How power law gamma calibration can lead to crushed blacks - Page 8 - AVS | Home Theater Discussions And Reviews
Baselworld is only a few weeks away. Getting the latest news is easy, Click Here for info on how to join the Watchuseek.com newsletter list. Follow our team for updates featuring event coverage, new product unveilings, watch industry news & more!


Forum Jump: 
Reply
 
Thread Tools
post #211 of 231 Old 12-03-2012, 04:15 AM
Advanced Member
 
Light Illusion's Avatar
 
Join Date: Aug 2010
Posts: 592
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 35 Post(s)
Liked: 263
In a calibration system where profiling is separate from calibration there is no reason for Gamma to affect Gamut...

The problem is when the two are interlinked, so that altering gamma causes a global change to the calibration - which it shouldn't.

Steve

Steve Shaw
LIGHT ILLUSION

Light Illusion is offline  
Sponsored Links
Advertisement
 
post #212 of 231 Old 12-03-2012, 04:54 AM
Senior Member
 
Jeroen1000's Avatar
 
Join Date: Nov 2008
Posts: 317
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 12
I had actually missed page 8 of this thread where there is an explanation. Except, judging from the explanation, gamma DOES affect the hue. Check out post #182 by Sotti. This going to sound either understandable or stupid:-):

Changing gamma does only not only change Y(luminance) for white, which we check by measuring the greyscale, but for all the colours (I found out you are British hehe).
Changing luminance is actually nothing more than altering the ratio of R, G and B at a certain point of stimulus (adding/removing green will have the biggest effect on Y).

The above tells me gamma will always affect gamut. Since I know you are a very clever person in the matter, either the above is wrong or I'm missing a crucial bit of information?

edit: I'm reading the previous page but it will take a few more reads to really get it...

My gear: Panasonic TH-42PF11EK pro plasma display. -- Iscan Duo video processor -- i1 display 3 colorimeter -- i1 pro 2 spectrometer
Jeroen1000 is online now  
post #213 of 231 Old 12-03-2012, 05:37 AM
AVS Special Member
 
stevekale's Avatar
 
Join Date: Nov 2006
Location: London
Posts: 1,147
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 128 Post(s)
Liked: 24
Quote:
Originally Posted by Light Illusion View Post

We have done a lot of work with ICC's (CMM's), and have found time and again just how inaccurate they can be...
Let's face it, any calibration process that has a 'Use Black Point Compensation' as part of its options isn't really looking like an accurate calibration method.
eek.gif
Seriously, we have been called into facilities time and again to rescue them from issues caused by trying to use ICC profiles for accurate calibration.

For video? I wasn't aware that videos were capable of embedding an ICC profile and that there were CMMs capable of using them (yet). I understood video to be stuck in the intransigent world of hard coding calibration which of course is annoying if you regularly switch between SD and HD media (and your display can't adapt to the changing content). ICC profiling is very effective in graphics (non video) world. As for BPC, it makes immense sense to scale luminance for a difference in black point.

(Of course also we need to distinguish between calibration and profiling.)

Demand greater honesty and transparency in high-end audio

With Theta, Egglestonworks and Krell since 1998!
stevekale is offline  
post #214 of 231 Old 12-03-2012, 06:36 AM
Advanced Member
 
Light Illusion's Avatar
 
Join Date: Aug 2010
Posts: 592
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 35 Post(s)
Liked: 263
Jerone, see my comment above your last post - if profiling is separated from calibration there is no interlinking.

With LightSpace you can do a profile and then set a Rec709 target with 2.2 Gamma, or 2.4 Gamma, and the changes in gamma will not alter the Gamut calibration...

Steve, Adobe attempted to force ICC calibration (and we do indeed understand profiling vs. calibration) on to their video users - it didn't work.

And there is no sense at all for 'Use Black Point Compensation' within an accurate calibration workflow.

Switching between colour spaces is very, very easy.
The professional video world does this all the time.

Steve

Steve Shaw
LIGHT ILLUSION

Light Illusion is offline  
post #215 of 231 Old 12-03-2012, 08:26 AM
AVS Special Member
 
sotti's Avatar
 
Join Date: Aug 2004
Location: Seattle, WA
Posts: 6,692
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 78 Post(s)
Liked: 207
Quote:
Originally Posted by Light Illusion View Post

Jerone, see my comment above your last post - if profiling is separated from calibration there is no interlinking.
With LightSpace you can do a profile and then set a Rec709 target with 2.2 Gamma, or 2.4 Gamma, and the changes in gamma will not alter the Gamut calibration...
Steve, Adobe attempted to force ICC calibration (and we do indeed understand profiling vs. calibration) on to their video users - it didn't work.
And there is no sense at all for 'Use Black Point Compensation' within an accurate calibration workflow.
Switching between colour spaces is very, very easy.
The professional video world does this all the time.
Steve

How do you transfer your profile to a TV's settings if there is no LUT box involved?

