or Connect
AVS › AVS Forum › Display Devices › Display Calibration › Setting contrast with Spears and Munsil disc
New Posts  All Forums:Forum Nav:

Setting contrast with Spears and Munsil disc - Page 3

post #61 of 484
Quote:
Originally Posted by PlasmaPZ80U View Post

if there are clear examples i'd still like to see them, but at this point i'm familiar with more practical considerations than any real-world video examples that might put much emphasis on maintaining a tight grayscale above reference white.

+1

It's not about grayscale.

Video is YCrCb
Your display is RGB

YCrCb is based on overall luminance, there thousands of colors that are dimmer than 235,235,235 that cause R,G or B to have excusions beyond 235.
post #62 of 484
Quote:
Originally Posted by sotti View Post

Forgive my ignorance, I'm focused mainly on the formula side of the equation. Where is the list of valid YCrCb that are 16-235 for Y and 16-240 for the chroma channel that are invalid?

Also I don't know what you mean by invalid and reserved for sRGB, every computer system I've ever seen goes 0-255 for R,G,B.

I don't have a list, but most possible YCbCr triads map to an out-out-gamut or nonsensical color and are invalid. There are only ~2.8 million valid, in-gamut YCbCr triads. See Poynton.

0 and 255 are valid for sRGB. They are invalid and reserved in StudioRGB. sRGB is not StudioRGB. sRGB uses 0 for black and 255 for white (8-bit). StudioRGB uses 16 for black and 235 for white.
sRGB makes no allowance for, or reservation of, the limited codewords for anything other than valid color information. Nor seemingly needs to. (witness the millions of beautiful hi-res images posted everywhere @ sRGB levels.)

Dave
post #63 of 484
Quote:
Originally Posted by sotti View Post

Forgive my ignorance, I'm focused mainly on the formula side of the equation. Where is the list of valid YCrCb that are 16-235 for Y or 16-240 for the chroma channel that are invalid?

Also I don't know what you mean by invalid and reserved for sRGB, every computer system I've ever seen goes 0-255 for R,G,B.

Argh, don't make my eyes bleed. 0 and 255 are reserved for video (SD or HD). We are talking about video right, as in BD and televisions?

Quote:
5.2 1080i, 1080p, and 720p
The default color space used by the high definition formats is that specified in ITU-R BT.709-5 [6].
ITU-R BT.709-5 Part 1, Section 4 [6] (or CEA-770.3-C Sections 5.4-5.7 [20]) shall be used for any color space conversion needed in the course of processing unless a different colorimetry is specified in the AVI.
The digital representation shall be as defined in Part 1, Section 6.10 of ITU-R BT.709-5 and is summarized below:
The coding shall be 8-bit coding (scale of 0 to 255). R, G, B, and Y signals shall have 220 quantization levels with the black level corresponding to level 16 and the peak white level corresponding to level 235. The signal level may occasionally move beyond level 235. CBCR signals shall have 225 quantization levels with a zero level corresponding to digital level 128 and full range corresponding with 16 to 240. For R, G, B, Y, CB, CR signals, 0 and 255 are reserved and should not be considered video. See section 5.4 for 640x480p and other IT video formats.
post #64 of 484
Quote:
Originally Posted by bodosom View Post

Argh, don't make my eyes bleed. 0 and 255 are reserved for video (SD or HD). We are talking about video right, as in BD and televisions?

What I'm refering to is the fact that your television is driven by RGB.

The YCrCb, valid reference data from 16-235 for lumanince data and 16-240 for valid chroma data is converted to RGB to drive the panel. It's that RGB data I'm concerned about. Bright saturated colors can easily both be described inside the valid YCrCb gamut and include values above 235.

I don't have the time or the equipment to digging to analyze a disc to find where transformed YCrCb values exceed 235 in RGB space.

