Ursa's "Guide to Basic Display Calibration for New Users" - a MUST read!!

I am flattered, and a bit embarassed by Bob's lead-in. However, I will do my best to live up to that introduction. Please fire away.

The file lives in the Calibration FAQ thread here: http://www.avsforum.com/avs-vb/showthread.php?t=585527

Later,
Bill

Personally I think GetGrey and Ursa should get together and sell their freeware offerings for $50 package deal - which is about what you pay for the AVIA video that teaches you user controls.

Sell? Freeware? Nah! Though as Scott can attest, this thought did cross my mind pretty early on.

You got it. The goal here was to try to see the relative magnitudes of changes for each control, i.e., was contrast a more granular control than brightness. The result? Not really. I still had clipping (pretty severely). Also, the end result of run 3 had my red color balance curve fairly obviously downward sloping. This indicated that I needed to further decrease brightness as well as backing down contrast. This then led to the changes I made for Run #4.


That sounds reasonable, but the question is how many tries it would have taken you to get the balance right. Since the primaries were inherently unbalanced to begin with, I did not try to maintain the ratio on the way down. Instead, my goal was to first maximize my limiting color (red), then to bring the other primaries in balance with it later.


The critical change was from Run #3 to Run #4. Since I had already set my master brightness and contrast, I did not want to make too radical of changes. So, Run #2 told me that I was not going to get a free lunch by tweaking a control for which I had no real documentation. Run#3 was similar, in that it showed that brightness and contrast were having about the same impact, and that I was really clipping red at that point.

The key was the downward slope in the color balance curve for red. If you think of it as a line, I needed to lower the low end a bit and bring up the top end a lot. However, since I was out of red at the top end, and needed to lower my contrast there, I needed to crash the other two primaries instead.

This may be an important point: lowering the other primaries will have a similar effect to raising a single primary (and vice-versa).

Thus, since I needed to a) lower the absolute level of red, and b) raise the relative level of red at the high end, this necessitated lowering the blue and green contrast almost as severely as I was changing red. If you look at the chart on p.17, you see that not only did I drop red contrast 16 points from my initial setting (Run #1), I also dropped blue 14 points and green 13 points. The result was reasonably flat color balance, but with red being significantly below blue and green.

At this point, I could have raised the red brightness a bit while lowering blue and green brightness, and the display should have been balanced relatively well. Of course, lowering green brightness would have probably lowered my black level beneath my ambient light floor and crushed my blacks. As a result, you see in Runs #5 - #7 my attempts to get back as much red as I could, while dialing-in the changes to blue and green to balance the primaries out.

Make sense?

Later,
Bill

Bob - No problem! Aside from you and Kevin, no one seems to have read the thing, so it's not like I'm getting inundated. I will be out of pocket most of tomorrow (one last day trip before my vacation), so I can't promise anything too responsive.

Later,
Bill

I'd like to second Bob's praise of Bill's pdf - it's really nice to have this kind of detailed account of calibrating. Given all the H77/78/79s out there, word of this document should be spread although the general principles can of course be applied to other pjs.

In my experience with the H79 and the Spyder2Pro 2.0, as well as with an OpticOne previously, I've found the same kind of behavior in the upper red, and even with a new bulb (with more red overall) I had to turn down red contrast to -6 or so to correct a bit of a peak and then dropoff at 80+% stimulus and then drop blue and green to -14 for results very similar to Bill's. (These are my rough recollections of the settings I used in the Image=>Advanced Adjustment section of the regular menu.) I had to turn red brightness down somewhat, too. I got good linear results for xy, using the TV and gamma 1 settings.

I see Bill is getting into the service menu for his cuts and gains. I haven't gone into the service menu since my early days with the H79 because using controls caused highly erratic responses reminiscent of the peaky graph he presents in fig 10. Gamma controls in the sm were notorious for causing it, but I think my adjusting contrasts/brightness controls did it, too. Just beware of this if you do go into the service menu.

