2.5 is display gamma, NOT 2.2 - Page 4 - AVS Forum
Forum Jump: 
Reply
 
Thread Tools
post #91 of 161 Old 10-07-2008, 12:53 PM
 
tbrunet's Avatar
 
Join Date: Jul 2005
Location: Savannah, GA
Posts: 2,750
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 12
Like quantization error, banding or contouring artifacts.
tbrunet is offline  
Sponsored Links
Advertisement
 
post #92 of 161 Old 10-07-2008, 12:58 PM
Senior Member
 
creatine64's Avatar
 
Join Date: Aug 2006
Posts: 295
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by Edo Gálvez View Post

Don't feel bad man, I think we IT guys are just hard wired against analog stuff and solving problems with other problems. XD

LOL

very true
creatine64 is offline  
post #93 of 161 Old 10-07-2008, 04:50 PM
Senior Member
 
delphiplasma's Avatar
 
Join Date: Mar 2008
Location: UK
Posts: 246
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 6 Post(s)
Liked: 23
Can't believe this discussion about gamma still goes on.

2.2 is mainly for multimedia computer use.

A professional cathode driven CRT monitor, as derived by the EBU specs, should be around 2.35-2.4

However, an old grid driven CRT of the sixties and seventies had an EBU spec of around 2.8!

I'm not sure where 2.5 came from?
delphiplasma is offline  
post #94 of 161 Old 10-07-2008, 05:28 PM
Senior Member
 
Edo Glvez's Avatar
 
Join Date: May 2008
Posts: 327
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 17
Quote:
Originally Posted by delphiplasma View Post

Can't believe this discussion about gamma still goes on.

2.2 is mainly for multimedia computer use.

A professional cathode driven CRT monitor, as derived by the EBU specs, should be around 2.35-2.4

However, an old grid driven CRT of the sixties and seventies had an EBU spec of around 2.8!

I'm not sure where 2.5 came from?

dewddddddd! why didn't you tell us sooner?
Elix likes this.
Edo Glvez is offline  
post #95 of 161 Old 10-07-2008, 05:59 PM
AVS Special Member
 
lcaillo's Avatar
 
Join Date: Feb 2002
Location: Gainesville, FL
Posts: 3,459
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by delphiplasma View Post

Can't believe this discussion about gamma still goes on.

2.2 is mainly for multimedia computer use.

A professional cathode driven CRT monitor, as derived by the EBU specs, should be around 2.35-2.4

However, an old grid driven CRT of the sixties and seventies had an EBU spec of around 2.8!

I'm not sure where 2.5 came from?

I agree with the first statement, but for a different reason. The best reason that it makes no sense is that a single number cannot adequately describe the response and that so many displays do not allow sufficient control. Let's follow Bill's suggestion and stop using a single value to describe luma response.

Yes, calibration is important...every user should be calibrated.

Need electronics repair? A great place to start looking for a shop in your area: http://www.tvrepairpros.com/
lcaillo is offline  
post #96 of 161 Old 10-08-2008, 06:06 AM
Member
 
WolfyA's Avatar
 
Join Date: Jul 2007
Posts: 192
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
@ tbrunet, alluringreality: Thanks for your replies. Things are starting to take root.
WolfyA is offline  
post #97 of 161 Old 10-08-2008, 06:15 AM
 
tbrunet's Avatar
 
Join Date: Jul 2005
Location: Savannah, GA
Posts: 2,750
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 12
Quote:
Originally Posted by WolfyA View Post

@ tbrunet, alluringreality: Thanks for your replies. Things are starting to take root.