If you want further clarification, just go ask Poyton! he's on spectracal's forum now.
post #65 of 484
Here’s one definition of what is valid in StudioRGB…

Quote:
Originally Posted by from Poynton View Post

The so-called valid colors encompass the volume that is spanned when each R’G’B’ component ranges from reference black to reference white. In Rec. 601, each component has 219 steps (risers) – that is, 220 levels. That gives 220*220*×220, or 10648000 colors: About 64% of the total volume of codewords is valid.

All the (and only the) BTB/WTW triads are outside of this definition of valid. If they aren’t classified here as valid, then what would that make them?

For YCbCr triads, there are much less that are valid (~2.8 million). YCbCr is recognized as have very poor valid codeword utilization.

Dave
post #66 of 484
Quote:
Originally Posted by alluringreality View Post

I tried going through some random commercial DVDs I own a couple times, and in the end I personally could not find anything that I could say was clearly intended to be brighter than 235 gray.

+1. This matches my personal measurements as well. I’ve never measured one pixel from mainstream studio material where even one component even approached 254. I also have never measured one pixel where all three RGB components were >235. I found thousands of pixels at 235,235,235 but never one at 236,236,236 or greater. That said a lot to me about how often the BTB/WTW range is used by mainstream studios. I haven’t done such measurements for many years though.

Dave
post #67 of 484
Quote:
Originally Posted by sotti View Post

What I'm refering to is the fact that your television is driven by RGB.

I assume at this point it's clear there's RGB and then there's RGB. sRGB is essentially irrelevant to broadcast video. StudioRGB is not sRGB. While my panel ultimately emulates sRGB that's useful only so far as my using it as a monitor for my computer. The thread and the motivating question are about clipping broadcast video signals to the levels I referenced from the ITU via CEA.

This is an area of frustrating contention and we should try write carefully and precisely. Or just refer people to Poyton. About which people will still argue but who cares.

I repeatedly make the point that beyond all of the numbers manufacturers will read sections 5.1 and 5.2 and conclude that it's acceptable to clip and build products that do exactly that.
post #68 of 484
Quote:
Originally Posted by dlarsen View Post

Here’s one definition of what is valid in StudioRGB…



All the (and only the) BTB/WTW triads are outside of this definition of valid. If they aren’t classified here as valid, then what would that make them?

For YCbCr triads, there are much less that are valid (~2.8 million). YCbCr is recognized as have very poor valid codeword utilization.

Dave

If that's what they do in the studio, setting everything to fit in 16-235 RGB before they encode to YCbCr then you're absolutely right that none of the values on a disc would fall outside of that range.

I'm not that familiar with studio workflow.

I've just been going over YCbCr conversion stuff lately and know what the "valid" range for YCbCr is, and that many valid values there map to outside RGB 16-235.
post #69 of 484
Quote:
Originally Posted by sotti View Post

If that's what they do in the studio, setting everything to fit in 16-235 RGB before they encode to YCbCr then you're absolutely right that none of the values on a disc would fall outside of that range.

I'm not that familiar with studio workflow.

You're putting words in my mouth. I never claimed here that this is what studio do or don't do. I quoted Poynton as to the definition of 'valid' as it applies to StudioRGB space. His definition that I quoted, clearly excludes each and every BTB/WTW triad as being classified as valid in StudioRGB. Personally, I’m more concerned about reproducing all the valid pixels and maximizing the valid dynamic range that I’m presented with with mainstream material. I have much less concern about any (theoretically) possible invalid or nonsensical pixels.

Dave
post #70 of 484
Quote:
Originally Posted by sotti View Post

If that's what they do in the studio, setting everything to fit in 16-235 RGB before they encode to YCbCr

That isn't what they do in post. But then, taking a DVD as an example, so many bad things happen in post that it's no exemplar of correct behavior. Odd things on disc are just another example of idiosyncratic application of technology. The questions in this thread regarding clipping are all about compromises and a punctilious hewing to some engineering ideal doesn't much help folks that wonder why they have trouble "Setting contrast with Spears and Munsil". I maintain that in the grand scheme of things there are bigger problems to solve than clipping AV gear (no disrespect).
post #71 of 484
Quote:
Originally Posted by dlarsen View Post

You're putting words in my mouth. I never claimed that this is what studio do or don't do. I quoted Poynton as to the definition of 'valid' as it applies to StudioRGB space. His definition that I quoted, clearly excludes each and every BTB/WTW triad as being classified as valid in StudioRGB.