Dan

Bill,

I just read it. Incredible. I actually feel like I understand everything much better now, as opposed to before reading it. You did a bang-up job writing that. I couldn't have asked for something more easy to follow.

Having read it, I have a question back regarding Run 4, and a possible fork in the road that could have been used. It seemed that blue and green both tracked pretty close to one another, although being 10-20% oversaturated, but fairly flat across the stimulus range. Red, on the other hand seemed to be about 25-30% undersaturated. The one thing I did notice was that it was fairly flat compared to other readings. Aside from needing to address the overall oversaturation of blue and green, would this have been a good place to try and boost the red saturation with the use of a CCR filter? It would seem that this might make things track better (flatter) and bring the stimulus of the three primaries to within closer values of one another.

I would really like to pick up a SpyderTV or other device. I understand that the GMB Eye One is supposed to be even better, but the price is quite a bit higher as well. Looks like I'll have to save some pennies and get at least something so that I can tinker and learn some more.

Careful on using "saturation" to describe these characteristics!

As for the CC filter, you could do it, but it would have led to about 1400:1 CR (the red level was at 74.6% and green was at 106.3%. Thus, the filter would need to have cut green by 30% to bring it level with red). I am currently sitting at about 1800:1, so as I indicated above, my goal after Run 4 was to get back as much of the CR as I could. In this case, it was without resorting to filters.


Jeff will probably disagree with me, but in so long as your expectations are set appropriately ahead of time, that would be a good approach in my book!

Later,
Bill

Figure 7 is my "base" reading, and if you want something really messed up, how blue "came up" relative to red and green from run 3 to 4, as well. I really have no explanation for why that happened. It could have been residual issues from mucking around with the DLP contrast and brightness, or it sould have been a measurement anomaly. Since I was focusing on getting red "flat", I was not too concerned about blue and green through run #3. (see below)

As for why I put figure 12 where it was: I wanted it to be all on one page, and that was the first place I could put a page break that did not destroy the flow or leave large amounts of white space. I'll try to get it closer to the measurement table in a future update. Great feedback, btw!


First, remember that the RGB levels are all relative. However, your guess is as good as mine as to why that happened. I had just mucked with some controls which I really should not have, and I did not know what to expect after that.

Do any real experts have any ideas?

Later,
Bill

Think of your RGB% chart as a see-saw - but with three seats instead of two.......

Hey guys. something i just noticed while reading thru the calibration guide. am i supposed to adjust color & tint before doing a full grayscale, gamma calibration? i adjusted my master brightness & contrast controls in the user menu first, but i'm not 100% sure about whether i'm supposed adjust master color & tint before grayscale/gamma.

thanks guys.
Carmine.

color & tint has no impact on greyscale.

Don't be too sure they're the only ones reading it. I found it to be incredibly informative, though I'm a novice at best so just learning and learning some more.

I just read Ursa's guide and I must say it is very informative, great work Ursa. I picked up on a lot of information that made me understand things instead of just accepting them and not knowing why they are the way they are. Keep up the good work!!

I would be more than happy to put it on the disc with Bill's permission. I'm trying to wrap that little snowball up over the holidays. Then maybe attempt a PAL version (I never learn ).

Scott - A fellow Houstonian AVSer, Tom Green, was nice enough to edit a slightly different version for me, so if you can give me a bit of time, I'd like to get a new revision out. We should take that discussion offline, though.

Later,
Bill

Understood. That would be awesome if it works out.

First, I want to say thank you so much for this manual. I know the amount of time it involves, I wrote one a few years back for the Sony 12xx projectors and it was a shwack of work.

I've just retubed my Ehome 8500 and am just starting to read you guide in preparation for calibrating it.

Again, thanks!

Wes

Wow. I've had only a couple of minutes to briefly skim the pdf, and already I can see that what little theory I thought I understood about calibration needs to be unlearned. Thank you so much for creating this document; it's a godsend.

Thanks! Bob's already asked a few questions, so if anything is unclear or confusing, please ask. I'd like to use this thread to help make the subsequent revisions (no promise on timeline) better.