No problem Wolfy, but delphiplasma post regarding industry standard reference CRT's is misleading..

http://www.displaymate.com/ShootOut_Part_2.htm

“Television, DVD, Web and computer based photographic content are generally color balanced on professional CRT studio monitors that are electronically adjusted to have a standard Gamma of 2.20, so you’ll get the most accurate images if your display also has this value”

The Gamma for the Sony CRT agrees perfectly with the 2.20 standard value. (CRT monitors from Ikegami, another major brand of professional studio monitors, also have a Gamma of 2.20 according to their Director of Engineering.)
Quote:
Dr. Raymond M. Soneira

CRT Sony PVM-20L5 (Gamma Measurements)
The Sony CRT had a ruler flat 2.20 response all the way down to 1%

tbrunet is offline  
post #98 of 161 Old 10-08-2008, 08:19 AM
Senior Member
 
Edo Glvez's Avatar
 
Join Date: May 2008
Posts: 327
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 17
Quote:

That link is awesome
Edo Glvez is offline  
post #99 of 161 Old 10-08-2008, 04:21 PM
Senior Member
 
delphiplasma's Avatar
 
Join Date: Mar 2008
Location: UK
Posts: 246
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 6 Post(s)
Liked: 23
Hi Guys,

Take a look at the technical documents published on the EBU website. They have technical documents for the requirements of 'grade1' broadcast studio monitors. The tolerance for gamma is between 2.35 - 2.4.

The BBC adheres to these standards. Before I saw those documents I was under the believe that 2.2 was the best to conform to, after seeing the display shoot out.

It would seem, however, that 2.2 is probably a good target to shoot for if the ambient room lighting is fairly high. The EBU state that 2.35 is for a dimly lit room.

Or is there some international mismatch between systems?
delphiplasma is offline  
post #100 of 161 Old 10-08-2008, 06:29 PM
Senior Member
 
jdoostil's Avatar
 
Join Date: Oct 2006
Posts: 280
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 4 Post(s)
Liked: 21
So lets assume black level, white level, and greyscale are set correctly, and the TV does not have gamma controls. If the avg gamma lands at 1.9 (just an example) is that just a limitation of that particular set, or can you do some reworking and achieve 2.2 (without crushing blacks)?
jdoostil is offline  
post #101 of 161 Old 10-08-2008, 07:33 PM
AVS Club Gold
 
Bear5k's Avatar
 
Join Date: Feb 2006
Location: Houston, TX
Posts: 2,239
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by jdoostil View Post

So lets assume black level, white level, and greyscale are set correctly, and the TV does not have gamma controls. If the avg gamma lands at 1.9 (just an example) is that just a limitation of that particular set, or can you do some reworking and achieve 2.2 (without crushing blacks)?

Start reading here at post #22:
http://www.avsforum.com/avs-vb/showthread.php?t=1039169

Color accuracy evangelist and CalMAN insider
Bear5k is offline  
post #102 of 161 Old 10-10-2008, 09:21 AM
Member
 
eiren's Avatar
 
Join Date: Jul 2008
Posts: 66
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Do you always need to aim for a flat gamma response?

For instance, could you achieve a bell shaped gamma curve that has the bottom and top end IRE at around the 2.2 level and then have the 20-80 range tracking along the 2.5 gamma?

Would this then give the extra pop in the midrange, and then keep dark detail looking good for dim viewing conditions... or is having a gamma that doesn't track uniformly across the entire IRE range going to cause big problems?
eiren is offline  
post #103 of 161 Old 10-16-2008, 02:10 PM
Member
 
Ciccio's Avatar
 
Join Date: Mar 2007
Posts: 47
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Quote:
Originally Posted by eiren View Post

Do you always need to aim for a flat gamma response?

For instance, could you achieve a bell shaped gamma curve that has the bottom and top end IRE at around the 2.2 level and then have the 20-80 range tracking along the 2.5 gamma?

Would this then give the extra pop in the midrange, and then keep dark detail looking good for dim viewing conditions... or is having a gamma that doesn't track uniformly across the entire IRE range going to cause big problems?

Very good questions ... I'd be interested in hearing answers to these as well.
Ciccio is offline  
post #104 of 161 Old 10-16-2008, 02:57 PM
Member
 
eiren's Avatar
 
Join Date: Jul 2008
Posts: 66
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
I've answered it myself now!

I calibrated my new Pioneer screen so that 0-30 IRE were set to 2.22 gamma, and the rest 40-100 IRE were set to 2.5 gamma.

The results are absolutely stunning! Real punch to the picture, and all dark detail retained.