Dave

Well I never said that those values would every be present, so now are you putting words in my mouth?

I'm talking about what happens in the TV as YCbCr gets processed for display, and what values would be legal. For instance 64,102,240 YCbCr would transform to 240,16,16 RGB, wich could be displayed on a set that is calibrated for WTW.

So I guess the question is do all DVD/BD's go through a StudioRGB step or not? If not then WTW data is inherntly valuable becaue each channel may indvidually excurse above 235 (although not in unision) and still create a valid YCbCr value. If they are run through a StudioRGB step then no data should exsist there at all.
post #72 of 484
Quote:
Originally Posted by bodosom View Post

That isn't what they do in post. But then, taking a DVD as an example, so many bad things happen in post that it's no exemplar of correct behavior. Odd things on disc are just another example of idiosyncratic application of technology. The questions in this thread regarding clipping are all about compromises and a punctilious hewing to some engineering ideal doesn't much help folks that wonder why they have trouble "Setting contrast with Spears and Munsil". I maintain that in the grand scheme of things there are bigger problems to solve than clipping AV gear (no disrespect).

I somewhat disagree. Obviously we can encode steps all the way to 254(5) for RGB in YCbCr, how else would we generate the ramps?

But even though 16-235 are the only valid values for white, the chroma channels are valid through 240 and you can legally describe 240,16,16 - 16,240,16 - 16,16,240. If the encoding to YCrCb comes from a source that hasn't been pre-clipped to 235 in RGB space these values should exsist in video.

The easiest way to validate that those values aren't being clipped to run your grayscale ramp all the way up.
post #73 of 484
just another conversion trial

the YCrCb legal value of Y-183 Cb-38 Cr-22 transforms to RGB 16,249,16.
post #74 of 484
Quote:
Originally Posted by Sotti View Post

Well I never said that those values would every be present, so now are you putting words in my mouth?

OK. The quote and my response here has made no reference to what is or is not present. Only what is classified by the Poynton quote as valid in StudioRGB.

Quote:
Originally Posted by Sotti View Post

So I guess the question is are all DVD/BD's go through a StudioRGB step or not?

Yes, I believe most all telicined material inherently starts out, or is placed early on, in StudioRGB space. I’m sure I’ll be corrected if I’m wrong. This is the color space that Poyntons definition of valid is referring to. All excursions into BTB/WTW are excluded from his classification of possibly being valid.

Quote:
Originally Posted by Sotti View Post

If not then WTW data is inherntly valuable becaue each channel may indvidually excurse above 235 (although not in unision) and still create a valid YCbCr value.

If the WTW StudioRGB space (235-254) is, or is supposed to be used for valid, intended, extended video information, why can’t they excurse in unison? Why would any excursion into StudioRGB WTW be specifically excluded from the definition of valid color triad?

Quote:
Originally Posted by Sotti View Post

If they are run through a StudioRGB step then no data should exsist there at all.

Again, I’m not arguing if it exists or not, or how much, or what it is, (but I can), I’m just pointing out that by the referenced Poynton quote, any and all BTW/WTW triads in StudioRGB (individually or in unison) are specifically excluded from being classified as a valid color pixel triad.

Being able to utilize the entire BTB/WTW range is a huge issue for calibration vendors. I do understand that.

Dave
post #75 of 484
Quote:
Originally Posted by sotti View Post

just another conversion trial

the YCrCb legal value of Y-183 Cb-38 Cr-22 transforms to RGB 16,249,16.

Legal doesn't necessary equate to a valid color. There are only ~2.8M valid color YCbCr triads. Where does 183,38,22 YCbCr map to on the rec.709 YCbCr gamut volume? Is it in-gamut valid?

