SDR Display EOTF (Gamma 2.2/2.4 or BT.1886).

[James Freeman]

First step download HCFR from here: https://sourceforge.net/projects/hcfr/

This software will let you plot EOTF curves and change their parameters like minimum light level which will plot the curve differently according to this.
HCFR also lets you change the Black output/input offset or the shadow compression level.


Peak luminance is always 100 nit, so set Override Black to 0.1 nit to have a 1000:1 typical contrast ratio display (100/0.1 = 1000).
You can play with it and just click Apply to see the new curve.
Lower gamma value = brighter shade.

Lets have a look:

BT1886:
Set to gamma to 0 and input offset to 100%, to use the BT1886 equation.
If you set Override Black to 0 nit, the gamma will be 2.4 spot on across whole luminance range.
http://www.mediafire.com/convkey/9fd7/j13i49to1x2tqjqzg.jpg

Black Compensated Gamma 2.2:
0 input offset for maximum black compression, 100 for less compression but less true gamma curve below 50% input.
http://www.mediafire.com/convkey/c830/7prvn021iqpc6kbzg.jpg

[James Freeman]

Clearly BT1886 is nowhere near 2.4 with a 1000:1 display.
It also doesn't follow 2.2 or 2.4 above or below 50% input.

SDR movies are graded on a typical 1000:1 LCD studio monitor calibrated to black compensated gamma 2.2.
If the grading monitor is an OLED, the minimum black level will be set to 0.1 or 0.05 to have contrast ration of 1000:1 or 2000:1 according to ITU specification of Grade A monitor.
If you watch this content on a BT1886 calibrated display you WILL have wrong shadows and more importantly wrong colors.

I don't think anyone can tell what input/output offset % the studios use but it's definitely not 100% input.
I personally use 70% output offset (30% input offset) Gamma 2.2 when I calibrate with DisplayCAL and ArgyllCMS because it resembles to my native monitor gamma curve and because it does not compress the blacks as much as 100 output offset.
Don't confuse these, HCFR uses input offset whether DisplayCAL uses output offset, same thing different name and proportional to one another.

[e-t172]

Clearly BT1886 is nowhere near 2.4 with a 1000:1 display.

Nobody said it was. Standard BT.1886 has a technical gamma of 2.4, in other words, 2.4 is used as the exponent. However, because a BT.1886 curve (in general) is not a pure power curve, the effective gamma ("average" gamma, or best fit for a pure power curve) is closer to 2.2 when a display with a non-perfect black point is used. This is completely normal and is working as intended - the whole point of BT.1886 is to compensate for such displays by lowering effective gamma.

It also doesn't follow 2.2 or 2.4 above or below 50% input.

Sure. BT.1886 is not a pure power curve. So what? Obviously the people who wrote BT.1886 standardized on that shape on purpose, otherwise they would just have mandated a pure power curve. Are you claiming that they're wrong, and if so, do you have specific evidence to back up your case?

SDR movies are graded on a typical 1000:1 LCD studio monitor calibrated to black compensated gamma 2.2.

They're not supposed to. They're supposed to be calibrated to BT.1886. That's kind of the point of having a standard.

If the grading monitor is an OLED, the minimum black level will be set to 0.1 or 0.05 to have contrast ration of 1000:1 or 2000:1 according to ITU specification of Grade A monitor.
If you watch this content on a BT1886 calibrated display you WILL have wrong shadows and more importantly wrong colors.

You're not taking viewing conditions into account, which are extremely important when determining the correct viewing gamma. If you take a movie that was graded in a dim room (which AFAIK is the norm), and watch it in a pitch-black room, you need to increase gamma to compensate. This is why the ArgyllCMS docs (http://www.argyllcms.com/doc/Scenarios.html#TV1) recommend an effective gamma of 2.4 for dark conditions (higher than BT.1886, at least for non-perfect displays), and 2.2 for dim conditions.

To be fair, color experts would probably raise eyebrows at this debate and would recommend that the video industry ditch this whole gamma business and switch to more modern and robust viewing condition management workflows, such as CIECAM02, along with standards that actually define viewing conditions in the first place (there is no mention of viewing conditions in BT.1886). That would quickly put an end to the pointless debates and speculation. I think that's asking for ponies and unicorns at this point, though. After all, what's the point of using proper color management frameworks when we're having so much fun arguing to death about the shape and parameters of a gamma curve?

[iSeries]

From my own personal experience, I have a TV that has awful blacks and a contrast ratio of less than 1000:1, and 1886 looks terrible on it. Washed out and lifeless. I settled on 2.2, with a black offset (92% output offset in displaycal) so blacks aren't crushed. Looks good, waaaaaay better than 1886. I wonder if those who sing the praises of 1886 have seen just how bad it looks on TVs with poor blacks/contrast ratios.