Gordon stated elsewhere he uses this on calibrations too (hopefully he wont mind me quoting him):

Quote:
Originally Posted by Gordon @ Convergent AV View Post

Eiren,

You have done exactly what I do with Lumagens parametric gamma capability on displays that have extreme contrast and can lose/crush dark detail when you try for lowest black levels....ie 9G's, CRT projectors and a couple of digital PJ's......so, although it might not be completely to standard I've only ever had letters of thanks from the customers I've done it for....

eiren is offline  
post #105 of 161 Old 10-17-2008, 02:31 PM
AVS Special Member
 
daMaster's Avatar
 
Join Date: May 2004
Location: Montreal, Canada
Posts: 1,407
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
Lots of technical discussions in here, but I'd like to get the crux of the issue and what it all means for us DIY calibrators:
- Should I be changing the gamma setting in ColorHCFR to 2.5 from the default 2.22?
- My average gamma is currently 2.33 and my black levels look just right, and could probably do better going a little lower. If I recalibrate to a gamma of 2.5, then everything will be very washed out and my blacks will be grey.
- For those with displays where gamma can't be adjusted, we are forced to play with contrast-to-brightness ratios on our displays. That is the only way to adjust gamma for us, right?
daMaster is offline  
post #106 of 161 Old 10-17-2008, 03:59 PM
Senior Member
 
scooper750's Avatar
 
Join Date: May 2005
Location: Jacksonville, FL
Posts: 245
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Actually, just the opposite. If the gamma is too low then the blacks will tend to look gray and washed out. High gamma will look darker and may risk black crush. All depends on the set. Higher gamma is great for dark environments (very well light controlled). Your lower gammas are better for well lit environments. It all boils down to the capabilities of the set and your viewing environment.
Quote:
Originally Posted by daMaster View Post

...
- My average gamma is currently 2.33 and my black levels look just right, and could probably do better going a little lower. If I recalibrate to a gamma of 2.5, then everything will be very washed out and my blacks will be grey....


Stephen Cooper
THX Certified Video Calibrator
Digital Video Calibrations LLC
http://www.digitalvideocalibrations.com
scooper750 is offline  
post #107 of 161 Old 12-29-2008, 08:57 AM
Advanced Member
 
andrewfee's Avatar
 
Join Date: Mar 2004
Posts: 785
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
I don't mean to bump an old topic, but it seemed like this belonged here.

After having a Pioneer KRP-500M last week and calibrating it to a "near-perfect" 2.2 Gamma in Pure mode (report) I found that the low-level shadow detail was somewhat lacking and being swallowed up by black when brightness was set "correctly".

I found myself wanting to raise brightness to around +4 rather than the +1 test patterns showed it should be. The shadow detail being revealed was all above black information, but it was just being "hidden" by the display.

After being told that if gamma was reading 2.2 and brightness was set to the point where 1% and 2% grey were visible, the display was showing exactly how things should look I decided I must be doing something wrong, so I decided to do some more reading up on gamma.


Now, I knew that the BT.709 curve was not a simple power function, but everything I have seen so far suggests that you should not be using this to calibrate your display. Certainly when using 2.2 Gamma in CalMAN using the "ITU/EBU Standard" option looks terrible.


I've spent a good portion of time yesterday and today going over forum posts and technical papers, reading up on gamma. There's certainly a lot of confusion/disagreement of what should be used out there!



From looking over things, I think one of the biggest mistakes seems to be the simplification of gamma to a simple power function. Or more specifically, the way people are doing so.

ITU-R BT.709-5 specifies the opto-electronic conversion (gamma curve) to be:



And many people are just simplifying this to be a power function of 1/0.45, or 2.22


Now I've done the math, and have verified that it's done properly, as the numbers match up to CalMAN's target luminance numbers with the same peak white, unless we both have it wrong. (well, my results were slightly different as I used 1/0.45 rather than 2.22 in my calculation)


The BT.709 curve is actually much closer matched to a 1.96 power function (1/0.51) rather than 2.22 as people commonly state, but not only that, it is brighter at the low end:



Note: I'm only going down to 10% grey here, as the linear section gets involved below there and (apparently) that should not be reproduced at the display.

Now this is the camera gamma, rather than what should be reproduced at the display.

