AVS Forum banner

Setting Contrast - why above white does not clip

30K views 164 replies 26 participants last post by  Stereodude 
#1 ·
Calibrators:


The #1 most asked question by new users trying to set their contrast on non-CRT displays when using the GetGray Digital Video Calibration DVD (or any) is this:


"When adjusting my [insert display model] I am unable to make the above white bars disappear..."


Of course this issue will apply to all calibration tools (DVE, Spears, AVS HD709, etc., signal generators, etc.), so I thought a separate thread might be appropriate. If it's good enough, it may qualify as a sticky in the calibration forum.


Could those of you who are expereinced calibrators comment on the whys and how-to's of setting the correct contrast point on these fixed pixel (non CRT) devices that do not clip above white when adjusting contrast? There have been a lot of threads on the subject, so maybe this thread can be populated as a reference.


I'm confident your answers would be better than mine.


Thanks!
 
See less See more
#2 ·
Here are some related quotes from other threads. It appears Michael is the most prolific in answering. There are other issues to be foudn from display model to display model so a search or your model number in the forum might help.


Quote:
Originally Posted by Michael TLV /forum/post/9625374


Greetings


1. Look for clipping ...

2. Look for discoloration ... (another form of clipping)

3. Eye fatigue factor. If the best contrast ratios still give you a headache... then turn the contrast down more. No sense having the best picture performance (technically speaking) if you get sick watching it.


Regards


Quote:
Originally Posted by CT_Wiebe /forum/post/9626649


You look at the grayscale ramps. Any tinting is an indication of incorrect color temperature (blue tinting is too cool, red tinting is too warm. This is hard to see, because your eye/brain will accomodate to the color shift very quickly. You can also see it in flesh tones (not on video games).


Many displays (especially flat panel LCD and Plasma units) will not allow you to set the Contrast (white level) controls correctly (the WTW bars cannot be made to disappear). In those cases, you adjust the Contrast so that the brightest parts of the picture are at a comfortable level and you do not get any colot tinting of the whites (the red color usually, but not always, runs out of gas first, which causes the whites to look bluish = "cool"). Most sets come set way too bright (illumination, not to be confused with the Brightness, black level, control).


Note: Usually TV and video games are designed to run on displays that are set cooler than is correct for movies. I would guess that normal or warm1 are probably closer to D65 for most displays (at least that's what seems to be reported in most reviews).
http://www.avsforum.com/avs-vb/showt...light=contrast says:
Quote:
Originally Posted by Q: High contrast setting for Samsung DLP? /forum/post/0


I'm having an extremely hard time adjusting the settings (contrast, brightness, color, sharpness) for my Samsung DLP HLN5065W. Any tips on how to use the Avia disc? I get no blooming at all when I try to set contrast, and I find it very hard to properly set contrast. I had read somewhere on the Web that Samsung DLPs need high contrast settings (80% or higher). Is this true? Any tips on using Avia? Are other discs better for setting contrast? Also is it better to go with "warm 1" or "warm 2?" DVIe on or off? Please help! Thanks.

Quote:
Originally Posted by Michael TLV /forum/post/9762932


Greetings


Blooming is for CRT sets. Not DLP.


Adjust contrast up until you get white crushing / clipping ... (use a grayscale ramp pattern)


Or just plain discoloration ... whites turn pink ... then back off.


DNIE is a bad thing on these sets ... leave it off.


Brightness pattern is straight forward enough.


Regards

Quote:
Originally Posted by Michael TLV /forum/post/13072727


Greetings


If you do not crush/clip ... then move to the next part of the contrast equation ... discoloration.


If the near whites still look white to you ... then say the second parameter is now met.


Now onto #3 ... eye fatigue factors. If you can maintain 1 & 2 while at max ... and you do not find the image to be fatiguing to look at ... then


Yes ... there is no disadvantage to setting it to max contrast.



I have to doubt that all 3 conditions can be met on this Sony set. I have the 46V version and the calibrated state has the contrast much lower than 100.


Regards
 
#4 ·

Quote:
Originally Posted by GetGray /forum/post/16280182


Calibrators:


The #1 most asked question by new users trying to set their contrast on non-CRT displays when using the GetGray Digital Video Calibration DVD (or any) is this:


"When adjusting my [insert display model] I am unable to make the above white bars disappear..."