Later,
Bill

Alright, now that we are out in the open, I'll defer the discussion of Gamma for a bit to tackle RGB balance and white points.

Picture, if you will, the CIE 1931 Gamut chart. For each standard, there is a defined location for each of the three primaries. However, for all current video standards, the same white point is used (D65, or x=0.313, y=0.329). Since the 1931 Cartesian system does not include luminance (Y), all points in the grayscale should hit D65. After all, since the difference between black and white is the amount (intensity) of light, the "color" of each point in the grayscale does not really change.

Let's emphasize the key point here: all points on the grayscale should be on D65.

Think of intensity, or luminance, as being orthogonal to the xy plane.

With me so far? Alright, then there is one additional piece to the puzzle: not only does each standard have a different defined point for their primaries, but they also have a different balance to the primaries that constitute white light. This is why you just can't calibrate to one standard and think it will work for all standards. Of course, for display manufacturers, this also drives you nuts because at a fundamental level, the color of a primary is going to be relatively fixed (e.g., dichroic filters, phosphor, etc. all have one color!).

So, where does that leave us for color balance? There are two issues at play here. The first is that there is a relative mix of red, green and blue that should be maintained. Second, this mix should be maintained as close to the defined white point as possible. Unfortunately, because of the pre-defined color issue mentioned above, you often cannot satisfy both of these two goals at the same time.

So, the color balance charts in my guide refer to the relative levels of red, green and blue that are required to bring the colors into balance given the defined mix for the HD standard (Green: 71.5%, Red: 21.3%, Blue: 7.2%). Why don't I look at my absolute deviation, rather than this relative deviation? Because to me, it is more important that the colors are balanced, than that they are at the right white point. Also, I can use both the color temperature and color error to tell me how far off of D65 I am, and how noticeable it will be.

Submariners have an expression about driving a sub. Bad sub drivers "chase the bubble", and as a result, they are always putting in steering corrections based on where the instruments said you were, rather than based on where they are indicating you are going. For me, focusing on the absolute deviation is a lot like chasing the bubble. Algae in Theodin's beard bugged the snot out of me. Slightly blue snow does not.

So, back to Bob's original question: are there several mixes of RGB that will meet D65 and be properly balanced for a given standard? The specific answer is no. However, the nuanced answer is that there is an error tolerance around this (dE), and I aim to make sure the balance is correct more so than hitting the theoretical level for a primary to be at D65.

Now if that is perfectly clear, then I definitely need to pay for the second course of Alan Greenspan's Guide to Business Communications!

Later,
Bill

Drapp, can you post your final setting for your h79 after calibration?

Jim, I could do that although the settings wouldn't necessarily be applicable to your situation. The general pattern that has cropped up with the H7* series is described well in the following link:

http://www.hometheaterhifi.com/volum...05-part-2.html

The main issues that seem to be pretty consistent with the H7* pjs are excessive red in lower IREs that is relatively easy to correct, and the drop in red at higher IRE that is less easy to correct without overall loss in contrast due to the necessary accompanying drop in green and blue to bring one back to D65. In my early calibration runs I'd create a red bump around 80 IRE that came when trying to turn up red to compensate for the 100 IRE fall-off; plus, red clips overall when turning it up too much. Steve Smallcombe used a red filter to correct the upper IRE red falloff, but lost 35% brightness and had to then use the brightmode to compensate. This approach maintains contrast.

With my new bulb I've used compromise settings in which I've turned down red until the fall-off is less severe at higher IREs, then turned down blue and green further, so my settings are something like R contrast - 6, B contrast -13 and G contrast -14, with TV and gamma 1. My regular contrast may be around 4 and and brightness 1-2, but this is from memory (I'm not home right now.) I have a High Power screen so I don't perceive a brightness problem now. I might try a CC10R and do a re-calibration, turning up my B and G contrasts to regain some overall contrast/dynamic range.

Dan

Ursa,

What a great job....this should be a sticky!!! How do you make that happen??

Larry