So what should be reproduced at the display? Well many people seem to have stated that a CRT's response is a 2.5 power function (Poynton states 2.4) and that should be what your target is.

From ARIB TR-B28 1.1, it would appear that this is not the case with pro monitors. They have measured a Sony BVM-D24E1WJ, and not only is its response not equivalent to a 2.5 power function (it's closer to 2.4) it brightens up at the low end just like the BT.709 curve, so it is not represented simply by a power function either: http://sr-388.net/images/av/gamma/ARIB.pdf

So that's hopefully a 2.5 power function dispelled.


If that isn't it, what then? Well, the end-to-end gamma should not be linear (1.0) and I've seen recommendations between 1.125 and 1.25 mentioned here.

EBU-Tech 3320 specifically recommends an end-to-end gamma of 1.2 when viewing in a dark room, and someone here also mentioned that Poynton suggests an end-to-end gamma in the range of 1.1-1.2 depending on the viewing environment.

That EBU document also specifically recommends a gamma of 2.35 as being ideal, with a tolerance of ±0.15.

So how did they arrive at 2.35? Well, it's certainly not based off BT.709 being 2.22. If you take the camera gamma to be 1.96 rather than 2.22, you arrive at 2.35 with an end-to-end of 1.2.

So, is a 2.35 power function correct then? Subjectively, I would have to say no. Midtones look good, but the low end is too dark.


Surely though, you should be using the camera gamma as a reference though, rather than a simplified power function. Perhaps the linear section is not intended to be reproduced at the display, but the rest of the curve is different as well. It is not simply a linear section followed by a 2.22 power function.

If you were to use an end-to-end gamma of 1.2, that would mean using 0.375 in the BT.709 function rather than 0.45. If you do this, you end up with a curve that, when approximated, ends up very close to a 2.35 power function, however, the low end 25% grey and below is actually brighter than even the "recommended" 2.22 power function.



It's not particularly clear at the low-end, so here are the numbers using 80cd/m² for white:



If you were to use CalMAN, it would be equivalent to using a target exponent of 2.67 with the "ITU/EBU Standard" formula, though you should not use the 5% target in CalMAN as it does appear to include the linear section near black. Assuming, of course, that the linear section is not intended to be reproduced on the display. (does anyone have a source for that?)


Am I misunderstanding things, or should this be the target for calibration with Rec.709 sources, rather than a 2.22, 2.35 or other power function, assuming a target of 1.2 end-to-end for viewing in a darkened room.
andrewfee is offline  
post #108 of 161 Old 12-29-2008, 12:35 PM
AVS Special Member
 
sotti's Avatar
 
Join Date: Aug 2004
Location: Seattle, WA
Posts: 6,639
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 43 Post(s)
Liked: 175
Wow, this is exactly where my head is right now as well. Since I'm building my PC gamma and greyscale app.

I just changed it up to use camera gamma, and I'd have to agree it doesn't look right at all.

I went to poynton's website and found this PDF
http://www.poynton.com/PDFs/GammaFAQ.pdf

In it, he says the idealized monitor would be an inverse of the 709 encode function (under section 6). So I coded that into my program and it doesn't work well, and like you said the gamma is more like 1.9 versus a power function and much brighter on the lowend.

I calibrated my display the best I can, it doesn't have much gamma control. Then I'm using my app on my PC to alter my PC's LUT so I have fine control of up to 256 points if I wanted it (usually only using 15-25).

I don't have any answers, just validating that your question is valid.

Joel Barsotti
SpectraCal
CalMAN Lead Developer
sotti is offline  
post #109 of 161 Old 12-30-2008, 03:07 AM
Advanced Member
 
andrewfee's Avatar
 
Join Date: Mar 2004
Posts: 785
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
Well from what I've read, it does look like the display should be an inverse of the BT.709 curve, rather than it being simplified to a power function.

However, you don't want it to be exactly the inverse, as you found out. It needs to be offset to provide an end-to-end gamma greater than 1.