Of course this issue will apply to all calibration tools (DVE, Spears, AVS HD709, etc., signal generators, etc.), so I thought a separate thread might be appropriate. If it's good enough, it may qualify as a sticky in the calibration forum.


Could those of you who are expereinced calibrators comment on the whys and how-to's of setting the correct contrast point on these fixed pixel (non CRT) devices that do not clip above white when adjusting contrast? There have been a lot of threads on the subject, so maybe this thread can be populated as a reference.


I'm confident your answers would be better than mine.


Thanks!

The main problem with the described methodology is that it does not tell you the RIGHT contrast setting, it tells you the highest possible Contrast setting you should use on a given video display (assume the display does eventually crush the above white step(s)). That's useful to know when you are setting up a "Day" mode for viewing with some light in the room (how bright can the picture get before losing information). But when you are setting up for viewing in a dark room, especially with panel displays, you will want a Contrast/Peak White setting much lower than the max setting. For a projector setup, you MAY want the highest possible light output that doesn't lose above-white steps... projectors are often operating in the 8-16 fL range depending on the projector, lens, lamp and screen being used so anything that helps a projector get more light is usually a good thing. But panel displays, even plasmas, have an abundance of light for dark-room viewing.


THX recommends 30 fL for panel displays in a dark room. I find a range of 30 to 35 fL produces excellent results in a dark room. To get 30-35 fL, I display a 100% White window pattern and measure the white level with a meter. A lot of people don't have a meter and can't identify 30 fL or 35 fL. So there's no easy way to know you have the best contrast setting for a dark room unless you have a meter or unless you have someone with a meter measure a display like yours and they share what contrast (and other settings if applicable) produce 30 or 35 fL.


Some LCDs can hit 150 fL or more without losing any above-white steps. You CERTAINLY don't want to have that much light in a dark room... unless you really enjoy bad headaches.
 
#5 ·
If one does not have a light meter to measure contrast, I have personally found that the Monster/ISF HDTV Calibration DVD very good at setting both brightness and contrast. I know that Michael TLV has mentioned it before. You can see the immediate effect of contrast on the white shirt the person is wearing.


As long as you can mute the narration by Jenna....
 
#6 ·
Is there an inexpensive light meter that is popular? Amazon has a $70 Extech 401027-Pocket Sized-foot Candle Light Meter and an Sekonic L-208 Light Meter (Black/White) for $95. I've got a feeling those are only for photography. I wouldn't know what to look for?


Also, does the room have to be pitch black to get a good reading?
 
#8 ·

Quote:
Originally Posted by cage22 /forum/post/16287843


Also, does the room have to be pitch black to get a good reading?

If you are trying to get a peak white reading... how dark the room needs to be to get an accurate reading depends on the type of meter you are using and how far it is from the display... and on how much light there is. Puck-type meters may block some ambient light, but there are limits. Even with those, the room should be fairly dark.


When trying to get an accurate black reading... the room really does need to be very dark - and the meter needs to be pretty good. Some don't read as low as you'd like. And even some expensive meters aren't too great at reading black level.
 
#10 ·
I'm not a professional calibrator, but I'll put my 2c in since this is a subject that interests me as well.


As others have already mentioned, the white clip, needle pulse, and blooming tests are not really designed to achieve the optimum white levels for viewing. They are designed merely to test the display's limits in this area. The optimum setting for white level depends primarily on 3 things:

1. The ambient/background light levels in the room.

2. The display's gamma.

3. The brightness of the content being displayed.
This is why eyeballing really is the best method. And simply assigning an arbitrary value such as 30 or 40 fL may or may not deliver the results you want.


If you don't trust your eyes to make the call unassisted, then the next best thing is probably an ambient light reference. This will take into account the first 2 items (room lighting & display gamma), but some adjustment will probably still be necessary to compensate for the different brightness of content being displayed.


Several of the calibration discs include some type of ambient light reference in their video setup patterns. The SD Component video edition of DVE for example includes one at Title 12/Chapter 16. The idea is to insure that the room lighting does not excede the brightness of the gray swatch in this pattern. It does not really give a specific target for optimum room lighting though.


