New calibration disc

Cool! Any plans for adding patterns similar to 3no's black bar and white bar patterns? You can find his DVD patterns in the HTPC forum.

http://www.avsforum.com/avs-vb/showt...&&#post5413832

Scott - Can you post what specific values you are encoding for grayscale in either RGB or YPbPr form (IRE only has meaning if you are authoring analog)? Also, can you add 100% primaries and secondaries?

Alternatively, I'd love to discuss beta testing as well!

Later,
Bill

this sounds great, I can't wait til it is released.

Scott - No big dog here! I'm just a humble bounty hunter (If you've seen Cowboy Bebop).

For the patterns I use, I set my reference white to 236 to handle some of the issues which you have discussed (22 code values between each decile of stimulus in RGB). Getting IRE to work out evenly to %stim in a NTSC environment requires 10-bit integer math (the last two bits are "decimalized" add-ons to the 8-bit values in some uses; however, we consumers get none of that!). Despite what would be convenient, we just can't assume away set-up in the playback environment (sorry to rant unnecessarily - it's a pet peeve).

As for run-out, I specifically want to try to provoke it. When I calculate my 1931 gamut, I capture the luminance information to calculate a dE between the actual primary and the expected value. When run-out happens, the dE will spike if the colors are accurate. Ironically, my H77, as delivered from the factory, had a significant run-out in red. However, it is also significantly oversaturated. As a result, the dE for 100% red was 14. If the red did not run out, the dE would have been over 30!

Later,
Bill

Scott, your disc sounds great - I hate navigating the other ones.

Heh, glad you mentioned red runout Bill - my Benq 8700 has run out of red too - 800 hours isn't kind on the bulb!

Did you author your disc using 8 bit RGB patterns? I hope it doesn't impact the YCrCb conversion when encoding - quite a few folks now have HDMI upscaling DVD players that can output 4:2:2 and 4:4:4 YCrCb (I hope I am using the right terminology) and I wonder if it would be visible? Hopefully not!

Anyways, I can't wait for the beta...

This sounds great. Can't wait to see the final product. I have a 52HM84 and since this disc is geared towards DLP sets, this will be very handy.

RGB->YCrCb conversion is done using a matrix that varies depending upon the standard used (602/SMPTE-C, 709, etc.). An HTPC using VMR should be able to validate that the RGB->YCrCb->RGB conversion has been done correctly. The problem is that I am not sure what matrices are being used by TT/ZP for high def. I am pretty sure that they do not actually re-convert when outputting at above-480p resolution.

Of course, these issues apply to the CE universe as well, and let's not get into the source of error in the mastering chain. (Per Stacey Spears, IIRC, one should probably use the HD matrices exclusively... )

Later,
Bill

Just curious...how is a calibration disc geared more towards DLP than versus say a plasma? I ask because Avia/DVE are not geared for one technology...is there a reason this would not work for a plasma?

Plasmas will be fine. What you have to worry about are the patterns that are designed to test specific design compromises/weaknesses in CRT-based displays (e.g., line bending due to HVPS issues, etc.). A digital is a digital in so far as that goes. What may make a difference is whether you are using an analog or a digital signal. D/A and A/D conversions can introduce visible errors if done poorly.

One could do some low light patterns to show DLP dither, but I'm not sure that would be that desirable! (Oddly, my H77 has greater issues with a 5% Stim pattern, than the 0% stim pattern!).

Later,
Bill

Does not make sense - as the component values to dE (dL*du*dv*) are magnitudes squared not +/- amplitudes. Your post implies that a negative dL* (runout) lessened the dE which is not possible.

Kevin - It was counterintuitive to me as well. I expected dE to go down when grayscale error (dL*) was eliminated. What happened in the calculations:
- dL* went to zero
- du* increaased significantly (6.5 -> 31!!)
- dv* decreased a bit

As near as I can tell, the error in red becomes more pronounced the brighter it gets (L* is a scaler for u' and v', after all). At least, that is the way the math is working (barring an error - going to have to bug hunt that one).

BTW, I am using the 1976 method here, with all of its flaws.

Later,
Bill

Ursa

Are you talking about measuring the Red Primary? Are you trying to fix the video red push, the greyscale red push, or the oversaturated Red chroma?

Are you using the measured white point or the standard white point in your calculations? If you change the reference - then these dE have nothing to do with each other since they are w.r.t. different references. So you should stick to the standard white so everything is on the same scale.