Also how do you implement BT.1886 in your software since it does indeed use the black point as part of formula.

Interesting that you condem black offset, yet the only approved standard for HDTV playback includes it.
http://www.itu.int/dms_pubrec/itu-r/rec/bt/R-REC-BT.1886-0-201103-I!!PDF-E.pdf

Adobe RGB also has many provisions for the fact that displays have non-zero black level
http://www.adobe.com/digitalimag/pdfs/AdobeRGB1998.pdf

Joel Barsotti
SpectraCal
CalMAN Lead Developer
sotti is offline  
post #216 of 231 Old 12-03-2012, 08:32 AM
Advanced Member
 
Light Illusion's Avatar
 
Join Date: Aug 2010
Posts: 592
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 35 Post(s)
Liked: 263
Quote:
Originally Posted by sotti View Post

How do you transfer your profile to a TV's settings if there is no LUT box involved?
Also how do you implement BT.1886 in your software since it does indeed use the black point as part of formula.
Interesting that you condem black offset, yet the only approved standard for HDTV playback includes it.
http://www.itu.int/dms_pubrec/itu-r/rec/bt/R-REC-BT.1886-0-201103-I!!PDF-E.pdf
Adobe RGB also has many provisions for the fact that displays have non-zero black level
http://www.adobe.com/digitalimag/pdfs/AdobeRGB1998.pdf

Umm, we are ONLY interested in LUT based calibration... As is says on the website - turn off all the internal rubbish, as it does more harm than good.

We can and do implement ANY colour space as we will have profiled the display in advance... that's kind'a obvious, isn't it?

'Use Black Point Compensation' is not black offset... very, very different animals.

It is impossible to 'add' black to a display, so the calibration needs to be 'float' based. Again that has to be obvious, doesn't it?

wink.gif

Steve

Steve Shaw
LIGHT ILLUSION

Light Illusion is offline  
post #217 of 231 Old 04-17-2013, 07:57 AM - Thread Starter
AVS Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 5,572
Mentioned: 61 Post(s)
Tagged: 0 Thread(s)
Quoted: 788 Post(s)
Liked: 698
Quote:
Originally Posted by Jeroen1000 View Post

sorry for kicking this up when the dust has settled. Why is it indeed, that "gamma" affects the gamut?

Not a direct answer (it's been covered in this thread already) but here are some statistics on what type of errors can be generated across the gamut when your assumption of source gamma does not match display gamma. I also threw in a color space mismatch (601 vs. 709) for comparison. These are measured values (2500 colors) on my display with dE calculated based on the assumptions in the legend.

zoyd is online now  
post #218 of 231 Old 04-20-2013, 12:33 PM
ADU
AVS Special Member
 
ADU's Avatar
 
Join Date: Apr 2003
Posts: 6,802
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 201 Post(s)
Liked: 242

^ Interesting.

 

The 601/709 mismatch is pretty small. I've often wondered if the difference between 601 and 709 would be big enough to catch in a simple visual color-bar decoder test.

 

I wonder what a Rec. 709 vs. Rec. 2020 (wide gamut) mismatch would look like btw? (Pretty substantial I'm guessin.)


ADU
ADU is offline  
post #219 of 231 Old 04-21-2013, 05:52 AM - Thread Starter
AVS Special Member
 
zoyd's Avatar
 
Join Date: Sep 2006
Location: Planet Dog
Posts: 5,572
Mentioned: 61 Post(s)
Tagged: 0 Thread(s)
Quoted: 788 Post(s)
Liked: 698
Quote:
The 601/709 mismatch is pretty small. I've often wondered if the difference between 601 and 709 would be big enough to catch in a simple visual color-bar decoder test.

I can see it in color bars when switching between 709/601 calibrated primaries but picking it out stand-alone is pretty tough.
Quote:
Originally Posted by ADU View Post

I wonder what a Rec. 709 vs. Rec. 2020 (wide gamut) mismatch would look like btw? (Pretty substantial I'm guessin.)

zoyd is online now  
post #220 of 231 Old 04-21-2013, 10:32 AM
AVS Special Member
 
HDTVChallenged's Avatar
 
Join Date: Jan 2002
Posts: 8,851
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 3 Post(s)
Liked: 180
Quote:
Originally Posted by ADU View Post

^ Interesting.