E.g. 1.96 / 1.96 = 1.0
1.96 * 1.2 = 2.35

So rather than using a 1.96 power function at the display to match the BT.709 transfer function, you should use 2.35 at the display. (simplified to power functions to illustrate the point from what I have seen, you should be using the proper BT.709 curve with 0.375 in the equation rather than 0.45)


In addition, I'm not sure where the notion that the linear segment should not be displayed comes from. To quote Poynton:

Real monitors are not as exact as this equation suggests, and have no linear segment


That simply suggests to me that the monitors available in 1998 were not capable of reproducing the inverse of the BT.709 curve exactly. Modern displays do not respond like CRTs and we've come a long way with digital processing now, and have devices that offer 21-point, 10-bit gamma control, so (in theory at least) they should be capable of properly reproducing the inverse of the BT.709 curve or at the very least, something much closer than a 2.22 or 2.35 power function.
andrewfee is offline  
post #110 of 161 Old 12-30-2008, 06:41 AM
AVS Special Member
 
alluringreality's Avatar
 
Join Date: Jul 2006
Posts: 3,203
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 46 Post(s)
Liked: 39
Quote:
Originally Posted by andrewfee View Post

from what I have seen, you should be using the proper BT.709 curve with 0.375 in the equation rather than 0.45

From what I've read, no matter what you have to make some assumptions. One assumption I would make is that when the current specifications were written CRT was still a common consumer display. I haven't seen that consumer CRT is capable of this sort of detailed response, so I really don't get where any basis for this conclusion comes from.

I also find the discussion of CRT having one gamma response sort of odd, because I've seen that increasing brightness will lower CRT gamma. CRT gamma also seems to fluctuate depending on APL.
alluringreality is offline  
post #111 of 161 Old 12-30-2008, 09:22 AM
 
tbrunet's Avatar
 
Join Date: Jul 2005
Location: Savannah, GA
Posts: 2,750
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 12
Quote:
Originally Posted by andrewfee View Post

..That simply suggests to me that the monitors available in 1998 were not capable of reproducing the inverse of the BT.709 curve exactly.

Actually the linear segment (near black region) is grafted onto the .45 power function to reduce real world noise artifacts. By reducing gain there it effectively minimizes sensor noise, be it tube, CCD, ect. So a reciprocal display response i.e. adding near infinite gain in this near black region would simply negate and undo the intended de-emphasis.
tbrunet is offline  
post #112 of 161 Old 12-30-2008, 09:57 AM
Advanced Member
 
andrewfee's Avatar
 
Join Date: Mar 2004
Posts: 785
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
Quote:
Originally Posted by tbrunet View Post

Actually the linear segment (near black region) is grafted onto the .45 power function to reduce real world noise artifacts. By reducing gain there it effectively minimizes sensor noise, be it tube, CCD, ect. So a reciprocal display response i.e. adding near infinite gain in this near black region would simply negate and undo the intended de-emphasis.

I had a feeling that might be the case, but if you are using an end-to-end greater than 1, you are already reducing the appearance of this.

And again, it's not simply a 0.45 power function, there is an offset of 0.099 included as well as the linear section, which brings the camera gamma closer to an 0.51 power function.
andrewfee is offline  
post #113 of 161 Old 12-30-2008, 10:33 AM
AVS Special Member
 
sotti's Avatar
 
Join Date: Aug 2004
Location: Seattle, WA
Posts: 6,639
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 43 Post(s)
Liked: 175
well the other thing that crosses my mind in this whole discussion is that by modifing the power curve and toe slope if you .

So the basic camera gamma function looks like:
if (stim% <= .081)
{
luma = sitm% / (.45 * 10)
}
else
{
luma = ((stim% + .099)/1.099)^(1/.45)
}

if you change the .45 to something else it would effect the toe slope and the power curve, meaning that you also need to find the new intersection of these two points, so the cross over from toe slope to the power function is smooth.

I also have no idea what the .099 offset is for, is it to align the to slope crossover point, in which case should we be examining that offset as well? Or is it to alter the shape of the power curve to smooth the lower stim% range to come out faster.

My gut feeling is to just go back to the basic power function, since this was the expected output when the camera gamma function was created.

All I know is that my eyes do not like the basic camera gamma function, and I don't know enough about it to start modifying the numbers inside it willy nilly.

