Basic Guide to Color Calibration using a CMS (updated and enhanced) - Page 4

I take it is the Y of the 100 IRE Gray.
"Start" file Y=1108.55
"After" file Y=873.84

Tom,
Thank you for the comments. The errors you have observed are probably due to my having attached the wrong "after" file to my original post. I have already corrected the mistake (edited the post).
I apologize for the confusion this mistake may have caused.

Tom,
Thanks, this is great news.
What do I do now with contrast, brightness and gamma? If I try to max the dynamic range by decreasing the brightness to the best possible black level and then after I adjust contrast per S&V calibration dvd, the gamma curve will be negatively impacted and one or more colours will most likely get clipping badly and/or the high IRE gray levels will take a colour shade.
Is there a way to adjust contrast by using colorimeter measures and HCFR and thereby preventing clipping and/or colored grays?
What should I go for: a nice ,flat gamma or the best possible dynamic range w/o clipping and coloured grays? I do not think I can have both.
All comments/opinions welcome.
Thanks.
Fermin

A question about this advice:
Is it necessary to find the exact point that you can raise contrast while not affecting colours at 100 IRE? I mean, could contrast be set slightly lower than that point and still provide maximum contrast for the display?

Tom,

On the GetGray DVD there are step patterns (0 to 256?? digital) for each of the primary colors (RGB). I have noticed that when I set contrast as usually directed (2 IRE just visible, or equivalent), the upper steps of one or more RGB patterns are crushed. Is this of significance? Should contrast be lowered to remove the crush or is there some other adjustment that is needed.

Also, I have read about determining the "limiting color" and then when adjusting the grayscale to not increase the gain of the "limiting color." Could you comment on the concept of "limiting color," its significance, and how to determine it.

Thanks much for taking on this thread.

Hey guys,

A short but sweet update. My new 12k has much better uniformity, so far no flicker (only 5 hours though), and better OOTB color accuracy. The Spyder2 and the DTP-94 are very close in performance as long as you keep the light levels up for the Spyder (by placing it closer to, and facing the pj). I'm assuming the DTP is more accurate, but in this configuration, the Spyder is much faster. So, I'll use the Spyder to get a preliminary calibration, then fine tune it with the DTP.

On another subject, I beleive it was Tom that mentioned that my RGB levels (especially Green) were lower than should be necessary for a proper calibration on this unit. The reason they were so low was my method, posted on the forum quite awhile back, by Guy Quo. His method was basically, center Brightness/Contrast, set Red and Blue Offset too low and use Green Offset to set Brightness (make BTB match backround). Return Red and Blue Offsets to default. Then, set Red Gain to highest that doesn't cause clipping. Then use only the Green and Blue Gains, and the Red and Blue Offsets to calibrate. I've always assumed this was correct. I mean Guys knows what he's talking about...right?

Rob

robbyc30

That is they way I have always done things. Some calibrators prefer to calibrate at 80% rather than 100% though - and are willing to put up with the color shift on hot whites for the better contrast numbers. I get annoyed when highlites shift or clip - so I do it Guy Kuo's way. Also it leaves user brightness/contrast for source variation adjustment without having to worry about greyscale interactions.

Of course you want to generalize it and find out what these clip limits are - it is not always red on gain and green on offsets.

Thank you, you've answered my question.

Krasmuzik... Just out of curiosity, how would one determine what are the clip limits?

If your sensor software has a RGB% histogram then whenever it stops changing when you increase the RGB gain - it is clipped. Or you can use RGB ramp test patterns and try to sense the clipping by eye same as you would for the contrast white levels. If 90% white is on D65 target but 100% white is too cyan - it is likely clipped on Red.

HD-DVE has RGB (10IRE) step patterns which can help diagnose gross clipping as well.

Mark

I and many others have made the assumption that if the display decodes in Rec709, even if the the original source was Rec601 transcoded to Rec709 by a dvdplayer, that we should calibrate primaries to Rec709. Now I think that may be incorrect, consider the two cases:

1. RGB->YCbCr(601) (at telecine) : MPEG2->YCbCr(601) (480i/p at dvdplayer)-> : YCbCr(601)->RGB (at display)

2. RGB->YCbCr(601) (at telecine) : MPEG2->YCbCr(601)->Transcode->YCbCr(709) (720p/1080i at dvdplayer) : YCbCr(709)->RGB(at display)

We all can agree that case 1. should yield Rec601 primaries. In case 2. the transcode step is necessary for a display which assumes the Rec709 colorspace whenever an HD transport stream is used but the end result, the RGB vector should be identical to the one in case 1, shouldn't it? Otherwise we will have changed the intent of the original color vector. In other words, the 1st encoding step defines what Red Green and Blue are and subsequent decode/encode steps are not allowed to change that original definition. Thoughts, counter arguments? I am now thoroughly confused on this.