[James Freeman]

iSeries, this what I'm talking about.
On a typical LCD bt1886 looks like crap, which is funny because it was made for low contrast displays.
Not a single doubt that a black compensated gamma 2.2 looks better even if it's not a standard.
I use 70% output offset gamma 2.2.

e-t172, do you have any sub 1000:1 display calibrated to 1886?
You are wrong about studio grading monitors being calibrated to bt1886, they are not and will never be.

[JarrettH]

From my own personal experience, I have a TV that has awful blacks and a contrast ratio of less than 1000:1, and 1886 looks terrible on it. Washed out and lifeless. I settled on 2.2, with a black offset (92% output offset in displaycal) so blacks aren't crushed. Looks good, waaaaaay better than 1886. I wonder if those who sing the praises of 1886 have seen just how bad it looks on TVs with poor blacks/contrast ratios.

Yep, I agree

I wouldn't even call it that blacks are crushed, just not reproduced; and for it to look good, blacks need to start later

[baii]

I have read some suggest that it is fine to let black crush a bit, ie. One may not necessarily able to differentiate step 16,17,18. I forgot about the exact argument but I have since follow this when picking my black compensation value and I am satisfied. Bt1886 may be technical right but does not look right for alot of contents. I think I have had content(the hobbit?) that look good/right on bt.1886.

Bt.1886 on a oled is flat 2.2 iirc.

I don't think bt1886 arguments have much to do with viewing condition, it is the fact that it is trying to compensate for different black level and lifting the shadow like mad.(when rest of the world aren't doing the"standard".)

Sent from my XT1575 using Tapatalk

[James Freeman]

Bt.1886 on a oled is flat 2.2 iirc.
I don't think bt1886 arguments have much to do with viewing condition, it is the fact that it is trying to compensate for different black level and lifting the shadow like mad.(when rest of the world aren't doing the"standard".)


Professional OLED grading monitors like the Sony PVM/BVM series have minimum output level setting that usually set to 0.05 or 0.1 nit resulting in a contrast ratio of 1000:1 or 2000:1.
With infinite black level BT1886 will result in exactly Gamma 2.4 power curve.

Watching content that was graded on a BT1886 calibrated studio monitor on a black compensated gamma display (most of them) will crush the black even more.
When using bt1886 calibrated grading monitor the colorist will pull the shadows down when grading till it looks good to him, this content will look as it should ONLY on a bt1886 calibrated screen.
Since most movies look good on a black compensated gamma screen (no crushed shadows) and bad on a bt1886 screen, it's only obvious to say that the studios don't use the BT1886 with their monitors.

[Asmodian]

If your display has a very high black I agree that pure BT.1886 looks bad but on decent displays it is very nice.

BT.1886 is not for your average crappy LCD, it is for nice displays using LCDs. It was needed because of the switch to high-end LCDs from CRTs or plasmas. Trouble resolving shadow details is a problem on high-end LCDs calibrated using the old standards.

On bad displays it is better to move to more output offset and let black crush a bit. By black crush I mean being unable to resolve 17, 18, etc. or having less difference between steps than expected so it is harder to resolve shadow details. Trouble resolving shadow detail is better than completely washed out.

It is physically impossible to have the correct gamma over the entire range, no black crush, and a high black level so you have to sacrifice something. BT.1886 (pure input offset) sacrifices the gamma curve and pure output offset sacrifices shadow detail. Partially input and output offset sacrifices a bit of both; you can tune the ratio to look better on your display under your viewing conditions.

Edit:
The exact exponent used in the gamma equation is not that important and it is definitely flexible depending on viewing conditions.

[baii]

But the best ips displays are barely breaking 1500:1 uncalibrated, and they are somewhat abundant from mid range to high end. Let alone PC monitor/display, which should make up a pretty huge portion of htpc/PC/madvr user.

Sent from my XT1575 using Tapatalk

[huhn]

but the user that is buying am IPS is usually not the user that really cares about good picture quality.

if you would care you wouldn't buy it in the first place.

btw. IPS TV screen are pretty rare these days. most of them don't reach ansi CR of 1000/1. but them we have local dimming and other stuff making judging this CR hard.

only LG really likes IPS screen these days and this is reflected in the usual test source too: http://www.rtings.com/tv/reviews/lg

[chros]

From the madvr thread:
Does it matter what curve you use on this imagined display? You do not want to watch Alien on a display with 0.14 cd/m^2 black.
This isn't an imaginary display, this is my TV! :D (see my signature) That was the best CCFL IPS 6 years ago.
Well, this is what I got now (and I hope it will last for 1-2 years): it's big, beautiful colors, shitty blacks :D This is what it is.

Your choice is black crush or washed out, using 2.4 or 2.2 gamma doesn't change much.
It does, and a lot (with the proper input/output offset)!

At night, in the dark, 2.4 looks better to me, 2.2 is for a well lit room.
Agreed.