I'm not really sure what the industry standard is for ambient/background room light. All I've read is that SMPTE recommends no brighter than 10% of the display's peak white level (probably measured from a 100% white window). I'm guessing that this "10%" value is in cd/m^2 or nits, because the max. ambient light pattern on the SD edition of DVE is actually about a 35% gray. In my own tests, I found the best level for background lighting for my display (34" CRT) is probably about half that, or roughly the equivalent of a 15% to 20% gray on the screen. (YMMV of course.)


Instead of adjusting the room light though, you can simply tweak the white level on the display until a flat field gray pattern in that brightness range closely matches the background lighting in your viewing area. And that should put you pretty close to the ballpark for comfortable viewing, with no light meter required.


There will probably be some variation from display to display, so if the picture is still too bright and fatiguing to the eyes with a 15%-20% ambient/background reference, then try adjusting the white level with a slightly brighter ambient gray reference more in the 20%-25% range. Or if you're having trouble making out shadow detail (with gamma and black level set correctly on the display), then you might possibly want to try adusting the white level with a slightly darker ambient gray reference. I would not go too far below a 15% gray though because the peak whites on the display may begin to appear quite bright (unless the display has some type of contrast limiter that automatically clamps down on high APL images).


You also want to try to stay within the white level limits defined by the other test patterns. If, for example, the ambient light reference calls for a higher white level setting than the other peak white level tests recommend, then you may need to reduce your ambient/background room lighting a bit to help bring the peak whites down to a level that's more manageable for your display. This might occur with some direct-view CRTs.


With LCDs you could run into the opposite problem, and might need a little more ambient/background light to get more contrast out of the display, and some more depth in the blacks. Plasma displays could go either way, depending on how good the blacks and contrast are on a given model.


Anyway that's my layperson's suggestion for folks still having some difficulty finding an appopriate white level for their direct-view displays, with no special equipment needed. Hopefully some of it makes sense.
 
#11 ·
A couple addenda to the above...


On CRT and plasma displays, the white level is adjusted using the PICTURE* control.


On LCDs, the white level is adjusted using the BACKLIGHT control on the display. And the PICTURE* control is used to set the white clip point, which should usually be adjusted just below the threshold where white details start to become crushed or shifted in color towards the native temperature of the LCD's backlight.


Generally speaking, you would set the white clip point first with the LCD's Picture control. Then use the Backlight control to adjust the white level of the LCD to an appropriate level for your room lighting (using either an ambient light reference as described above, or just your eyes).


Plasma displays may begin to crush, or more likely shift the color of white details above a certain Picture setting as well. So you'll want to try to the keep the Picture control on plasma displays below that threshold too. Plasmas have no backlight though, so the Picture control is pretty much the only basic user adjustment you get for both white level and white clipping on those displays.


[*Sometimes labeled CONTRAST.]
 
#12 ·

Quote:
Originally Posted by ADU /forum/post/16409723


A couple addenda to the above...


On CRT and plasma displays, the white level is adjusted using the PICTURE control.


On LCDs, the white level is adjusted using the BACKLIGHT control on the display. And the PICTURE control is used to set the white clip point, which should usually be adjusted just below the threshold where white details start to become crunched or shifted in color towards the native temperature of the LCD's backlight.


Generally speaking, you would set the white clip point first with the LCD's Picture control. Then use the Backlight control to adjust the white level of the LCD to an appropriate level for your room lighting (using either an ambient light reference as described above, or just your eyes).


Plasma displays may begin to crunch, or more likely shift the color of white details above a certain Picture setting as well. So you'll want to try to the keep the Picture control on plasma displays below that threshold too. Plasmas have no backlight though, so the Picture control is pretty much the only basic user adjustment you get for both white level and white clipping on those displays.


Increasing numbers of LCD panels are appearing with absolutely NO problems setting the Picture control to any value within it's adjustment range... there is no clipping or crushing anywhere.


This can make it difficult to determine where to set the Backlight and Picture/Contrast controls.


But... setting backlight higher, harms black level (makes black lighter). So if the panel has no problems with the Picture/Contrast control range and setting backlight to the lowest available setting doesn't cause any linearity or off-color issues, setting Backlight to the lowest setting will produce the best black level. However, you may or may not be able to achieve the desired peak white level with the Backlight set to the lowest possible setting (varies from brand to brand and model to model).


