

Have only just noticed a reference to a new LCD flatscreen technology that replaces the backlight with LEDs, thus making localised dimming of particular areas of the display possible. Such technology reportedly can provide high contrast within a single frame - an exciting possibility. Due to be released overseas in July this year. If you haven't already read about this technology, see the following thread: Samsung 81 Series Lcd (led Backlight), contrast ratio -> 50,000:1.

[alanh]

All,

What you are proposing is similar to the way optical sound was recorded in the old days prior to laser optical sound readers in cinema projectors. The optical recorders would us a shutter to cut off the light used to create the sound track when there was silence. This was to minimise the noise from the old PE cell systems.

Since the DTV system does not have a control signal for average luminance, the decoder will have to guess at one. Where this can be a problem is where there is large areas of the screen is flashing for example in Science Fiction movies, flashing police lights etc. MPEG compression systems minimise the data transmitted data from frame to frame, so there is an averaging effect. So if the decoder is measuring the average picture level to determine the back lighting the response time is critical for the previous example.

Gamma Graphs

The backlight system should stay at illuminant D (6500 K) unless the whole screen is one colour. Otherwise the small areas of colour will change hue.

AlanH

[Owen]

Have only just noticed a reference to a new LCD flatscreen technology that replaces the backlight with LEDs, thus making localised dimming of particular areas of the display possible. Such technology reportedly can provide high contrast within a single frame - an exciting possibility. Due to be released overseas in July this year. If you haven't already read about this technology, see the following thread: Samsung 81 Series Lcd (led Backlight), contrast ratio -> 50,000:1.

This system is not new, its been done before in professional grade LCD monitors, at a price or around $35,000 from memory, and it apparently works quite well.

It will be interesting to see how well Samsung can implement such a system. Maybe good, maybe not.

LCD needs this technology to stay competitive.

[Owen]

All,

What you are proposing is similar to the way optical sound was recorded in the old days prior to laser optical sound readers in cinema projectors. The optical recorders would us a shutter to cut off the light used to create the sound track when there was silence. This was to minimise the noise from the old PE cell systems.

Since the DTV system does not have a control signal for average luminance, the decoder will have to guess at one. Where this can be a problem is where there is large areas of the screen is flashing for example in Science Fiction movies, flashing police lights etc. MPEG compression systems minimise the data transmitted data from frame to frame, so there is an averaging effect. So if the decoder is measuring the average picture level to determine the back lighting the response time is critical for the previous example.

Gamma Graphs

The backlight system should stay at illuminant D (6500 K) unless the whole screen is one colour. Otherwise the small areas of colour will change hue.

AlanH

Only LCD panels use a dynamic backlight.

Projection systems use a fixed backlight and a dynamic iris so the light source stays at a constant colour temperature.

[MLXXX]

Re the Samsungs to be released later this year

I presume the backlighting LEDs will operate very fast, and will be capable of changing from fully on to fully off from one frame to the next, or to any required intermediate brightness.

I presume the light level averaging would be calculated very fast, from frame to frame, for each 'illumination zone'.

I presume there would be a great number of independent 'illumination zones' (judging from the quality of the sample pictures!).

One artefact that could be expected if the illumination zones are coarse, is that a video source frame containing a bright line against a black background would display as a bright line, surrounded by a halo of grey, all on a black background. The halo would correspond to illumination zones that included part of the bright white line. It is inescapable that the native contrast ratio of each pixel in a multi-pixel illumination zone will limit the ratio of contrast that illumination zone can display in any given frame.

To get around that limitation there could be a separate illumination pixel for each display pixel. That would be a bit like a plasma panel serving as the backlight to an LCD display! A very expensive proposition, if it were practical at all.

Have just seen Owen's post. I do hope these Samsung panels will cost less than $35,000:-

This system is not new, its been done before in professional grade LCD monitors, at a price or around $35,000 from memory, and it apparently works quite well.

It will be interesting to see how well Samsung can implement such a system. Maybe good, maybe not.

LCD needs this technology to stay competitive.

[Owen]

As no-one has provided a quantitative answer to the question raised in the topic title for this thread, I will provide a simple qualitative answer to the first part of the question. Current LCD flatscreen panels provide insufficient contrast within a single frame.

I base that on simple observation - the picture of typical LCD flatscreens has noticeably insufficient contrast at the dark end of the intensity range when displaying a bright scene. Even a bright scene can include rather dark picture content, but such content tends to look a lightish grey, rather than a dark grey approaching black.

The Panasonic Viera TH-65PV600A 65 inch 1920x1080 pixel Plasma TV is specified as having a 5000:1 contrast ratio. To my eyes, just viewing the display in a showroom, fed an HD movie, its contrast performance was very good, noticeably better than the other, non-projection, displays. The performance seemed to be in the ballpark of what is needed.

However the 5000:1 spec is reportedly a dynamic ratio, which raises some doubt as to whether the display would have sufficient contrast to display frames that contain both very bright and very dark content.

Plasma’s can’t sustain full brightness when most of the screen area is brightly illuminated. Full contrast ratio or dynamic range is only attainable with relatively small areas of the screen illuminated.

The drop of in peak output with a full white screen is in the order of 50%.

Cheaper CRT’s also suffer a similar dynamic compression.

A correctly calibrated Plasma with it’s contrast set so as to avoid white clipping or gamma curve irregularities also has significantly less light output and contrast ratio then the specifications suggest. That’s why contrast ratio numbers are almost always meaningless. They are only ever achieved under unrealistic or specific circumstances.

LCD’s don’t suffer dynamic compression but have poor native contrast ratio’s in stead.

They also loose peak light output when correctly calibrated to avoid clipping or gamma curve distortion.

LCoS and SXRD has similar characteristics to LCD, but has up to 5 times the native contrast ratio and can also take advantage of a dynamic iris.