Kevin - Yes, I am talking about measuring the red primary. I'm not trying to do anything with it yet (at least, not until after dark), just describe how accurate it is at 100%. The surprising answer from a dE perspective is that it is less inaccurate now, than if it did not run out of light.

The reference is to the established primaries in Rec709.

Later,
Bill

Hi guys,

I just wanted to drop in to tell everyone that Scott has worked really hard putting this all together, and though the disc is quite simple to navigate and is not chock full of patterns, it is the single most useful calibration disc I have ever seen or used...period!!

This project was born out of our disgust of the horrible navigation that we must endure when using any of the other commercial discs. These are the ones that I know of and their problems (I own all of them except Avia Pro):

1. DVE consumer - horrible navigation and incomplete set of gray stimulus windows
2. Avia consumer - bad (but not horrible) navigation, but levels were affected in the encoding process, making them useless (for Scott and I).
3. S&V Home Theater TuneUp - corrected encoding problems, but gray windows are incomplete (not enough per cent stimulus)
4. DVE Pro - has all of the patterns correctly encoded, but once again is horrible to navigate - and it costs $300!!
5. Avia Pro - I didn't buy it for $400, but I understand that it has all the right patterns in a somewhat better menu setup, but do you really want to pay that much just for a complete set of patterns that should been included on the consumer versions?

Scott's disc is much easier to navigate than any of them! That's because it is not designed with fancy intros and a ton of other patterns that you will probably never need or use. Those patterns are still available - just buy one of the commercial discs! Scott's disc is all "meat and potatoes" - a very small disc with just the good stuff that you will need...

Scott designed all his own patterns, and though they resemble others (how many different ways are there to create a window?), some are unique and more useful than any others I've seen, especially his version of the "moving bars" brightness and contrast patterns. Scott created a nice main menu where you can access any group of patterns directly, but if you just start at chapter one, each pattern is in a moving video endless loop, so you just have to hit the "next" or "previous" button on your remote to move forward and backward through the sequence. The main patterns are all arranged in the exact order needed for UMR's i1 Pro DCS software (my idea... ), so taking a complete profile now only requires a couple of minutes, as you don't have to search for any patterns.

The 75% values were suggested by Jeff (UMR) and thus we went with them. (I use the term "we" loosely - Scott did ALL of the work, while I just made the suggestions and did a lot of testing.)

Scott went through hell and lots of time and work to insure that all levels remained correct, as apparently the MPEG encoding process likes to play with color levels. After a couple of weeks of failures, Scott got everything perfect, with the levels measuring perfectly using the "eye drop" method in MS Paint and by using DGIndex. We came to these conclusions by carefully comparing DVE commercial patterns to make sure that the values matched. Scott even found some very slight off values in the DVE patterns and corrected them in his. As a final check, we would like to have some of the "big dogs" (as Scott calls you) take a look at the patterns and provide one last check on his work to insure that all levels are 100% correct.

Just to be clear, once again I am not taking ANY credit for this other than for making suggestions and testing. This is ALL Scott's very hard work and he deserves a ton of credit!

So... I guess my dumb question is... How do we get a copy of this Disc?

I just bought my first HD Plasma (Panny 42px50u), and have started to research calibration discs.

This sounds like the perfect thing for a n00b. I will still be doing more research however.

Thanks!

This is a pretty handy mini-calibration disc, one thing that I think is pretty important to include is some resolution patterns/wedges etc, because especially if you're trying to 1:1 pixel map to a digital display you need to be able to check this for instance.

GetGray,

What most of us (in PAL land) are looking for is a PAL disk just like this. DVE PAL is the only game in town these days.

Any chance a PAL version (720x576/50HZ) is also in the works?

BTW, are you going to publically offer an ISO of the disk or will it eventually be for sale?

GetGray,

Everything in PAL is identical (the color encoding is different, but the MPEG encoder should take care of that). There is below black in PAL as well, it uses 50HZ and there's no 3:2 (there's only 2:2 or video). It's also 720 x 576 instead of by 480.

From an encoding standpoint, those are all the differences...

You can't mix PAL and NTSC on the same disk, so it would basically have to be a PAL version and an NTSC version (just like DVE).

Scott - Definitely kudos are in order on the disc. Two thumbs up! I sent you a few requests for labeling enhancements that I would find useful (e.g., digital values on all of the window patterns). Also, TT may be letting me down on the window color patterns.

If you would like, I can send you my full-field patterns that are corrected to require no more than a 50:1 instantaneous CR to display properly.

Later,
Bill