So... when the Picture/Contrast control has no measured "issues" when set to "max", set it to max then set the backlight to get the peak white level you want (say 30 fL for a dark room). This will produce the best (darkest) blacks and prevent eyestrain in a dark room. If more luminance is needed for a room with some light, increase the backlight setting to move the peak white level past 30 fL and turn the backlight back down for dark-room viewing.


It is too simplistic to say there is just 1 way to adjust an LCD panel... too much depends on the age, brand, and model to make any blanket statements.
 
#13 ·

Quote:
Originally Posted by ADU /forum/post/16409723


A couple addenda to the above...


On CRT and plasma displays, the white level is adjusted using the PICTURE control.


On LCDs, the white level is adjusted using the BACKLIGHT control on the display. And the PICTURE control is used to set the white clip point, which should usually be adjusted just below the threshold where white details start to become crunched or shifted in color towards the native temperature of the LCD's backlight.


Generally speaking, you would set the white clip point first with the LCD's Picture control. Then use the Backlight control to adjust the white level of the LCD to an appropriate level for your room lighting (using either an ambient light reference as described above, or just your eyes).


Plasma displays may begin to crunch, or more likely shift the color of white details above a certain Picture setting as well. So you'll want to try to the keep the Picture control on plasma displays below that threshold too. Plasmas have no backlight though, so the Picture control is pretty much the only basic user adjustment you get for both white level and white clipping on those displays.


Increasing numbers of LCD panels are appearing with absolutely NO problems setting the Picture control to any value within it's adjustment range... there is no clipping or crushing anywhere.


This can make it difficult to determine where to set the Backlight and Picture/Contrast controls.


But... setting backlight higher, harms black level (makes black lighter). So if the panel has no problems with the Picture/Contrast control range and setting backlight to the lowest available setting doesn't cause any linearity or off-color issues, setting Backlight to the lowest setting will produce the best black level. However, you may or may not be able to achieve the desired peak white level with the Backlight set to the lowest possible setting (varies from brand to brand and model to model).


So... when the Picture/Contrast control has no measured "issues" when set to "max", set it to max then set the backlight to get the peak white level you want (say 30 fL for a dark room). This will produce the best (darkest) blacks and prevent eyestrain in a dark room. If more luminance is needed for a room with some light, increase the backlight setting to move the peak white level past 30 fL and turn the backlight back down for dark-room viewing.


It is too simplistic to say there is just 1 way to adjust an LCD panel... too much depends on the age, brand, and model to make any blanket statements. And there is another issue also... some backlight ballasts make noise when they are set very low... so the strategy of using the lowest possible setting can be problematic if the backlight ballast hums audibly when set very low. Lots of issues have to be accounted for.
 
#14 ·
Speaking of whites...


I'm doing a calibration best as I can until my calibrator arrives..


In setting the contrast, it seems I am clipping whites.. Can only see up to bar #246 on the Spears and Muncil Blu Ray test disc.. Blacks are fine. SO, I seem to be clipping whites..


Is this something that requires service menu and / or a calibrator's expertise? No matter how high / low I move my contrast (and brightness), I can't see all bars... Now on the 11 pt scale, all is good. This problem test pattern has at least 20 white gradations and I cannot see them all - I can see 14/20... is this nomal or should I see all?


Thanks for the help / explanation.
 
#15 ·

Quote:
Is this something that requires service menu and / or a calibrator's expertise? No matter how high / low I move my contrast (and brightness), I can't see all bars... Now on the 11 pt scale, all is good. This problem test pattern has at least 20 white gradations and I cannot see them all - I can see 14/20... is this nomal or should I see all?

the above could easily be caused by the DVD player clipping the signal prior to it being sent to the display. Look for a contrast setting on your player and turn it down 1-3 clicks.


On the same note, I am finding that most BD players clip white when using their factory default settings.
 
#16 ·

Quote:
Originally Posted by thebland /forum/post/16478171


In setting the contrast, it seems I am clipping whites.. Can only see up to bar #246 on the Spears and Muncil Blu Ray test disc...