BT.1886 avoids black crush..
That's true, but it can produce elevated whites (washed out image), like other guys experienced it.

...while pure power 2.2 or 2.4 will crush black some with that high of a black level.
That's why input/output/black offset can be used to compensate the effect manually and not just relying an another bad automatic option provided (BT.1886).

I tried Gamma 2.2 and this it looks a lot better to me (shadows aren't as lifted/elevated looking). And yep, you have to take so much into account for your user experience (I view at night, matte IPS, what target brightness I want, my display is also not natively capable of 2.4 gamma - closer to 2.2).
Good job!

This 2.4 vs 2.2 battle is overall stupid.
Rec709 movies are supposed to be seen with 2.2 gamma.
True.
All major studios calibrate and grade to 2.2, as noted by many professional colorists on the AVSForum board.
That's, again, simply not true. See below/above. (James, you wrote something completely different in this thread :) )

If you want to raise the gamma anyway, enable "dynamic contrast" on your TV, just be sure it's originally calibrated to 2.2.
Unless ur TV is very old or a monitor calibrating to 2.4 has no point as modern TVs with dynamic contrast can handle "dynamic" higher gamma better than pure 2.4 gamma (atleast 2013+ models), basically less black/white crush.
I don't know anything about new TVs, I tried it out on mine (low/mid/high settings) and it produced different results than the calibrated one, so I disabled it completely.

That would only make sense if consumer TV's were used to grade those films instead of ultra high-end non-consumer TV's with very deep black levels.
There's the point!

BT.1886 is terrible (IMO).
No two displays will result in the same curve.
Black Compensated Gamma curve makes sure that from a certain point all displays are equal.
That's right, at least close to each other as they can by nature.


From this thread:
Clearly BT1886 is nowhere near 2.4 with a 1000:1 display.
It also doesn't follow 2.2 or 2.4 above or below 50% input.
That's the whole point! Every gamma curve is changing by the black level of the device!
There's no such thing that 2 different device use the same curve by default (no matter what curve you use).
Btw: you don't have to change the black offset in HCFR, you only have to provide an Y value for the black. see it here: http://www.avsforum.com/forum/139-display-calibration/1471169-madvr-argyllcms-139.html#post40108362
And pretty good explanation that you've provided!

Nobody said it was. Standard BT.1886 has a technical gamma of 2.4, in other words, 2.4 is used as the exponent. However, because a BT.1886 curve (in general) is not a pure power curve...
This is the first misunderstanding: none of them are in real life (except for maybe OLEDs)! All of them are taking the black level of the display into account!
http://www.avsforum.com/forum/139-display-calibration/1471169-madvr-argyllcms-139.html#post40108362

, the effective gamma ("average" gamma, or best fit for a pure power curve) is closer to 2.2 when a display with a non-perfect black point is used. This is completely normal and is working as intended - the whole point of BT.1886 is to compensate for such displays by lowering effective gamma.
Take a look at the image of BT.1886 (along with the other 2) in B.4.c. section in the above link. I don't think that is normal at all. And this is where the device capability comes in.

Sure. BT.1886 is not a pure power curve. So what? Obviously the people who wrote BT.1886 standardized on that shape on purpose, otherwise they would just have mandated a pure power curve. Are you claiming that they're wrong, and if so, do you have specific evidence to back up your case?
They're not supposed to. They're supposed to be calibrated to BT.1886. That's kind of the point of having a standard.
In an ideal world where every display would be the same, yes. But we aren't living in an ideal world. :)
That made me think about: indeed, for what displays did they created this curve? :)
See below, or in B.4.a. section in the above link: with OLEDs you'll get 2.4 almost pure (!) gamma curve!!!! So much about standard :D

You're not taking viewing conditions into account, which are extremely important when determining the correct viewing gamma. If you take a movie that was graded in a dim room (which AFAIK is the norm), and watch it in a pitch-black room, you need to increase gamma to compensate. This is why the ArgyllCMS docs (http://www.argyllcms.com/doc/Scenarios.html#TV1) recommend an effective gamma of 2.4 for dark conditions (higher than BT.1886, at least for non-perfect displays), and 2.2 for dim conditions.
That's perfect statement. Then, again, how could be BT.1886 be the right choice all the time?

From my own personal experience, I have a TV that has awful blacks and a contrast ratio of less than 1000:1, and 1886 looks terrible on it. Washed out and lifeless. I settled on 2.2, with a black offset (92% output offset in displaycal) so blacks aren't crushed. Looks good, waaaaaay better than 1886. I wonder if those who sing the praises of 1886 have seen just how bad it looks on TVs with poor blacks/contrast ratios.
The problem is they didn't see any other display. And this is the whole point: you can't state that there are universal (best) option for everybody. They have to experiment what the best is for their setup.