Joel Barsotti
SpectraCal
CalMAN Lead Developer
sotti is offline  
post #114 of 161 Old 12-30-2008, 02:56 PM
 
tbrunet's Avatar
 
Join Date: Jul 2005
Location: Savannah, GA
Posts: 2,750
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 12
Quote:
Originally Posted by andrewfee View Post

And again, it's not simply a 0.45 power function, there is an offset of 0.099 included as well as the linear section, which brings the camera gamma closer to an 0.51 power function.

http://books.google.com/books?id=NsR...um=6&ct=result

Figure 2.6
Quote:


As a true power function requires infinite gain near black, a linear segment is substituted. The result is that noise picked up by the video signal has less effect on dark areas than bright areas. After a gamma function at the display, noise at near-black levels is compressed with respect to noise near white levels.

In practice the system is not rendered perfectly linear by gamma correction and a slight overall effect is usually retained in order to further reduce the effect of noise in darker parts of the picture. A gamma correction factor of .45 may be used to achieve this effect.

Gamma is the average slope i.e. the average slope of a line plotted on a log vs log graph. It also represents/defines (single gamma number) the most important "midtone" region, which also contains the bulk of the image content.
tbrunet is offline  
post #115 of 161 Old 12-30-2008, 04:14 PM
Advanced Member
 
andrewfee's Avatar
 
Join Date: Mar 2004
Posts: 785
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
Quote:
Originally Posted by tbrunet View Post

http://books.google.com/books?id=NsR...um=6&ct=result

Figure 2.6
Gamma is the average slope i.e. the average slope of a line plotted on a log vs log graph. It also represents/defines (single gamma number) the most important "midtone" region, which also contains the bulk of the image content.

It seems fairly widely accepted that a power curve of 1/0.51 (1.96) is the closest match to the BT.709 curve. Using an average gamma is not a good representation of it, however.

Perhaps throwing the numbers into HCFR better illustrates my point:


Red is a 2.22 power function (1/0.45)
Yellow is the BT.709 curve
Cyan is the average gamma that HCFR calculated. (1.89)

Interestingly, 50% grey has a gamma of 1.95, which is the most important number, according to that quote.

Given that the BT.709 curve is considered to be best represented by a 0.51 power function, that would mean you should use a 2.35 (0.425) power function at the display when viewing in a darkened room. (end-to-end of 1.2) If you feel that a power function is suitable that is.


Thanks for the link though, I'll have a look over it.

Quote:
Originally Posted by sotti View Post

well the other thing that crosses my mind in this whole discussion is that by modifing the power curve and toe slope if you .

So the basic camera gamma function looks like:
if (stim% <= .081)
{
luma = sitm% / (.45 * 10)
}
else
{
luma = ((stim% + .099)/1.099)^(1/.45)
}

if you change the .45 to something else it would effect the toe slope and the power curve, meaning that you also need to find the new intersection of these two points, so the cross over from toe slope to the power function is smooth.

I also have no idea what the .099 offset is for, is it to align the to slope crossover point, in which case should we be examining that offset as well? Or is it to alter the shape of the power curve to smooth the lower stim% range to come out faster.

My gut feeling is to just go back to the basic power function, since this was the expected output when the camera gamma function was created.

All I know is that my eyes do not like the basic camera gamma function, and I don't know enough about it to start modifying the numbers inside it willy nilly.

You're right, I shouldn't have disregarded the linear section before. I didn't realise just how much that changed things.

I'm still (somewhat) convinced that you should be using the inverse of the BT.709 curve at the display, but with an end-to-end of 1.2.

I'm just not sure how you end up with that yet.
andrewfee is offline  
post #116 of 161 Old 12-31-2008, 08:21 AM
 
tbrunet's Avatar
 
Join Date: Jul 2005
Location: Savannah, GA
Posts: 2,750
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 12
Quote:
Originally Posted by andrewfee View Post

Interestingly, 50% grey has a gamma of 1.95, which is the most important number, according to that quote.

..Given that the BT.709 curve is considered to be best represented by a 0.51 power function...