I'm still BD-less, so don't really know how the Spears/Muncil disc works ( these folks should ), but if #246 refers to the color level, then that's probably in the whiter-than-white range, which is less crucial than colors in the 16-235 range. I don't think anyone wants this to become another debate on the importance of head and foot room being displayed, but since it's relevant to the discussion, fwiw here's my layperson's take on the subject...


By definition, in the Y'CbCr video color space, Y'=16 is black, and Y'=235 is white. Colors below Y'=16 are considered "blacker-than-black" (BTB), and colors above Y'=235 are considered "whiter-than-white" (WTW).


On most fixed pixel displays which have less than perfect black levels, it's usually desirable to clip all the BTB color info off by setting the black clip point on the display as close to the Y'=16 color level as possible, to achieve the deepest looking blacks possible. (Crushing a small amount of shadow detail above Y'=16 may even be worth considering if the black level is particularly poor on the display, though I'd recommend trying to increase the background/ambient light around the display first, if that's feasible.) CRTs work a bit differently, but the basic principal is the same. And the black level should also be adjusted so BTB colors below Y'=16 are not visible, and shadow detail slightly above Y'=16 is just barely visible using some type of PLUGE pattern. Direct-view/single-tube CRTs users should make this adjustment with the normal room lighting used for viewing turned ON.


As far as WTW is concerned, it sort of depends how much contrast there is to spare on the display. If, for example, you've got a rather dim front projector, then you'll probably want to clip as much WTW info in the 236-254 range off as you can to get the brightest whites and highest contrast ratio possible from the display, without sacrificing the vital color info in the 16-235 range.


If you have a relatively bright display (e.g. an LCD or plasma) in a darker room, which seems to have plenty of contrast to spare, then there may be little or no benefit to clipping the WTW info above Y'=235 off. If you're in this situation though, with an LCD, then the first thing you should probably do is try to adjust the BACKLIGHT setting on the display lower. If the Backlight setting is already at or near its darkest level, then try to make the background/ambient room lighting around the display a bit brighter, if possible. (This is what I meant earlier when I said: "With LCDs you could run into the opposite problem, and might need a little more ambient/background light to get more contrast out of the display, and some more depth in the blacks.") The latter would also apply to an overly bright plasma display.


If the Backlight setting (on LCD) is as dark as it goes, and the background/room lighting is as bright as you care to make it, and you still have excess contrast available on the LCD/plasma display, then you'll want to adjust the Picture/Contrast level lower on the display anyway to a point that's more comfortable for your eyes... which may also allow some WTW info to be seen (if there is any).
 
#17 ·

Quote:
Originally Posted by thebland /forum/post/16478171


Speaking of whites...


I'm doing a calibration best as I can until my calibrator arrives..


In setting the contrast, it seems I am clipping whites.. Can only see up to bar #246 on the Spears and Muncil Blu Ray test disc.. Blacks are fine. SO, I seem to be clipping whites..


Is this something that requires service menu and / or a calibrator's expertise? No matter how high / low I move my contrast (and brightness), I can't see all bars... Now on the 11 pt scale, all is good. This problem test pattern has at least 20 white gradations and I cannot see them all - I can see 14/20... is this nomal or should I see all?


Thanks for the help / explanation.

If the Blu-ray player is connected directly to the TV, there can be only 2 sources of the "problem"...


1) Your TV simply won't discriminate between the digital steps 246-255

2) As Gregg mentioned, there's a setting in the Blu-ray player that's changing 247-255 to 246 so they all appear at the TV as the same digital level.


It's no tragedy if the TV won't display discrete steps for 247-255. You'll probably never see anything that would be noticeable. It's more of a psychological problem than anything else. The only time it really matters is if you are displaying video or images from a PC that is in 0-255 mode. And if your TV supports PC 0-255 mode you'll probably find all the steps are visible anyway. Consumer discs and TV are not supposed to contain any data below16 or above 235 anyway. If there's any data out there, it's bogus info. Your TV will be setup to not show any change in black level if it receives data below 16 - so if there happens to be blacker-than-black bit in there somewhere, it will just show up as black. If something similar happens on the top end, there's really no issue since video is supposed to "stop" at 235 anyway.


There are some very impressive video displays that display 237-255 as if all those values were 235 and nobody ever complains about "missing something" on those displays.
 
