Guide: GREYSCALE CALIBRATION FOR DUMMIES - Page 3

Gregg, I was about to say the same thing.

Thank you for saying "field of view"; that really clears it up for me.

Is there a ratio of brightness to FOV?

Thank you,

-Brian

thread cleanup

if you guys want to fight: don't do it on AVS

As far as I know there is no standard for the transfer function of studio monitors so this argument over 2.2 or 2.5 display gamma is kind of moot. IIRC Rec. 709 implies a 2.5 display gamma, but with 2.2 or 2.5 you will end up with an end-to-end gamma > 1 anyway so I would say it comes down to viewing conditions and personal preference. Tom Huffman said it best in this thread IMHO.

Yes, indeed it's very unfortunate that the display transfer(or the de-gamma) has not been explicitly defined, and that's why we have some confusion today. It's also why many digital displays have their LUT at gammas that are often quite low, and image depth is lost.

Until it is clear that mastering practices change, I stick with the CRT as the reference, and CRTs when properly setup will have gamma higher than 2.2. I think 2.4ish 2.5 is common and a reasonable range to target as a reference.

But again, I myself tweak my own gamma (on my CRT) slightly lower for those subjective reasons, and think it is fully reasonable to do so. It's just important to know that if you're hitting 2.2, you're already lower than standard, so going even lower than that is likely going to begin to become more strongly deviant and detrimental to image fidelity because the departure from standard is more severe.

Again, I'd really like to thank kal for putting this guide together. Thank you.

I have a bit of a "silly" question that I'm probably overlooking the obvious answer to.

I'm wondering why my brightness setting (black level) is so WILDLY different between sources.

For instance, on my Xbox 360 I have say a 50% brightness, and on my PS3 I have a 30% brightness and on my PC I have a 40% brightness.

However my contrast and color controls across the board are all exactly the same.

I'm using an i1Display LT, so I'm going by measurments not guess.

Thank you,

-Brian

kamuls question above is very similar to what troubles me also. Being that we are using a dvd player as our source for test signals, do we assume that the factory settings i.e. brightness, contrast, color, tint, etc. will give us correct levels? I have an XA-2 and all of these settings are user tweakable. Do I set all of these back to '0' or factory default before I adjust my projector?

Let me also thank kal for an excellant document.

How are you connected?

If you're connected some via HDMI/DVI and some analog, then that makes sense. Additionally, even if all connected the same, there are different output standards, and devices aren't always perfect on top of that. So it's not surprising that black level is different, however it may also signal that sources are not setup properly, particularly if they're connected digitally, where they all basically should be 16-235 for video. But you're dealing with three dual-use graphics-game/video platforms, so unless they're set properly they may be outputting 0-255 for instance, or something else, so that might be worth figuring out first.

As to white level, that's a bit harder to see than black level, so there may be differences there that you're not detecting visually as easily.

Well, I just started typing this long winded message, when it finally sunk in. Thanks for the help, Chris.

I haven't got an answer to my earlier question though, but do you (or anyone else) know a way to calculate field of view to maximum light output?

I'm using a 32" LCD and I'm sitting 6 feet from the screen (best comprimise for audio / video, SD and HD material).

I would like to know what foot lambert I should be aiming for. Also as I've mentioned my room is light controlled with an Idealume bias light. I've seen suggestions ranging all the way from 30 fL to 60 fL!

When Gregg Loewen said "field of view" that really cleared my confusion, but now I'm wondering if there is a mathmatical way of calculating this difference.

Thanks in advance,

-Brian

I agree with you and especially for viewing in a dim surround, which the standards are tailored for.

Subjectively I prefer a display gamma > 2.2, at least with my Pio Kuro PDP.

Hi, I just followed this guide tonight with an Eye-One Display 2 LT on my Mitsubishi HC3000 DLP projector. Thanks very much for the guide; it was enormously helpful and the process was even fun!

After my first round of adjustments, which only took a few minutes, my results seemed pretty good from 20-90, but then at 100 they look pretty nuts:



Any idea what could account for that or how to correct it? They are so far off in that range, and the rest of the range is mostly very good, that it doesn't seem like I could tame them without messing up the rest of the range.