Dave
post #76 of 484
Quote:
Originally Posted by dlarsen View Post

Legal doesn't necessary equate to a valid color. There are only ~2.8M valid color YCbCr triads. Where does 183,38,22 YCbCr map to on the rec.709 YCbCr gamut volume? Is it in-gamut valid?

Dave

You simply clip the values to legal numbers, all values out of gamut are mapped to their nearest in gamut equivelent.

All though all that out of gamut data is what will be used to extend the gamut with xvYCC.

If StudioRGB is THE standard and YCbCr is just a transport mechanism, then there really isn't even a need to calibrate higher than 235.
If YCbCr is THE standard, then you need to calibrate all the way to 254(5).

There has to be a standard right?
post #77 of 484
Quote:
Originally Posted by dlarsen View Post

Here's one definition of what is valid in StudioRGB

All the (and only the) BTB/WTW triads are outside of this definition of valid. If they aren't classified here as valid, then what would that make them?

For YCbCr triads, there are much less that are valid (~2.8 million). YCbCr is recognized as have very poor valid codeword utilization.

Dave

This is the same semantic bulls**t you've been trying to play for years.
post #78 of 484
Quote:
Originally Posted by dlarsen View Post




Yes, I believe most all telicined material inherently starts out, or is placed early on, in StudioRGB space. I'm sure I'll be corrected if I'm wrong. This is the color space that Poyntons definition of valid is referring to. All excursions into BTB/WTW are excluded from his classification of possibly being valid.




Dave


I regularly handle telecined material and its never been clipped to 235 in RGB.
post #79 of 484
Quote:
Originally Posted by dlarsen View Post

OK. The quote and my response here has made no reference to what is or is not present. Only what is classified by the Poynton quote as valid in StudioRGB.



Yes, I believe most all telicined material inherently starts out, or is placed early on, in StudioRGB space. I’m sure I’ll be corrected if I’m wrong. This is the color space that Poyntons definition of valid is referring to. All excursions into BTB/WTW are excluded from his classification of possibly being valid.



If the WTW StudioRGB space (235-254) is, or is supposed to be used for valid, intended, extended video information, why can’t they excurse in unison? Why would any excursion into StudioRGB WTW be specifically excluded from the definition of valid color triad?



Again, I’m not arguing if it exists or not, or how much, or what it is, (but I can), I’m just pointing out that by the referenced Poynton quote, any and all BTW/WTW triads in StudioRGB (individually or in unison) are specifically excluded from being classified as a valid color pixel triad.

Being able to utilize the entire BTB/WTW range is a huge issue for calibration vendors. I do understand that.

Dave

Actually, Poynton doesn't really refer to R'G'B' as being "valid" or not at all, because it's kind of redundant. He only refers to R'G'B' signals as being "legal." A different term, "valid" is only useful when you move to Y'CbCr space where we have to distinguish what signals are appropriate within that space ("legal") and then what those signals represent in R'G'B' space ("valid/not valid")

The standards explicitly allocate this range for video, and it's quite useful. For instance, you can't make a non-neutral white without excursion beyond reference white. That's something that you cannot do in sRGB without lowering the luminance of the white point in such a scene.

You've been asked for years why you have such a problem with studio levels, and what benefits would be gained by clipping them off, and you have never come up with any reasonable answer for that.

It gets old when you continue reposting the same crap over and over, and when you won't answer anybody's questions.
post #80 of 484
Quote:
Originally Posted by sotti View Post

You simply clip the values to legal numbers, all values out of gamut are mapped to their nearest in gamut equivelent.

All though all that out of gamut data is what will be used to extend the gamut with xvYCC.

If StudioRGB is THE standard and YCbCr is just a transport mechanism, then there really isn't even a need to calibrate higher than 235.
If YCbCr is THE standard, then you need to calibrate all the way to 254(5).

There has to be a standard right?

And even if we're talking pure 'Studio RGB' you'd still want to preserve all the way up to 254. If it's in the content, you want to see it. And most all content excurses beyond 235 in RGB. There is no positive reason to clip content at 235. You don't gain anything by doing that, which is why it usually isn't done.
post #81 of 484
Look there is a really easy way to sort this out...