#18 ·

Quote:
Originally Posted by Doug Blackburn /forum/post/16485119


If the Blu-ray player is connected directly to the TV, there can be only 2 sources of the "problem"...


1) Your TV simply won't discriminate between the digital steps 246-255

2) As Gregg mentioned, there's a setting in the Blu-ray player that's changing 247-255 to 246 so they all appear at the TV as the same digital level.


It's no tragedy if the TV won't display discrete steps for 247-255. You'll probably never see anything that would be noticeable. It's more of a psychological problem than anything else. The only time it really matters is if you are displaying video or images from a PC that is in 0-255 mode. And if your TV supports PC 0-255 mode you'll probably find all the steps are visible anyway. Consumer discs and TV are not supposed to contain any data below16 or above 235 anyway. If there's any data out there, it's bogus info. Your TV will be setup to not show any change in black level if it receives data below 16 - so if there happens to be blacker-than-black bit in there somewhere, it will just show up as black. If something similar happens on the top end, there's really no issue since video is supposed to "stop" at 235 anyway.


There are some very impressive video displays that display 237-255 as if all those values were 235 and nobody ever complains about "missing something" on those displays.

No, video is not supposed to stop at 235. It is expressly designed to extend to 254, and that is explicitly stated in the video standards. Further, this discussion we've had before is something of a simplified version that only deals with RGB. The YCbCr that's on the disc is 16-240 for chroma, and even assuming a hard clip right at 240 and at 235 for Y, this will still yield RGB values that are outside the nominal range upon conversion to RGB.


This occurs with a great deal of content, probably all content that has been handled well and has not been errantly clipped off. And even content that may have been hard-clipped in component will still yield values that are beyond 235.
 
#19 ·

Quote:
Originally Posted by thebland /forum/post/16478171


Speaking of whites...


I'm doing a calibration best as I can until my calibrator arrives..


In setting the contrast, it seems I am clipping whites.. Can only see up to bar #246 on the Spears and Muncil Blu Ray test disc.. Blacks are fine. SO, I seem to be clipping whites..


Is this something that requires service menu and / or a calibrator's expertise? No matter how high / low I move my contrast (and brightness), I can't see all bars... Now on the 11 pt scale, all is good. This problem test pattern has at least 20 white gradations and I cannot see them all - I can see 14/20... is this nomal or should I see all?


Thanks for the help / explanation.

Ideally your system wouldn't be clipping them. It may be a display issue, or a source issue, or some other processing in the chain. There are a variety of settings that can impact this, so without knowing what your system is, it's hard to say. Hopefully your calibrator will be able to spend some time chasing this down and rectifying it.


However, as ADU said, and I would agree, this is just one part of the system performance. If you can't get all of the peak whites, this may not be that big a deal. Further, if the 'fix' causes other more serious side-effects, then maybe stick with a little bit of clipping. and if you're seeing all the way to 246, that should be pretty much everything you'll see on a normal disc. If it was hard right at 235, I might be a little bit more concerned, but even still not hugely concerned if it wasn't fixable. In other words, it isn't the end of the world, but it is unfortunate.
 
#20 ·
I also made the error (imo) of referring to RGB in the first draft of my last post above. However, in the context of this discussion, I think it makes sense to focus primarily on Y'CbCr (since that's what's used for HDTV, Blu-ray discs, DVDs, and the calibration discs), and more specifically on the Luma or Y' component of Y'CbCr, because that's where all the B&W information on the discs is contained. I think the chroma components are just offsets of Y'. So as I mentioned above, the black and white points should be defined solely by Y', which has a nominal range of 16-235.


Other things may happen to the signal and those values once they're translated back into the RGB domain in the display, but that'll be out of most users' hands (unless they're tinkering with other controls in the service modes on their displays). And unless I'm mistaken, it's the Y' component from which the PLUGE and white clip tests should be derived on calibration discs like the one Jeff is using. So for purposes of setting the black and white points on the display via the Brightness and Picture/Contrast controls, the Y' value is probably all you need to be concerned about.


On some discs and broadcasts, there may indeed be some info in the WTW range above Y'=235. Since that falls outside of the nominal range for Y'CbCr video (as defined by the black and white points of Y'=16 and Y'=235), it's open to debate how valuable/important it is to be able to see that information. Ultimately I think the individual user has to make that call.