For what it's worth, in my "before" measurements, the delta E of the 10-90 range was all 11 to 17, but the 100 was only 9.1 (compared to 12.1 in my "after").

Also, my gamma looks bad, but I don't know what to do about it. I adjusted the user Gamma settings on the HC3000 (high, mid and low controls), and ended up maxing out the high and mid which improved things a bit, but this is still the best I got:

curttard, it looks like you ran out of red (and to a lesser extent green). The only remedy is to turn down your contrast. While diplaying the 100% pattern, run a continuous measure and keep backing off on contrast until x (red) hits .313. This will be the highest you can run your contrast and maintain red at 100% (you cannot raise the red drive above the point it is at either, you will have to adjust around it using green and blue). Running out of green may be why your gamma dove off a cliff at 100%.

Thanks for the tip, I'll try that tomorrow. Do you think it would be worthwhile, though? It seems like I'd probably be giving up some accuracy across the vast majority of the range, and definitely a lot of contrast (overall picture brightness) as well, just to flatten out that dip at the very top end.

One other question I had was that the guide recommends setting contrast to reach a certain ftL, but doesn't that ignore white clipping? And with a projector, wouldn't the ftL be entirely dependent on the projector's lumen output, age, etc?

You shouldn't be giving up any accuracy, but you are correct that you will be giving up some of your contrast ratio and light output. But in return you should get a much better gamma response at the top end and the color of white would be correct for anything between 90% and 100%. Also keep in mind that most comercial DVD/Blu-Rays contain information that can extend out to what the AVS HD disc lists as 109% under the misc. patterns. That drop you see on your graphs extends all that way out to 109%. If you want to protect all the way out to 109%, you would substitute the 109% pattern for the 100% in my instructions.

As most of the calibrators here will tell you, calibration often requires deciding which trade offs to make to get the best picture from a particular set up. You'll have to decide whether getting your top end correct is worth the drop in contrast ratio and light output. You should probably measure it, see the extent that it happens and then make the call.

Yes the recommendations totally ignore any real world problems you may have getting to those results. You have the situation of color run outs if you have insufficient light output for example. You can also have a case where you have far more light output available, an iris or a backlight can't be adjusted and you end up throwing away contrast ratio to get to the recommendations. In the case of a projector, there is the option of using a neutral density filter to cut your light output while maintaining contrast ratio.

Thanks for all the information. I'm looking forward to trying it out tonight.

That's a really well done guide you made kal. I wish there had been something as thorough when I first tried figuring out this stuff out years ago.

You definitely might want to make an addendum about color run outs on digital displays. Using 80% is a good way to set your grayscale, but it can leave you with problems like curttard had. Unless your contrast is set ridiculously high, you will not see a color run out problem measuring at 80% until you make a grayscale run. Most people end up scratching their heads wondering why their color temperture/gamma lines dive of a cliff by 100%. It's not something that is immediately intuitive since they are not looking at 100% when you are adjusting the bias/gain controls. A good way to find the ceiling on digital displays is put up a 100% pattern, run a continous measure and raise and lower the contrast watching x for when hits a ceiling. There will be a point where x will start dropping as you raise your contrast, and you will have to stay below that point (all this also applies to green and blue which are less likely to run out). If you decide to use that point for your contrast, you will have to leave the red drive alone since moving that control will change where red runs out.

I'm glad this thread was created because I'm intrigued about using a color correction filter to improve the contrast ratio (and hopefully black levels) on my RPTV. I used a neutral density filter in a prior RPTV, so I've already gotten a little experience under my belt.

I know the guide is specifically for greyscale calibration, but I'm curious about how the gamma controls on my projector (adjustments for High, Mid and Low) would interact with those. From some experimenting, I got the gamma looking better in terms of average (2.2 spot on), but the color breakdowns were pretty wild (red). I'm going to try again tonight, but I suspect I'll end up leaving things how they are now -- I definitely can't afford to give up any more light output by lowering the contrast. Here are my stats as they stand now:

Is there a way to calibrate a satellite source? How are you determining the black levels?

