or Connect
AVS › AVS Forum › Display Devices › Display Calibration › Spears & Munsil HD Benchmark Blu-Ray 2nd Edition
New Posts  All Forums:Forum Nav:

Spears & Munsil HD Benchmark Blu-Ray 2nd Edition - Page 10

post #271 of 636
Stacey or Don,

I've discovered banding on all of the RGB ramp examples contained on your program. The only method I can find to eliminate this is to have the Blu-ray player output RGB @ PC level, (which of course is otherwise undesirable). No other setting (on the Blu-ray player or on my display) seems to have any effect on this. While experimenting with this, I've found the banding appears to be completely a result of the luminance portion of the signal, as significant banding is completely visible in Red and Blue ramps with the color control on the display reduced as far as it will go (black and white image). Banding in the green ramp is significantly more subtle. I originally thought that the banding was the result of some error in the way my display was processing YCbCr (quantization or rounding errors). However, I have found that reducing the color setting on the Blu-ray player (Oppo BDP-103) to it's minimum (with display color set normally) also results in the banding being obvious the Red and Blue ramps displayed in black and white. Viewing in black and white (color reduced at the blu-ray player), Y Cb and Cr ramps appear smooth without any obvious banding and R G and B all exhibit banding. I am assuming that with the Oppo set to a -16 color saturation that it is outputting no (or very little) Cb or Cr signal. Would you consider this an indication that whatever is causing the banding on the RGB ramps is occurring in the Blu-ray player? Or is there still some way that this could be something in the processing on my display?

Thanks,

Brett

Red portion of Monotonicity RGB - Saturation set to -16 on Oppo


Red portion of Ramps Double RGB - Saturation set to -16 on Oppo
post #272 of 636
Converting between YCbCr and RGB is not lossless. In order to optimize each channel for the best possible ramp, we manipulate the other channels up and down to ensure we get the expected value back. This all happens below 16 and should not be visible when brightness is set properly. Some displays are doing something odd with values below 16 which make the banding stand out when brightness is set correctly.

My VT20 has two picture modes that produce the correct ramp and the rest of the picture modes produce the banding you see.

At this point we have not root caused what the display is actually doing wrong. We do know that when color conversion and brightness are properly implemented no banding occurs. I have seen this banding on the JVC projectors as well as my VT20 in some picture modes. One day we will probably sit down and figure out what the display is doing that it should not and write an article on it so manufacturers stop doing it like we did for the chroma bug and the color conversion shortcut.
post #273 of 636
You might try turning off processing options and changing display modes to see if you can get the banding to go away. If you see that banding, it generally means that some kind of non-linear "enhancement" is being done in the display, and it's messing up the ramp. Or possibly it's just a bug. That's part of the reason we built the ramp the way we did; it's a perfectly legal ramp that lots of displays show perfectly. But some displays do odd things with it. Whether it's a bug or an attempt to enhance the image, it's not correct.
post #274 of 636
Thanks for pointing out those patterns as I have been interested if my eeColor 3dLUT box produces any banding artifacts so I gave it a test but need help in interpretation. I'm using a PS3 to play back the test disk with the following results using the double ramps:

1. eeColor off - PS3 set to YCbCr output: No banding present in either RGB or YCbCr test ramps.
2. eeColor on - PS3 set to YCbCr output: some banding present in green channel of RGB, no banding present in YCbCr test ramps.
3. eeColor on - PS3 set to RGB limited: no banding present in RGB ramps, no banding present in YCbCr test ramps.

comments?
post #275 of 636
Does eeColor work in RGB or YCbCr? If it works in RGB, it might be best to send it RGB.

In RGB limited is the >235 stuff still present? I don't recall if the Super White setting applied to YCbCr only or both.

How does YCbCr compare to RGB on the other color space tests?
post #276 of 636
Thread Starter 
Quote:
Originally Posted by sspears View Post

Does eeColor work in RGB or YCbCr? If it works in RGB, it might be best to send it RGB.

eeColor is working in RGB internally.
post #277 of 636
Quote:
Originally Posted by sspears View Post

Does eeColor work in RGB or YCbCr? If it works in RGB, it might be best to send it RGB.

It works in RGB (10 bits/channel) but it does not expand YCbCr to full-range so it preserves BTB/WTW and then sends YCbCr back out.
Quote:
In RGB limited is the >235 stuff still present? I don't recall if the Super White setting applied to YCbCr only or both.

No, BTB/WTW is clipped in both limited and full-range output from the PS3
Quote:
How does YCbCr compare to RGB on the other color space tests?

The same results were obtained for the other ramp options, monoticity and ramps horizontal/vertical.
Edited by zoyd - 6/13/13 at 2:51pm
post #278 of 636
If head/toe are clipping with RGB out, then I would use YCbCr. I thought with Superwhite enabled that RGB also passed both. Its been years since I used RGB out from a PS3.
post #279 of 636
Quote:
Originally Posted by sspears View Post

If head/toe are clipping with RGB out, then I would use YCbCr. I thought with Superwhite enabled that RGB also passed both. Its been years since I used RGB out from a PS3.

Thanks Stacey, yes - superwhite is only an option for YCbCr.
post #280 of 636
Quote:
Originally Posted by sspears View Post

Converting between YCbCr and RGB is not lossless. In order to optimize each channel for the best possible ramp, we manipulate the other channels up and down to ensure we get the expected value back. This all happens below 16 and should not be visible when brightness is set properly. Some displays are doing something odd with values below 16 which make the banding stand out when brightness is set correctly.

My VT20 has two picture modes that produce the correct ramp and the rest of the picture modes produce the banding you see.