What you quoted were my words btw..and they seem to align with Panavision here:

http://www.panavision.com.au/PDFs/In..._Explained.pdf
Perceptually Uniform Grayscale Coding In the Panavision Genesis Electronic Cinematography System

Quote:
Empirically an inverse value of 1/2.2 or 0.45 has been standardized as the “average slope” of the most recent incarnation of this video signal definition as incorporated in ITU-R BT.709 (See page 5)

This basic gamma function embodied in this standard is almost identical to the original 1953NTSC standard that required a camera output with the slope of a power function of .45

FWIW gamma is computed by measuring the “slope” of a log-log plot of the transfer function, comparing the log of the measured screen luminance to the log of the input signal.
tbrunet is offline  
post #117 of 161 Old 12-31-2008, 08:45 AM
AVS Special Member
 
jimwhite's Avatar
 
Join Date: Sep 2001
Location: Saint Petersburg, FL USA
Posts: 5,372
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
so we're back to 2.2 ?


Jim White
St. Petersburg, FL
jimwhite is offline  
post #118 of 161 Old 12-31-2008, 08:49 AM
 
tbrunet's Avatar
 
Join Date: Jul 2005
Location: Savannah, GA
Posts: 2,750
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 12
Quote:
Originally Posted by jimwhite View Post

so we're back to 2.2 ?


According to Dr. Raymond Soneira it never left 2.2
Quote:
http://www.displaymate.com/ShootOut_Part_2.htm
Sony PVM-20L5
Measured Gamma 2.20

tbrunet is offline  
post #119 of 161 Old 12-31-2008, 10:55 AM
Advanced Member
 
andrewfee's Avatar
 
Join Date: Mar 2004
Posts: 785
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 11
Quote:
Originally Posted by tbrunet View Post

According to Dr. Raymond Soneira it never left 2.2

So how is it that the ARIB measures a Sony BVM-D24E1WJ to have a response that doesn't match up to either 2.2 or any simple power function?

Aren't the BVM class of monitors supposed to be more accurate than the PVM ones?


According to EBU-Tech 3320:

BBC R&D Report RD 1991/6, Methods of measuring and calculating display transfer characteristics (gamma)' by Alan Roberts, indicates a method of performing such measurements, and has yielded results which indicate that

the gamma of a grade 1 CRT monitor is typically in the region of 2.3 to 2.4.



The luminance gamma characteristic (electro-optical transfer function) of the screen should be equivalent to those of a reference CRT with the rendering intent (dim-surround) expected of a TV system. It is believed that a nominal value of 2.35 is appropriate.



It has been found that the end-to-end or system gamma for images captured in nominal daylight conditions, adapted for the dim-surround consumer-viewing environment is approximately 1.2, i.e. definitely not linear.

The system gamma can be expressed as:

System gamma = camera encoding gamma (OETF) x display gamma (EOTF)

It has been found from measurement techniques, progressively refined over several decades, that a correctly designed CRT display has an EOTF gamma of approximately 2.35. This is part of the immovable legacy effect of the CRT.

Therefore our system gamma equation is rewritten as
System gamma = 1.2 = OETF gamma x 2.35
It would seem to me that the display's gamma should be an inverse of the camera gamma, with an end-to-end of 1.2. (OETF = BT.709 curve/camera gamma)
andrewfee is offline  
post #120 of 161 Old 12-31-2008, 11:29 AM
 
tbrunet's Avatar
 
Join Date: Jul 2005
Location: Savannah, GA
Posts: 2,750
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 12
All I know is Sony and Ikegami intentionally designed their reference CRTs to have a ruler flat 2.20 response, which btw is a result of custom electronic processing. Testing carried out by Hewlett-Packard, MS and Sony on VGA computer monitors from many brands has shown the average CRT gamma to be indeed 2.2, with a standard deviation of about 0.2.

Quote:


http://download.microsoft.com/downlo...rCaseStudy.doc

• Sony provided studies that proved optimally calibrated CRTs in their native state are 2.2 gamma and have HDTV primaries. [Katoh and Deguchi, "Reconsideration of CRT Monitor Characteristics"]

tbrunet is offline  
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