Bt.1886 on a oled is flat 2.2 iirc.
With infinite black level BT1886 will result in exactly Gamma 2.4 power curve.
James's right. (see in B.4.a. section in the above link) And that's what you almost get with the new OLED TVs.

I don't think bt1886 arguments have much to do with viewing condition, it is the fact that it is trying to compensate for different black level and lifting the shadow like mad.(when rest of the world aren't doing the"standard".)
It does, and again, depending on the device. (Take a look at the image of BT.1886 (along with the other 2) in B.4.c. section in the above link.)

Professional OLED grading monitors like the Sony PVM/BVM series have minimum output level setting that usually set to 0.05 or 0.1 nit resulting in a contrast ratio of 1000:1 or 2000:1.
With infinite black level BT1886 will result in exactly Gamma 2.4 power curve.

If your display has a very high black I agree that pure BT.1886 looks bad but on decent displays it is very nice.
And that's the point (again): we can't say that 1 setting is (should be) good for every display!

BT.1886 is not for your average crappy LCD, it is for nice displays using LCDs. It was needed because of the switch to high-end LCDs from CRTs or plasmas. Trouble resolving shadow details is a problem on high-end LCDs calibrated using the old standards.
:) There's no trouble at all with black crush. (see my reply to you earlier and below)

On bad displays it is better to move to more output offset and let black crush a bit. By black crush I mean being unable to resolve 17, 18, etc. or having less difference between steps than expected so it is harder to resolve shadow details. Trouble resolving shadow detail is better than completely washed out.
And that's what you can avoid with using the proper input/output offset with gamma 2.2/2.4 like curves.

1 more thing about BT.1886 and viewing condition: take the example of a state of the art OLED display today. You'll get almost pure 2.4 gamma with it which is perfect for night. Unless you have a dedicated cinemaroom that always provides the same condition (I don't :) ), I don't think this setup will be good for watching something during a bright day in the living-room: you'll get crushed blacks with the exact same setup the only difference is the viewing condition.

In summary:
BT.1886 is not the solution for all our problem but an alternative option to use.

My whole point is this: we can't say that 1 setting is (should be) good for every display!
If somebody asks for help the first question should be "what is your black level of your display?". Then "What's your viening condition?".
They have to understand how the whole thing is working and after that They have to decide which is the right option for themselves. :) (And that's the user preference part at the end).

Who is unsure about this topic, please, go through this post (if you didn't do it already, along with the link for an article at the beginning), it took me quit some time to understand it, and more to write it :)
http://www.avsforum.com/forum/139-display-calibration/1471169-madvr-argyllcms-139.html#post40108362
You can even play with it (as James suggested it) that helps understand the whole concept.

My preference (but this is just my opinion, you can have different) quoted from the above linked post:
"Personally (I don't like to darken my viewing environment, I prefer the way as it it during the whole day), I agree with Vincent Teoh's opinion, I want to get a:
- for evening/night usage (with just a dim environment light on): 2.4 like gamma with slightly elevated shadows at the lower end of greyscale
- for cloudy day usage: 2.2 like gamma with slightly elevated shadows at the lower end of greyscale
- for sunny day usage: 2.2 like gamma with a little bit more elevated shadows at the lower end of greyscale"

Final note: if we think that this is confusing then just take a look ahead, what will come with the new HDR capable displays :) It will be a huge mess, especially in the beginning.

[Asmodian]

You seem to think using an input offset is different than BT.1886 or is your point that a gamma of 2.2 is that much better than 2.4? I think of 100% input offset with 2.2 gamma as BT.1886 like for well lit rooms.

I suppose I agree with you, you should not simply use BT.1886 on all displays, but I think BT.1886 is great because I remember calibrating before input offset was available.

I do not think gamma is confusing at all, it is actually pretty simple once you realize there isn't a "correct" way to handle gamma. Balance input and output offset so it looks nice, my setting of 100 input offset and 2.4 gamma (BT.1886) looks good on my IPS LCD display and I have tried the full range of options in Argyll and I also use HCFR to see what the results are according to my meter (i1d3). I have another display on which I use a 60% input offset. I still use 2.4 gamma (technical) though, my room is very dark and 2.2 looks a bit washed out.

[chros]

You seem to think using an input offset is different than BT.1886 ...
It's not just my opinion, it's a fact :) You can take a look at the pics in B.4.b. : they are completely different (and that is a WLED panel, not a CCFL IPS!!! :) ).
And you can also try it out (it's just 5 mins in HCFR.)

or is your point that a gamma of 2.2 is that much better than 2.4?
No. Again, it depends. :) I edited the above post in the summary to make it more clear. (But that is just my opinion :) )

I think of 100% input offset with 2.2 gamma as BT.1886 like for well lit rooms.
Again, it depends on the black level of the device. See above.