The DVD player has to do the transcode properly - rather than stretching REC601 out to REC709 - it needs to limit RGB to within the REC601 gamut within the REC709 gamut - which means limiting the YCbCr signal. An SD picture in a HD signal should appear to be a desaturated HD signal - and not made 'extra super vivid bonus' picture. Now which option do you think marketers choose that they 'think' customers prefer - especially if it cost no engineering resources to do?

This sounds like you agree that for color accuracy when we use SD source patterns, even those that go through the transcode step, we should tweak the display to reproduce SD primaries? Is that correct?

No - the signal is in REC709 so you should use REC709 primaries - however the signal itself is going to be desaturated so it never exceeds SMPTE-C gamut. Even if they choose to do it right (unlikely) - very likely they will not give you the control over the transcode process.

Since it is likely that REC709 sources itself was mastered on a SMPTE-C monitor - there are those who say don't even using REC709 primaries to begin with. If you do that - then the upconverted DVD problem solves itself if transcoding done wrong - but get even more desaturated if they did it right. These cross format codec and display isssues makes it near impossible to guarantee you see what the mastering engineer saw.

I think that just restates my original post. If it's done right and you will never see colors outside the original SMPTE-C gamut then as a calibration issue you can only use this test pattern to align to the original SMPTE-C gamut. If it's done wrong, by stretching one gamut to the other, then it that case I agree you should align to Rec709.


This was another issue that I was going to throw into the OP but glad you brought it up. Does there exist an equivalent to a SMPTE-C monitor for Rec709? If HD is being mastered with a calibration to SMPTE-C phosphors, then regardless of the encoding matrices we should be calibrating everything to SD primaries.

We are both right - I was thinking using RGB to align the gamut to aviod issues - you are using upconverted YCbCr to align the gamut.

Guy Kuo was insistent that nobody uses HD gamuts in mastering only SMPTE-C. He no longer posts here - maybe he will if they ever do start mastering HD with HD gamuts. Hopefully you have enough calibration memory to go both ways!

So does this mean that when playing an upconverted DVD (let's say through a HTPC) than the display should be on a mode that is calibrated to 601; while when playing HD material than the display should be on a mode that is calibrated to 709?

Or shoudl the upconverted DVD be played on modes calibrated to 709 as well?

This post by Glimmie indicates to me that for movies made from the digital intermediate process, home video is now coming from an underlying CIE XYZ colorspace..

CIE XYZ colorspace is founded on human perception limits - so I am not sure what Glimmie is claiming it does alien-only colors - rather it would be correct to say it can store colors that currently cannot be displayed unless we have multi-primary laser systems. The new support for outside gamut colors in HDMI1.3 could take advantage of full CIE XYZ if we ever get a media format founded on Digital Cinema - neither high def DVD format supports this - so you can forget home movies in this format - maybe PS3 will do video games this way though. But you still need something to display it on.

All that RGB is is an abstraction that later assigns the RGB primaries to an XYZ color- the true measure of color seen by humans - so an XYZ colorspace is not limited by intended primaries but its RGB based display most certainly is..

The color space of DigitalCinema is irrelevant - the question is what monitor are they color correcting it on? Is it the same as the Digital Cinema Projector - or is it a SMPTE-C monitor? Home video formats are founded on D65 - film and Digital Cinema are founded on different white points - this means home video has to be color corrected for its own format. Which means somebody will look at it it and sign off on it looking right - unless they want to rely on transcoders getting it right via 'perfect' math - and the directors don't want the opportunity to make artistic compromises to fit the larger gamut into a smaller one.

Under the assumption that when dvd players upconvert they map 601 primaries to 709 primaries then for calibration purposes you can use a 601 encoded gamut to calibrate your display for 709 material. i.e. if the transcode process is simply this:

YCbCr(601)->RGB->YCbCr(709) with no tinkering of the RGB values.