If retaining the WTW information has no other negative impact on the picture quality (especially the contrast ratio), then it's a non-issue. And there's really no point in clipping the WTW info off.


However, in many instances (such as the dim projector scenario I gave above), this will not be the case. And displaying WTW will mean sacrificing some useful contrast, and brightness in the display's whites. In this case, it's really up to the user to decide which will deliver a more satisfying picture... a little brighter whites and a bit more contrast, or a bit more detail which may or may not show up in the WTW range. For some folks a compromise between the two may make the most sense.
 
#21 ·

Quote:
However, in many instances (such as the dim projector scenario I gave above), this will not be the case. And displaying WTW will mean sacrificing some useful contrast, and brightness in the display's whites. In this case, it's really up to the user to decide which will deliver a more satisfying picture... a little brighter whites and a bit more contrast, or a bit more detail which may or may not show up in the WTW range. For some folks a compromise between the two may make the most sense.

Yup, in a nutshell.
 
#22 ·

Quote:
Originally Posted by ChrisWiggles /forum/post/16485966


No, video is not supposed to stop at 235. It is expressly designed to extend to 254, and that is explicitly stated in the video standards. Further, this discussion we've had before is something of a simplified version that only deals with RGB. The YCbCr that's on the disc is 16-240 for chroma, and even assuming a hard clip right at 240 and at 235 for Y, this will still yield RGB values that are outside the nominal range upon conversion to RGB.


This occurs with a great deal of content, probably all content that has been handled well and has not been errantly clipped off. And even content that may have been hard-clipped in component will still yield values that are beyond 235.

YCbCr is converted to 16-235 RGB. Just because Cb and Cr can carry values up to 240 does not mean those values are present in the resunting RGB conversion. Your ability to obfuscate a simple process and concept is beyond compare (and beyond comprehension).
 
#23 ·

Quote:
Originally Posted by Doug Blackburn /forum/post/16493066


YCbCr is converted to 16-235 RGB.

Right. But what I'm saying is that YCbCr triplets that are clipped (to 16-235 for luma and 16-240 for chroma) will still yield RGB triplets that exceed 16-235 RGB range.


For instance, the perfectly legal YCbCr triplet of (200, 160, 150) in Rec.709 decodes to an RGB triplet of (250,181,240). Note that the component values don't even come anywhere close to the legal limit for video luma or chroma, yet they already exceed the nominal range in RGB for both red and blue. Unless an additional clip is imposed on the playback side in RGB (for no good reason), video values will regularly fall outside the nominal RGB range. An imposed clip at the playback side in RGB would change the color that was encoded on the disc, and I would argue that this is a detriment and a deviation from accuracy since this is not how it was observed(arbitrarily clipped) when mastered.

Quote:
Just because Cb and Cr can carry values up to 240 does not mean those values are present in the resunting RGB conversion.

Right. The component video cube is significantly larger than the RGB cube, and values that are much farther from the extremes of YCbCr will approach the extremes of the RGB cube. Values significantly lower in YCbCr can and do exceed the RGB nominal range. Surely you would not characterize an YCbCr triplet of (200, 160, 150) as being at the extremities of the component video space? In my example above for instance, observe that I got a value of 240 for B, and my chroma values aren't anywhere near 240.


If my simple and arbitrary (but effective) example is not convincing, lets look at some real content, this time provided by Stacey Spears, in his thread here:
http://www.avsforum.com/avs-vb/showt...5#post16475585






The histogram shading indicates the nominal reference range for each channel, as stacey describes in the thread:
Quote:
Originally Posted by sspears /forum/post/0


The histogram background has a light and dark gray. For YCbCr, the dark is 1-15 and 236-254 for Y and 1-15 and 241-254 for Cb/Cr. For RGB, it is 1-15 and 236-254.

Quote:
Your ability to obfuscate a simple process and concept is beyond compare (and beyond comprehension).

I'm attempting to clarify, not obfuscate. Your efforts to ignore all new information is pretty enterprising. If you put as much effort into expanding your knowledge rather than defending your existing understanding, it would be far more satisfying for everybody.