I suppose I agree with you, you should not simply use BT.1886 on all displays, but I think BT.1886 is great ...
It can be, it's not the argument :) And I'm happy that you (and not just you) like it.
BT.1886 is not the solution for all our problem but an alternative option to use.

[Stereodude]

But the best ips displays are barely breaking 1500:1 uncalibrated, and they are somewhat abundant from mid range to high end. Let alone PC monitor/display, which should make up a pretty huge portion of htpc/PC/madvr user.
Most LCD TVs are VA instead of IPS for good reason. Samsung's latest VA panels are 6000:1. For some reason VA computer monitors are not all that common.

[chros]

If you like the idea to use separate 3DLUT for day and night (I ended up just using 1 for day), I uploaded a script to pastebin that I'm using (since madvr doesn't support this by now): http://pastebin.com/ECBX5weQ
It can change different things based upon the given time interval:
- copy appropriate 3DLUT (I couldn't figure it out how on earth you use symlinks on Win without installing any other tools)
- change your display brightness if it supports DCC (commented out by default)
- change the theme of Win (commented out by default)

You can put any extra stuff in it if you like.
Just place a shortcut on the desktop for it and you can just double click on it to change the settings.
It can be also put into Task Scheduler but I didn't want to do it.

[Asmodian]

It's not just my opinion, it's a fact :) You can take a look at the pics in B.4.b. : they are completely different (and that is a WLED panel, not a CCFL IPS!!! :) ).
And you can also try it out (it's just 5 mins in HCFR.)

Using Argyllcms you only have options for % output offset, exponent, and technical or relative. I am not sure what HCFR is using for all their various gamma options but BT.1886 is simply a specific set of these options (0% output offset, 2.4, technical) in Argyllcms. Nothing unique about it, maybe HCFR means technical gamma by its BT.1886 option?

Technical v.s relative gamma is pretty different at the same gamma value but by tweaking the gamma value you can get their curves to pretty much match.

[chros]

Using Argyllcms you only have options for % output offset, exponent, and technical or relative. I am not sure what HCFR is using for all their various gamma options but BT.1886 is simply a specific set of these options (0% output offset, 2.4, technical) in Argyllcms.
That's correct. As in stated in B.3.a.

Nothing unique about it, maybe HCFR means technical gamma by its BT.1886 option?
No. The config dialog says 0=Absolute. (But that means you haven't tried it out! Why not?! :) )
They can't understand anything else, because it wouldn't be BT.1886 (as you/I stated above).
Those guys know what they're doing. :)

Technical v.s relative gamma is pretty different at the same gamma value but by tweaking the gamma value you can get their curves to pretty much match.
It can be, I haven't tried it. If you can provide some nice data/pictures about it it'd be nice :)
You couldn't/can't set other value than 0 for absolute gamma in HCFR (with 100% input offset means BT.1886) so you can't experiment with it by just entering numbers (I asked Zoyed half a year ago, but he wasn't interested in it).
What you can do to check is (but it's a longer process) generating a new 3DLUT with the selected setting, then measure it in HCFR or in Dispcal and compare the resulted gamma curve to the other 3DLUT.

[chros]

If you like the idea to use separate 3DLUT for day and night (I ended up just using 1 for day), I uploaded a script to pastebin that I'm using (since madvr doesn't support this by now): http://pastebin.com/ECBX5weQ
It can change different things based upon the given time interval:
- copy appropriate 3DLUT (I couldn't figure it out how on earth you use symlinks on Win without installing any other tools)
I found out, you can't create symlinks simply (for whatever reason) but only hardlinks (ppffff :) ).
So here's the modified script that uses hard links, so no more copy involved: http://pastebin.com/fKP2q96k

[iSeries]

Ok so after suffering the awful black level of my LG plasma for long enough I followed the pot tuning guide over at AVSforums. Pre-tweak my black level measured an embarrassing 0.2 cd/m2. Post-tweak, a much more reasonable 0.09 cd/m2. Recalibrated with Displaycal last night and BT.1886 looks MUCH better. Very close to 2.2 overall but with slightly lifted shadows compared to the 2.2 with offset I was using before, and it manages to do this without making the image washed out. To me, this confirms my suspicion that BT.1886 does not work well for TVs with bad black levels.

[nevcairiel]

Ok so after suffering the awful black level of my LG plasma for long enough I followed the pot tuning guide over at AVSforums. Pre-tweak my black level measured an embarrassing 0.2 cd/m2. Post-tweak, a much more reasonable 0.09 cd/m2. Recalibrated with Displaycal last night and BT.1886 looks MUCH better. Very close to 2.2 overall but with slightly lifted shadows compared to the 2.2 with offset I was using before, and it manages to do this without making the image washed out. To me, this confirms my suspicion that BT.1886 does not work well for TVs with bad black levels.

I'm confused, you say that BT.1886 looks much better for you now, and at the same time you say its not meant for bad black levels, which your TV evidently has? So does it look better now or what?