The 601/709 mismatch is pretty small. I've often wondered if the difference between 601 and 709 would be big enough to catch in a simple visual color-bar decoder test.

From my personal experience:

REC601 sources through a REC709 decoder is tolerable.

REC709 sources through a REC601 decoder will drive you completely nuts even on normal programming (blue is boosted significantly.)

Yes, you can detect it through a "simple" color decoder/filter test, if you know what to look for ... smile.gif
HDTVChallenged is offline  
post #221 of 231 Old 04-21-2013, 03:31 PM
AVS Special Member
 
dr1394's Avatar
 
Join Date: May 2002
Location: Mizar 5
Posts: 3,215
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 45 Post(s)
Liked: 24
Quote:
Originally Posted by ADU View Post

The 601/709 mismatch is pretty small. I've often wondered if the difference between 601 and 709 would be big enough to catch in a simple visual color-bar decoder test.

barsmatrix.jpg

Ron

HD MPEG-2 Test Patterns http://www.w6rz.net
dr1394 is offline  
post #222 of 231 Old 04-21-2013, 08:09 PM
ADU
AVS Special Member
 
ADU's Avatar
 
Join Date: Apr 2003
Posts: 6,802
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 201 Post(s)
Liked: 242

^ Nice. Thanks for posting those. They illustrate the difference quite well.


ADU
ADU is offline  
post #223 of 231 Old 04-21-2013, 08:10 PM
ADU
AVS Special Member
 
ADU's Avatar
 
Join Date: Apr 2003
Posts: 6,802
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 201 Post(s)
Liked: 242
Quote:
Originally Posted by zoyd View Post

 

Very cool! Thanks for posting this zoyd.

 

This graphically illustrates why it can be so difficult to get the color on some wide gamut displays close to Rec. 709 spec.


ADU
ADU is offline  
post #224 of 231 Old 06-19-2013, 01:25 PM
 
Koperfild's Avatar
 
Join Date: May 2011
Posts: 37
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
Could anyone tell how does Calman 5 calculate gamma when non-zero black level is applied? It doesn't look like simple black offset (Vin = Vout^gamma + black lvl) because the value is somehow scaled across the dynamic range (for example 1% IRE target is raised by almost 100% value of the black level, while 99% IRE is almost not affected at all)... Is it a standard way of calculating gamma?
JimP likes this.
Koperfild is offline  
post #225 of 231 Old 06-19-2013, 02:38 PM
AVS Special Member
 
sotti's Avatar
 
Join Date: Aug 2004
Location: Seattle, WA
Posts: 6,692
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 78 Post(s)
Liked: 207
Quote:
Originally Posted by Koperfild View Post

Could anyone tell how does Calman 5 calculate gamma when non-zero black level is applied? It doesn't look like simple black offset (Vin = Vout^gamma + black lvl) because the value is somehow scaled across the dynamic range (for example 1% IRE target is raised by almost 100% value of the black level, while 99% IRE is almost not affected at all)... Is it a standard way of calculating gamma?

It depends on the gamma formula. Some (like BT.1886) use the black level as part of the input value for the function. For those that don't we do a simple compression of the range.

Joel Barsotti
SpectraCal
CalMAN Lead Developer
sotti is offline  
post #226 of 231 Old 06-20-2013, 12:17 AM
 
Koperfild's Avatar
 
Join Date: May 2011
Posts: 37
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
Quote:
Originally Posted by sotti View Post

For those that don't we do a simple compression of the range.

Thanks for your answer Joel. Is, what you do, a common procedure, or is it your own solution?

Poynton in his "Gamma Rehabilitation" doc. mentioned following equation: L= (V’ +ε)^gamma, where ε is black offset. On the internet more popular equation is Luminance = contrast * value^gamma + black level. None of them is similar to yours as far as I understand.
Koperfild is offline  
post #227 of 231 Old 06-20-2013, 09:34 AM
AVS Special Member
 
HDTVChallenged's Avatar
 
Join Date: Jan 2002
Posts: 8,851
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 3 Post(s)
Liked: 180
Don't know what Joel's using, but in my spreadsheet I believe I used:

L = Lb + (Lw-Lb)* V^gamma

In other words, apply power law over the reduced range between your measured white and black levels then add the black level in as a linear offset.