Sample some BD/DVD and see if it contains level outside 235 in RGB ( this has actually been done by S&M amongst others)

See if the picture looks better with the display white point mapping 235 or mapping 255.

Rinse/repeat etc.
post #82 of 484
Just checked some NTSC material off D1.

Its clouds in bright daylight with the sun visible . Not the nicest material in the world looks like old timelapse stock footage telecined from film.

Level tops out at 254 RGB.

If I clip it at 235 and drag 235 up to 254 to mimic the display criteria of the original material...looks truly awful.

this is on a rec 709 lutted workstation
post #83 of 484
Pretty much everything I look at also exceeds 235 in RGB. Shouldn't surprise anyone.
post #84 of 484
Quote:
Originally Posted by sotti View Post

I somewhat disagree. Obviously we can encode steps all the way to 254(5) for RGB in YCbCr, how else would we generate the ramps?

But even though 16-235 are the only valid values for white, the chroma channels are valid through 240

Please note that I'm using your response as a bit of a springboard.

First: as I quoted CbCr is 16-240 (my emphasis).

Quote:


The digital representation shall be as defined in Part 1, Section 6.10 of ITU-R BT.709-5 and is summarized below:
The coding shall be 8-bit coding (scale of 0 to 255). R, G, B, and Y signals shall have 220 quantization levels with the black level corresponding to level 16 and the peak white level corresponding to level 235. The signal level may occasionally move beyond level 235. CBCR signals shall have 225 quantization levels with a zero level corresponding to digital level 128 and full range corresponding with 16 to 240. For R, G, B, Y, CB, CR signals, 0 and 255 are reserved and should not be considered video.

I don't choose to argue about or care much about technical minutia. I don't even care much what happens in post. I have no control over that. What I care about is what my equipment can do. And like any number of other bits of AV stuff it clips outside the STANDARD range. Yes, that's right, CEA-861-D is a standard followed by consumer electronics manufacturers world-wide. So my wonderful 9g Kuro monitor clips because Pioneer read the standard (it also obeys the quantization flags in 861/HDMI 1.3, possibly the only display to do so). My AVR used to clip (because you can program a REON to clip) until I convinced them it was a good thing not to clip.

All the obstruse technical arguments in the world are irrelevant unless they take operation standards into account.

Said standards are the reason for this and various other threads asking about "Setting contrast with Spears and Munsil disc" or any other reference material.

On the bright side these things can actuall be fixed (just join CEA) -- unlike all the crappy DVDs you own with bogus MPEG flagging and horked telecine.
post #85 of 484
Quote:
Originally Posted by bodosom View Post

Please note that I'm using your response as a bit of a springboard.

First: as I quoted CbCr is 16-240 (my emphasis).



I don't choose to argue about or care much about technical minutia. I don't even care much what happens in post. I have no control over that. What I care about is what my equipment can do. And like any number of other bits of AV stuff it clips outside the STANDARD range. Yes, that's right, CEA-861-D is a standard followed by consumer electronics manufacturers world-wide. So my wonderful 9g Kuro monitor clips because Pioneer read the standard (it also obeys the quantization flags in 861/HDMI 1.3, possibly the only display to do so). My AVR used to clip (because you can program a REON to clip) until I convinced them it was a good thing not to clip.

All the obstruse technical arguments in the world are irrelevant unless they take operation standards into account.

Said standards are the reason for this and various other threads asking about "Setting contrast with Spears and Munsil disc" or any other reference material.

On the bright side these things can actuall be fixed (just join CEA) -- unlike all the crappy DVDs you own with bogus MPEG flagging and horked telecine.

And the Kuro is wrong. There are many consumer displays that don't clip, as they shouldn't. NO professional displays clip like this, and studios don't set up their displays to clip this way.