[iSeries]

I'm confused, you say that BT.1886 looks much better for you now, and at the same time you say its not meant for bad black levels, which your TV evidently has? So does it look better now or what?

I don't know if I agree that 0.09cd/m2 (0.025ftl) is a bad black level . I'd say it's now an average black level. Previously, it measured 0.2cd/m2, which definitely is a bad black level. Yes it looks better now I have an average black level. Previously, BT.1886 looked absolutely terrible.

[chros]

Ok so after suffering the awful black level of my LG plasma for long enough I followed the pot tuning guide over at AVSforums. Pre-tweak my black level measured an embarrassing 0.2 cd/m2. Post-tweak, a much more reasonable 0.09 cd/m2.
Impressive! Can you share the link for the guide? Thanks

I'm confused, you say that BT.1886 looks much better for you now, and at the same time you say its not meant for bad black levels, which your TV evidently has? So does it look better now or what?
iSeries referred to the pre and post calibration settings.

Recalibrated with Displaycal last night and BT.1886 looks MUCH better. Very close to 2.2 overall but with slightly lifted shadows compared to the 2.2 with offset I was using before, and it manages to do this without making the image washed out. To me, this confirms my suspicion that BT.1886 does not work well for TVs with bad black levels.
Yep. And it can also be broken (like many things) :) : http://www.avsforum.com/forum/40-oled-technology-flat-panels-general/2132034-lg-55ef9500-65ef9500-oled-owners-thread-612.html#post43540586
If you are already in the experimental mood and if your TV has 2.4 gamma setting then you can try to calibrate it to that setting for night usage. After that you can decide which one you like.

I don't know if I agree that 0.09cd/m2 (0.025ftl) is a bad black level . I'd say it's now an average black level. Previously, it measured 0.2cd/m2, which definitely is a bad black level.
Well, I won't say that it's average.
My CCFL LCD has: 0.1593 (day setting) , 0.1351 (night setting) ; which are bad :)
I'd say the average black level today is around 0.027 (WLED panel)
In contrast an LG OLED has about 0.0005 :)

[iSeries]

Impressive! Can you share the link for the guide? Thanks


iSeries referred to the pre and post calibration settings.


Yep. And it can also be broken (like many things) :) : http://www.avsforum.com/forum/40-oled-technology-flat-panels-general/2132034-lg-55ef9500-65ef9500-oled-owners-thread-612.html#post43540586
If you are already in the experimental mood and if your TV has 2.4 gamma setting then you can try to calibrate it to that setting for night usage. After that you can decide which one you like.


Well, I won't say that it's average.
My CCFL LCD has: 0.1593 (day setting) , 0.1351 (night setting) ; which are bad :)
I'd say the average black level today is around 0.027 (WLED panel)
In contrast an LG OLED has about 0.0005 :)

Here's the link: http://www.avsforum.com/forum/167-plasma-flat-panel-displays/1372479-lg-plasma-panel-tweaks-better-blacks.html

Yes, 'average' is relative. It's an average black level, when compared between something that has a *bad* black level (cheap LCD), and something that has an *excellent* black level (the best plasmas, the very best LCDs with FALD, OLED etc). And price has to be a consideration when considering what is good, bad or average. I only paid $699cdn for my 60" plasma. For that price, I may even consider it to have a *good* black level.

The black level on my TV is actually worse for 24p sources. At 60hz, the black level measures lower.

To be clear, I only said BT.1886 looks better than it did. 2.2 with a small offset still looks better to me. 2.4 is too dark I find.

[chros]

Yes, 'average' is relative. It's an average black level, when compared between something that has a *bad* black level (cheap LCD), and something that has an *excellent* black level (the best plasmas, the very best LCDs with FALD, OLED etc). And price has to be a consideration when considering what is good, bad or average. I only paid $699cdn for my 60" plasma. For that price, I may even consider it to have a *good* black level.
Yes, you're right, mine was around 550 euros 5 (?) years ago (although it was a refurbished one, without any defect). But it still doesn't change the fact that what was average 6 years ago it's still now. Fortunately technology is evolving. :)

Here's the link: http://www.avsforum.com/forum/167-plasma-flat-panel-displays/1372479-lg-plasma-panel-tweaks-better-blacks.html
Woow! :D That was something! :) Good job!

[iSeries]

Yes, you're right, mine was around 550 euros 5 (?) years ago (although it was a refurbished one, without any defect). But it still doesn't change the fact that what was average 6 years ago it's still now. Fortunately technology is evolving. :)!

My TV black level measures even lower at 60hz than when playing 24p stuff (manual says 3:3 pulldown but my meter is measuring 96hz not 72hz). Black level is measuring 0.05cd/m2 at 60hz. Think i'll recalibrate at 60hz and use MadVR's smooth motion.