BT.1886 does this much more intelligently (not to mention aggressively.)

Or you could just follow power law down as far as you can, and let your brightness control setup handle the compression near black.

There's probably a "right" answer somewhere in all of this, but the truth is probably that there's no way to know what method was used on the mastering/editing monitor, until you see something that looks really wrong ... like "black" holes in the back of peoples heads where there should be blondish hair (e.g.: "Black Swan".)
HDTVChallenged is offline  
post #228 of 231 Old 06-20-2013, 11:03 AM
AVS Special Member
 
sotti's Avatar
 
Join Date: Aug 2004
Location: Seattle, WA
Posts: 6,692
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 78 Post(s)
Liked: 207
Quote:
Originally Posted by HDTVChallenged View Post

There's probably a "right" answer somewhere in all of this, but the truth is probably that there's no way to know what method was used on the mastering/editing monitor

Exactly. Studio's aren't all conforming to any particular standard, which means that there is actually no right answer. Quite a few just run standard 2.2 with no offset for black. Some are starting to use BT.1886.

Joel Barsotti
SpectraCal
CalMAN Lead Developer
sotti is offline  
post #229 of 231 Old 06-20-2013, 11:25 AM
AVS Special Member
 
Elix's Avatar
 
Join Date: Jun 2011
Location: Dungeon, Pillar of Eyes
Posts: 1,358
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 95 Post(s)
Liked: 51
I've read through the whole topic (quite a long time ago, actually) and have been conducting my own experiments on this for a while. Nobody seemed to ask some questions which seem quite obvious to me. When pursuing for ultimate shadow detail what level would you want to be able to see in the actual video content? Is it 1 or maybe 2 (on a 0-255 scale)? And when we say 'see' how distinctly that would be? Must we use this or that pattern to judge the visibility of low level signals? I believe that's practical questions.
A theoretical question will be: "Will every display calibrated to BT.1886 be able to distinctly reproduce these low level signals?".
Elix is offline  
post #230 of 231 Old 06-20-2013, 10:15 PM
AVS Special Member
 
HDTVChallenged's Avatar
 
Join Date: Jan 2002
Posts: 8,851
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 3 Post(s)
Liked: 180
Quote:
Originally Posted by Elix View Post

I've read through the whole topic (quite a long time ago, actually) and have been conducting my own experiments on this for a while. Nobody seemed to ask some questions which seem quite obvious to me. When pursuing for ultimate shadow detail what level would you want to be able to see in the actual video content? Is it 1 or maybe 2 (on a 0-255 scale)? And when we say 'see' how distinctly that would be?

Perhaps the reason that nobody asked that question is this: You should be able to see anything between 8 bit digital level 16 and 255 for video content. IOW: 17 to 254. By definition, black =16 and reference-white = 235. 236 through 254 is the overshoot (aka superwhite) range.
Quote:
Must we use this or that pattern to judge the visibility of low level signals? I believe that's practical questions.

????
Quote:
A theoretical question will be: "Will every display calibrated to BT.1886 be able to distinctly reproduce these low level signals?".

Yep ... assuming the display can actually pull off BT.1886.
HDTVChallenged is offline  
post #231 of 231 Old Today, 12:21 PM
Senior Member
 
fluxo's Avatar
 
Join Date: Oct 2012
Posts: 323
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 121 Post(s)
Liked: 101
Quote:
Originally Posted by zoyd View Post
ITU/EBU recommends that the end-to-end gamma of the video chain should be between 1.1 - 1.2 so we have a clearly defined target. It's commonly stated that video RGB is encoded with an exponent of 1/2.22 but the full function is:

v'=4.5*v for v < 0.018
v'=1.099*v^(.45)-0.099 for v >= 0.018

So what we want our displays to do is invert the function above and then apply a modest power of between 1.1 - 1.2.

What do we do in practice? We feed the display a linear input of 10% steps and (if we have a 10pt gamma control) try to get a nice flat fit to the equation v=v'^(gamma=2.2, 2.3, etc.) This completely ignores the fact that the real video signals we will eventually watch are encoded using the equation above.

So what are the consequences of the simplified power law calibration approach?



This is a plot of the result of feeding a linear signal which has been encoded using the BT.709/601 transfer function into a display with a simple power law response.
Which gamma formula are you using for that plot, please?

This is my plot, for various exponents, based on a log-log gradient definition for point gamma:



Thank you.
fluxo is online now  
Sponsored Links
Advertisement
 
Reply Display Calibration

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off