Thanks for the great tutorial. My eye-one is arriving today. [You're right. You should have bought stock!]

Curt,

think of the gamma graph as showing you the brightness response (in relation to 100%) of your white along the scale between 0 and 100 - while the grayscale graph is showing you the color response from 0 through 100. Your graphs show that red is dropping off rather fast at the high end while blue is spiking up. Essentially you're losing consistent grayscale color performance somewhere between 90% and 100%. Lowering contrast a bit is the only option I can think of to improve that. You may want to post your .chc file in the HCFR calibration file thread so folks can comment on it.

hope this helps,


--tom

curttard, if you look at the graphs you posted, you can see that you have a run out on red. You tried compensating by raising your high end gamma. You can see the hump at 80% that brings both your blue and green gamma to under 2.5 while your red has dropped to 1.8. At 90% and 100% the gamma on red is so low that it's under 1.0 (and the gamma on blue has gone to 2.9). You would be much better off lowering your contrast to stop the red runout and then adjust your high end gamma to closer to 2.2. You are not "2.2 spot on". You have an overall average gamma of 2.2. The average doesn't mean much if you have such wide variation along the curve. Until Chad B gave out some sage advice on adjusting gamma on my Sony HDTV, I had a 2.1 gamma average with my low end being 2.5 and sloping down to a 1.7 gamma at 100%. It looked no where near correct even though I had a 2.1 average.

You could most likely get a very good calibration out of that projector if you lower your contrast until you no longer get a run out on red and do the gamma and grayscale again. As it stands now, you are badly overdriving red and it's hard clipping above 90%. Your attempts to compensate with the gamma control have made the situation worse. Even if you don't want to lower your contrast, you should make an attempt to even out your gamma as far up as you can and just let 90%-100% go. You are far better off trying to get all 3 colors near 2.2 the whole way then to rely on just achieving an overall average gamma of 2.2.

I did try lowering the contrast on a 100% window until red came into line, but that point was just unacceptably low -- a white screen was grey, and not even a particularly bright grey.

One question about your method of finding the max contrast point before red starts dropping off -- where should the red RGBHighEnd be before starting to pinpoint the Contrast setting? Should it be left at 0, or should I get 30% and 80% correct and then go to 100% and adjust Contrast, then re-calibrate?

Just did another calibration. I went against one of the guide's recommendations and used the green controls. My results are much better now:






Huge thanks to Kal and Serron. I'm getting closer and think I'll just watch some movies for awhile before tweaking further

Curt what do those other two screens look like now from above in your new settings you just posted? The two colored curves...

For cburbs:

Does anyone know if the X-Rite Eye-One Display Lite EODLT is the same colorometer as the Gretag Macbeth Eye-One Display LT. Both are sold on amazon. I have a discount code but it must be sold through amazon not a third party vendor which the Gretag Macbeth Eye-One Display LT is.

Thanks

That looks tremendously better. It is worth pointing out that adjusting the green drive is the same as adjusting your contrast control. If you turn your green drive down, then you are in effect turning the contrast control down. It looks like you are no longer clipping red to the same degree which seems to have solved your problem.


It seems like the same product to me. They certainly display the picture of a display LT box.

I know, it just seemed to make more sense to use the green control. Just set Contrast at 0, then adjust whichever RGB controls need adjusting to get 80% and 30% right, verify the graphs to make sure everything's looking good, then raise Contrast until x starts to drop and back off one notch from there.

I have to say, though, that I might go back to the inaccurate settings. I can't say I actually see a difference in color except when I'm going from a 90% pattern to a 100% pattern, and I don't see a difference in details or anything like that. The difference I *do* see is that the picture when correct is a noticeably dimmer, and since I'm only getting 8 ftL or less anyway, I can't really afford to give up any light.

You might want to discuss the low light output issue in the projector forums (specifically in an owners thread for your model if one exists). Other may have suggestions or solutions for you.

Instead of going back to your prior calibration, you might as well just raise the contrast to where need it to be with your current settings and measure again. Just keep an eye on 80% and below because you are going to sacrifice somewhere above that where red hits the wall.