[chros]

My TV black level measures even lower at 60hz than when playing 24p stuff (manual says 3:3 pulldown but my meter is measuring 96hz not 72hz).
Which meter with which software can measure frequency?

Black level is measuring 0.05cd/m2 at 60hz.
That's strange. What's the reason behind it?

Think i'll recalibrate at 60hz and use MadVR's smooth motion.
I also tried this approach (I have problem with chroma subsampling with non-59/60p) but I didn't like the effect of it (and costs too much power as well), so I'm using 24/25/59/60p (I can't use 23p because I get massive frame drops, but it's fine).

[iSeries]

Which meter with which software can measure frequency?


That's strange. What's the reason behind it?


I also tried this approach (I have problem with chroma subsampling with non-59/60p) but I didn't like the effect of it (and costs too much power as well), so I'm using 24/25/59/60p (I can't use 23p because I get massive frame drops, but it's fine).

Displaycal shows the refresh rate when calibrating. A reasonable explanation for the rise is given here http://www.avsforum.com/forum/139-display-calibration/1311386-lg-plasma-60pk550-mll-50-higher-24-fps-mode.html.

Smooth motion works well enough for me. Not as smooth as feeding the TV 24p and having it do it's 3:3 (or 4:4?) pulldown (which can't be turned off) but smooth enough and the difference in black level is quite noticeable.

[chros]

Displaycal shows the refresh rate when calibrating.
Really? I've never noticed it. Was it always there or is it recently added?

[Asmodian]

Really? I've never noticed it. Was it always there or is it recently added?

It has been there for four years at least, I haven't found it to be very accurate but it does work.

[chros]

It has been there for four years at least, I haven't found it to be very accurate but it does work.
Sorry guys, but I don't see it anywhere (not in the interactive measurement nor when calibration starts). I'm using DisplayCal 3.1.3.1 with i1displaypro.
I've selected:
- office & web profile (but it's the same for a VideoLut profile with madvr as well)
- my TV at Display
- LCD (generic) at Mode
- hit calibrate & profile

[Asmodian]

Sorry guys, but I don't see it anywhere (not in the interactive measurement nor when calibration starts). I'm using DisplayCal 3.1.3.1 with i1displaypro.
I've selected:
- office & web profile (but it's the same for a VideoLut profile with madvr as well)
- my TV at Display
- LCD (generic) at Mode
- hit calibrate & profile

Not DisplayCal, I don't use the GUI, but the command line programs from Argyllcms, dispcal and dispread.

On a 144 Hz display:
C:\Calibration\XB270HU>dispcal -v -dmadvr -m -qh -ye -G2.4 -f0 -k0 -O "XB270HU matrix" -o XB270HU_1.icm XB270HU
Setting up the instrument
Product Name: i1Display3
Serial Number: -
Firmware Version: v1.01
Firmware Date: 12Apr11
Created MadVR window
Place instrument on test window.
Hit Esc or Q to give up, any other key to continue:
Display type is 'e'
Target white = native white point
Target white brightness = native brightness
Target black brightness = native brightness
Target effective gamma = 2.400000
Commencing display calibration
Measured display update delay of 3 msec, using delay of 104 msec & 0 msec inst reaction

[chros]

Not DisplayCal, I don't use the GUI, but the command line programs from Argyllcms, dispcal and dispread.

On a 144 Hz display:
"Measured display update delay of 3 msec, using delay of 104 msec & 0 msec inst reaction"
Thanks Asmodian, DisplayCal also shows this info, but how can we get the refresh rate of the display out of it?

[huhn]

what has the update delay to do with the refreshrate?

and on a LCD it is not really possible to measure the refreshrate without frame interpolation or BFI.

[Asmodian]

Thanks Asmodian, DisplayCal also shows this info, but how can we get the refresh rate of the display out of it?

Refresh rate = 1000 / update delay (ms)

what has the update delay to do with the refreshrate?

and on a LCD it is not really possible to measure the refreshrate without frame interpolation or BFI.

In this situation I believe dispcal sends a black, white, black frame and attempts to measure the time between frames with the colorimeter. The value reported does seem to match refresh rates better at lower refresh rates.

[huhn]

In this situation I believe dispcal sends a black, white, black frame and attempts to measure the time between frames with the colorimeter. The value reported does seem to match refresh rates better at lower refresh rates.

if you send a 24 fps signal and do this than the result would be at best 24 but the screen is most likely running at 120. so you can get the refreshrate of the screen this way.

[chros]

Refresh rate = 1000 / update delay (ms)
I believe this is far off from the original, right? :) 333 != 144. :)
so you can get the refreshrate of the screen this way.
Probably you meant "can't".

Btw, do we know about any software that can be used with our colorimeter to determine this? Or is our meter fast enough at all for this task?
(I just found this on youtube: with photocell and multimeter: https://www.youtube.com/watch?v=_Hmmb-RCfII)

[huhn]

that's a plasma it is a refresh based screen LCDs doesn't work like this.