It is asinine for you to say you only care what consumer devices do, and that you don't care what studios do. These are studio standards, and it is the practices of studios that are of crucial interest to anyone who cares about image fidelity. And nowhere do studio standards state that anything should clip to 16-235. On the contrary they imply quite the opposite.

The list of consumer devices and displays that regularly and continually and consistently screw things up (including the Kuros) is endless. If your bar for image quality is what an average consumer device does, then that is a VERY low bar indeed.
post #86 of 484
Quote:
Originally Posted by ChrisWiggles View Post

And the Kuro is wrong ... It is asinine for you to say you only care what consumer devices do ...These are studio standards

I'm going to ignore this bit of unpleasantness are reiterate my point. There are STANDARDS (as in printed documents) that consumer electronics companies (like Pioneer) can follow. Amazingly even intelligent people can read an ambiguous document and come to disparate conclusions. Where CEA standards and SMPTE Recommended Practice papers diverge there are going to be problems. But pretending that CE companies don't have a basis for their decisions is simplistic.

By the way I don't care what studios and post houses do -- good or bad -- because I can't change it. If I purchase a some AV gear with idiosyncratic behavior I don't like then I can do something. Just like I don't care about the feud between you and dlarson. There's nothing I can do about it either (however I suggest that if you two can agree that Poynton is the authority then send mail to his question/answer address at SpectraCal and have him resolve the issue).
post #87 of 484
Quote:
Originally Posted by bodosom View Post

There are STANDARDS (as in printed documents) that consumer electronics companies (like Pioneer) can follow.

That's right. And still they manage to screw things up with impressive regularity.

It took nearly a decade for many manufacturers to figure out how to handle chroma with progressive-scan DVD players correctly. And it is STILL commonly screwed up.

It's taken them practically a generation to figure out how to decode NTSC accurately.

The reference is the studio. Period. That is where we discern the artistic intent, and that is the image we are attempting to reproduce.

Consumer devices are often a cruel joke in comparison.

The Kuro is wrong, and it is unfortunate. It's not surprising though: it's not a professional display.
post #88 of 484
Quote:
Originally Posted by ChrisWiggles View Post

...

The reference is the studio. Period. That is where we discern the artistic intent, and that is the image we are attempting to reproduce
....

The Kuro is wrong, and it is unfortunate. It's not surprising though: it's not a professional display.

Well only the film maker gets to see the truth then unless you're suggesting that what comes out the DVD/BD production house is the same as what's on the film. And don't be so quick to blame CE companies. Anyone can look at a DVD and see that it's rife with MPEG errors because of bad encoders or pilot error.

Actually Pioneer tried diligently to do the right thing. They include a switch that turns off clipping and as I said they're the only company I'm aware of that properly processes quantization information in AV Info frames.

And finally I don't think pro/reference monitors can realistically be used as a model of behavior. And not because not even Runco has the nerve to charge $30k for 17" display but because the monitor is just one part of the production chain addressed by SMPTE RPs.

It goes without saying that none of this provides any help to the OP anyone else asking how to "fix" their problems.
post #89 of 484
Quote:
Originally Posted by bodosom View Post

By the way I don't care what studios and post houses do -- good or bad -- because I can't change it.

My understanding is that the only reason we calibrate our displays at all is to emulate what the post houses do.

Calibration in it's purest form is intrinsictly about what the post houses do.
post #90 of 484
Quote:
Originally Posted by bodosom View Post

Well only the film maker gets to see the truth then unless you're suggesting that what comes out the DVD/BD production house is the same as what's on the film.

Increasing film makes view the DVD/BD release as the definitive version of their works. They are there in post making tweaks and changes.

Abyss was recently brought to my attention as a film that was lit completely different in the movie house (tungsten daylight look) then on video (the blue tinted movie we remember).

I listened to Ron Moore talk about making BSG and how he frequently made decisions slanted toward the version that will persist after airing (BD).

George Lucas clearly views the most recent DVD release of SW as the most canonical.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Display Calibration
AVS › AVS Forum › Display Devices › Display Calibration › Setting contrast with Spears and Munsil disc