However, 601(or more properly SMPTE-C) mastered disks encoded in rec601 when upconverted with this transcode process will display colors not intended by the colorist. Of course you can avoid this whole issue by playing the disk in 480i/p and let your display do the upconvert. Your specific question though is "should I set the display to use rec601 primaries with upconverted SD material", and I think the answer to that is yes, again assuming the player truly maps one color space to the other. If after the transcode process all we do is this:

YCbCr(709)->RGB, the RGB values chosen during the mastering process are preserved.

Think of it this way, at the beginning of the chain is a SMPTE-C phosphor, if all the encode/decode steps are color neutral(meaning the RGB values are not mucked with) at the end of the chain (your display) you want SMPTE-C phosphors.


As far as HD material goes it's not clear what the correct color space to use is and somebody more familiar with HD mastering please chime in. But my thinking is that if the colorists are really using SMPTE-C reference then it depends on whether the encoding process is fudged to preserve this color within the wider rec709 gamut (in which case you would set your display to rec709) or whether, like the dvd player example, the primaries are mapped directly to rec709 in which case rec601 primaries at the display would be appropriate.

Hi, gang.

I've been reading up on the Plasma pages over the last couple of months and have finally settled on the Panasonic TH-58PZ750U. It's scheduled to be delivered this week, but I'm trying to line up everything I'll need beforehand...including calibration specifics.

I'll have to admit that the content of this thread is somewhat foreign to me, but I'm coming up to speed on devices, DVDs, and terms fairly quickly.

The question I have is fairly simple: What changes to calibration are needed, given this new "Studio Reference Mode" feature that is available on the new 750Us?

According to Panaconic, the Studio Reference Mode "enables the replication of colors that was originally intended by the directorexactly the way they are seen on reference monitors in professional film editing rooms."

The manual for the 750U (careful...it's 30M) has more specifics on pages 32/33, and 56. Here are the Picture menu and sub-menu options (pgs. 32/33):

Normal
Picture mode
Picture, Brightness, Color, Tint, Sharpness
Color temp.
Color mgmt.
C.A.T.S.
Pro setting
Zoom adjust
PC adjust
Other adjust
Video NR
3D Y/C filter
Color matrix
Block NR
Mosquito NR
Black level
3:2 pulldown
HD Size


Here is the text from page 56 Technical Information, regarding the "Pro Setting":

Normal: Resets all Pro setting adjustments to factory default settings. (Set/No)
Panel brightness: Selects the display panel brightness.
(High: Vivid, Mid: Standard, Low: Studio ref)
Contour emphasis: Adjusts the contour emphasis of the image. (On/Off)
Gamma adjust: Gamma correction. Adjusts the intermediate brightness of the image. (Normal/Mid/Full 1/Full 2)
Black extension: Adjusts the dark shades of the image in gradation.
W/B high R: Adjusts the white balance for light red areas.
W/B high B: Adjusts the white balance for light blue areas.
W/B low R: Adjusts the white balance for dark red areas.
W/B low B: Adjusts the white balance for dark blue areas.
AGC: Increases the brightness of dark signal automatically. (On/Off)


Is this new "Mode" just a marketing term for switching some Service Menu features to User Menu OR is this feature a realistic way of gaining a truer representation of colors for the end user? How does this mode change the calibration steps...if at all?

Thanks in advance!


P.S. Sorry for the lenthy post, but I'm sure many others have or will have the same question.

Based on the previous posts, I think it would be safe to think/state that when a dvd player does the upconversion AND outputs the signal through DVI(RGB) then it does not matter whether it has done the upconversion in the 601 or 709 space since , logically, when converting the signal back to RGB(DVI) it will use the same coding it used to upconvert it and will output in all cases the original /intended RGB signal. Right??
In the affirmative, wouldn't the display that gets an RGB signal tthroughDVI at its native resolution just display that RGB signal received in the same way, wwhetherit is SD or HD; yes??
In the affirmative, then, when using DVI, the primaries should be calibrated to rec 601.
When using a component connection, the problem remains whole: has the dvd player done the transformation to rec 709? since the display will receive a YCbCr signal that it will have to convert and has no way to know what colour space was used by the dvd player.
Kras, zoyd: does this make any sense?
Thanks ,
Fermin.

FGM

You cannot make blanket assumptions. What primaries does the display use with an RGB input - REC709 (same as sRGB)? Or does it use uncorrected native primaries? Or does it send RGB through a video codec to add back in video controls - and now it is REC601 and SMPTE-C?