[e-t172]

I've used a photoresistor (specifically, a Panasonic AMS3 (https://industrial.panasonic.com/cdbs/www-data/pdf/ADD8000/ADD8000CE2.pdf)) plugged into an audio input to measure actual LCD refresh rates, using a test video that alternates black and white frames. The audio input is basically used as a poor man's oscilloscope - it works as long as your audio input doesn't have an aggressive low-cut filter (24p generates a 12Hz signal).

It is possible to measure jitter and spot "problems" (such as unexpected 3:2 pulldown, inconsistent cadence, dropped frames, duplicated frames) by plotting the spectrum (e.g. using Audacity) and looking for harmonics that shouldn't be there (i.e. aren't multiples of the refresh rate). When I first tried it I didn't expect it to work. I was pleasantly surprised. I even noticed some interesting things, for example madVR frame timing seems to behave quite erratically when the refresh rate doesn't match (e.g. 24p@60Hz) and Smooth Motion isn't used (an issue madshi is already aware of, it's not what madVR is designed for anyway).

I would be interested to know if a colorimeter such as an i1 Display Pro can also measure time with enough accuracy to provide refresh rate estimates and spot cadence or jitter issues. I don't trust what ArgyllCMS reports because I have no idea how it's calculated, and I can't use my own test signals for the measurements.

[e-t172]

Answering my own question about how ArgyllCMS measures refresh rate with an i1 Display Pro, from Argyll sources (spectro/i1d3.c):

determining the refresh rate for a refresh type display;

Read 1300 .5 msec samples as fast as possible, and timestamp them.
Interpolate values up to .05 msec regular samples.
Do an auto-correlation on the samples.
Pick the longest peak between 10 and 40Hz as the best sample period, and halve this to use as the quantization value (ie. make it lie between 20 and 80 Hz).

If there was no error, return refresh quantization period.

If there is no apparent refresh, or the refresh rate is not determinable, return a period of 0.

I didn't know that a typical colorimeter like an i1 could measure as fast as 0.5 milliseconds, that's impressive considering the instrument is not really designed to do that in the first place. That would make it an excellent tool to measure all kinds of time-related behaviors, including refresh rates and video player cadence issues. The only thing missing is some software that can do the measurements and report detailed information when used in combination with a special test video. ArgyllCMS doesn't use a test signal, and therefore is only adequate for measuring the flicker frequency of displays that inherently flicker no matter what the signal is (CRTs, plasmas, etc.). If one were to use the same basic mechanism but with a test video signal and specifically designed software, we could measure the end-to-end video chain in very interesting ways.

[chros]

Interesting idea! It can be worth to post the same thing e.g. in DisplayCal thread on avsforums, maybe one of the devs (garym, foech, zoyd) will like this idea as well.

[Asmodian]

if you send a 24 fps signal and do this than the result would be at best 24 but the screen is most likely running at 120. so you can get the refreshrate of the screen this way.

Argyll uses its own display method and generates its own patterns which run at the refresh rate set in Windows, so it can only measure that.

I believe this is far off from the original, right? :) 333 != 144. :)

I did say it I didn't find it to be very accurate when I first mentioned it. ;)

Interesting idea! It can be worth to post the same thing e.g. in DisplayCal thread on avsforums, maybe one of the devs (garym, foech, zoyd) will like this idea as well.

DisplayCal is just a GUI for Argyll CMS so they already have access to and use this method. :)

[zoyd]

Colorimeters typically use light to frequency converters as the sensing element and can run up to 500 kHz at peak light input.

[chros]

Colorimeters typically use light to frequency converters as the sensing element and can run up to 500 kHz at peak light input.
Wow! That's a lot!
What do you think: can it be possible to measure a display internal (!) refresh rate with any kind of software and a colorimeter?
(E.g. I'm curious about my LCD real refresh rate, not that is reported by Windows.)

Thanks for checking in, zoyd!

[Lyris]

What do you think: can it be possible to measure a display internal (!) refresh rate with any kind of software and a colorimeter?
(E.g. I'm curious about my LCD real refresh rate, not that is reported by Windows.)

The software bundled with the Klein K-10 (and K-10a) can measure refresh rate, I used it all the time for reviews at HDTVtest.

[huhn]

is it possible to read the speed of the "impulse"/BFI mode of TV too?

the KD-65ZD9 is supposed to flicker a lot it would be nice to know at what HZ it does this.

BTW. it would be lovely if you guys could create colorimeter correction (if a colorimeter like a T1D3 is present too) and upload them to this data base:
https://colorimetercorrections.displaycal.net/

[Siso]

@chros what black output offset should I use with a monitor with 902:1 contrast, and black level 0.1116 cd/m²? The gamma will be 2.4 relative or absolute?