Now, it would be an interesting to have a discussion about the impacts and prevalence of data that exceeds the nominal bounds in component or in RGB, and the visual impacts of clipping that data. But if you're for some strange reason hell-bent convinced that it is simply impossible for data outside those bounds to exist in either the content (YCbCr) or after decoding (in RGB) despite ample evidence to the contrary, then it's impossible to have any kind of intelligent discussion about, for instance, where to set a display's peak white level in relation to the either content's nominal reference or peak white points (or some compromise between them as ADU mentions above). That's a fair and elucidating discussion to have. Sitting around and having a tantrum because you simple don't believe that such data could ever exist, or that no digital filtering occurs simply because you are under the impression that filters can only be analog, or made of cloth for your car or your vacuum cleaner doesn't really get us anywhere.


But that example that Stacey provided is just a random frame from a blu-ray movie. Who cares about blu-ray, that's not a benchmark of anything, right? Surely the animators of Cars just didn't know what they were doing, didn't care about video, the mastering engineers relegate new BD releases to second-rate assistants and interns. Right...


edit: all RGB and Y in this post should be primed. I left that out from laziness and since it should be obvious I don't mean true Y or true RGB.
 
#24 ·

Quote:
YCbCr is converted to 16-235 RGB.

That is incorrect. Neither PC or VIDEO color conversion will restrict YCbCr 16-235 into RGB 16-235.


Before anyone says something about Car's being an animation, I can post a frame from any HD title and show the same thing. An example I gave in the other thread was: Y 140, Cb 161 and Cb 202 would convert to R 254, G 100, B 200 using the BT.709 color matrix. If you used PC levels, which expand, you would get R 277, G 98 and B 214. Since we don't go above 255, R would get clipped to 255.


In case someone is not aware what Chris meant by primed, he is talking about Y' vs. Y or R'G'B' vs. RGB.
 
#165 · (Edited)
Quote:

YCbCr is converted to 16-235 RGB.


That is incorrect. Neither PC or VIDEO color conversion will restrict YCbCr 16-235 into RGB 16-235.

Before anyone says something about Car's being an animation, I can post a frame from any HD title and show the same thing. An example I gave in the other thread was: Y 140, Cb 161 and Cb 202 would convert to R 254, G 100, B 200 using the BT.709 color matrix. If you used PC levels, which expand, you would get R 277, G 98 and B 214. Since we don't go above 255, R would get clipped to 255.

In case someone is not aware what Chris meant by primed, he is talking about Y' vs. Y or R'G'B' vs. RGB.
This only proves the wrong matrix was used to convert Cars to YCbCr. They used a matrix for 16-235 and fed in input signals that went up to 255. It's easy to prove. Convert a RGB value of 255, 16, 16 to YCbCr using a matrix for 16-235. You'll get about 67, 100, 250 for the YCbCr value. Turn that back into RGB using a matrix that generates 0-255 RGB and you get 278, 0, 0 for your RGB values. Gee, that sure looks familiar...

You're advocating setting contrast in such a way to compromise the white level of the display and reduce the contrast of the display to compensate for the fact that some small percentage of content is incorrectly mastered. If the correct 3x3 matrices are used and the spec is followed you can't get out of bound values like your Cars example (beyond expected quantization / rounding / truncation errors which are minor). The solution shouldn't be to muck up the calibration of your display, but to beat on mastering houses to stop screwing up.

By the way, why don't you advocate doing the same thing for the black side setting the black level so we can see all the way down to 0? Why are you content to chop off mastering errors on the black side, but not the white side?
 
#25 ·

Quote:
Originally Posted by sspears /forum/post/0


An example I gave in the other thread was: Y 140, Cb 161 and Cb 202 would convert to R 254, G 100, B 200 using the BT.709 color matrix. If you used PC levels, which expand, you would get R 277, G 98 and B 214.

R254, G100, B200 would be invalid and out-of-gamut for StudioRGB. Invalid and out-of-gamut ‘data’ are considered errors or mistakes by many purists who respect and toe the correct line and gamut for .709.


Of course such invalid and out-of-gamut ‘data’ or errors would get clamped with sRGB. sRGB doesn’t reserve ~36% of codeword volume for invalid and out-of-gamut ‘data’ or errors.


Dave
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top