At this point we have not root caused what the display is actually doing wrong. We do know that when color conversion and brightness are properly implemented no banding occurs. I have seen this banding on the JVC projectors as well as my VT20 in some picture modes. One day we will probably sit down and figure out what the display is doing that it should not and write an article on it so manufacturers stop doing it like we did for the chroma bug and the color conversion shortcut.

Thanks Stacey!

The manipulation of data values below 16 (particularly in RGB channels prior to YCbCr conversion) seems to make sense in relation to the results I am seeing. I am only getting completely band free ramps with RGB PC Level output which is stripping the below 16 data before delivering it to the display, so that somewhat seems to fit. I can also see how the variation of values between 0 and 15 on one or two of the RGB channels when the third channel is at a high value could result in variances of the resulting Y value. If I am understanding this correctly, it would completely explain the banding I am seeing in the black and white image with the color saturation reduced. This banding is certainly visible, but very minimal with the color saturation set correctly. If the manipulation of these below 16 values are being done with the assumption of 8-bit per channel processing from YCbCr to RGB, is it possible that the visibility of the banding is the result of the YCbCr to RGB processing being accomplished at higher than 8-bits per channel? I'm fairly certain that the processing in my display is 10-bits per channel (possibly 12).

Quote:
Originally Posted by dmunsil View Post

You might try turning off processing options and changing display modes to see if you can get the banding to go away. If you see that banding, it generally means that some kind of non-linear "enhancement" is being done in the display, and it's messing up the ramp. Or possibly it's just a bug. That's part of the reason we built the ramp the way we did; it's a perfectly legal ramp that lots of displays show perfectly. But some displays do odd things with it. Whether it's a bug or an attempt to enhance the image, it's not correct.

Don,
I've changed every option I can find on the display, including restoring two service menu settings (for convergence and gamma) to their factory settings. So far I can only eliminate the banding by setting the Oppo to output RGB at PC level. I've found nothing else that even has any effect on the banding. (Setting to 30 or 36 bit output on the Oppo does smooth the banding slightly, but doesn't reduce it.) Considering what Stacey said above, and my speculation about it, I'm going try setting the Oppo to RGB Video Level and turn Deep Color Off. That's a combination I'm not 100% sure that I've tried before. That's the only thing I can think of that might result in the YCbCr conversion to RGB being processed at 8-bits, but I'm not certain that will prevent the Oppo from processing the conversion at a higher bit depth.

Thanks!
Brett
post #281 of 636
Quote:
is it possible that the visibility of the banding is the result of the YCbCr to RGB processing being accomplished at higher than 8-bits per channel?

No. Each 8-bit value is multiplied by 4 to convert to 10-bit. 16 becomes 64, which is reference black in 10-bit.

I am running my OPPO with deep color on, 30-bit dither into my Samsung display. As I said, on the VT20, the banding is present in some picture modes and not other. Custom and THX are banding free on the VT20.
post #282 of 636
Quote:
Originally Posted by sspears View Post

No. Each 8-bit value is multiplied by 4 to convert to 10-bit. 16 becomes 64, which is reference black in 10-bit.

I understand that that there is no change simply by up-converting from 8 to 10 bit/channel. What I was wondering about is if rounding or truncation makes any difference when the RGB values that result from the YCbCr conversion are not 8-bit integers. For example, 10-bit/channel processing from YCbCr to RGB somehow results in an R value of 66. At 8 bit this would be equivalent to 16.5 and would have to be rounded up or down.

I believe that my display is converting all RGB input into YCbCr as the color and tint controls remain effective regardless of colorspace setting on the Oppo. I did try RGB Video level output at 8-bit, and it still has the banding. Apparently the YCbCr to RGB conversion in my display is not 100% accurate as the visibility of the banding in the Y-channel (which I believe is the result of the manipulation of below-16 data as mentioned in a post above) is reduced with increase in color but never completely eliminated with any color saturation setting. Perhaps Cb and Cr are not being applied linearly to Y' in the RGB conversion.

Anyway, thanks again Stacey for the explanation of the below-16 data manipulation on the ramps. I at least think I understand now why I'm seeing the banding on the black and white Y-channel portion of the ramps when I turn the color controls down.
post #283 of 636
Just one more thing I noticed about the Y-channel banding I've mentioned in my past few posts. I know it is not a normal procedure to reduce the color saturation when viewing these patterns, but I was curious what effect it would have on the ramps and the banding, and I tend to like to experiment and see if I can derive any information from a pattern beyond what it was primarily intended for. (Which I've found the HD Benchmark Blu-ray is excellent for. I'm having a ton of fun with it.)

On the "Contrast Blue" pattern, if I turn the color saturation down completely (on either the Oppo or the TV) the brightness of the numbered blocks appears to increase gradually from 231 to 244. 244 appears to be at the same level as the rectangle surrounding the numbered blocks. 245 appears much darker than 244. Brightness again gradually increases from 245-253. 253 and 244 appear to be nearly the same brightness.

On "Contrast Green" brightness appears to gradually increase to 250, and from 251 to 253. 250 and 253 appear nearly the same brightness as the background rectangle.

On "Contrast Red" brightness appears to gradually increase to 248, from 249 to 251, and from 252 to 253. 248, 251 and 253 appear nearly the same brightness as the background.

If my interpretation of the manipulation of below 16 values on these patterns is correct, then it explains these results.

With color set to its normal setting, all three patterns appear correctly, showing gradually increasing brightness from 231 to 253 with only 253 appearing nearly the same brightness as the background.
post #284 of 636
Quote:
Originally Posted by sspears View Post

Try this www.handforgedvideo.com/sspears70/v2insert.zip
Can the booklet still be found somewhere?
I didn't find it on the spears & munsil Webpage.
post #285 of 636
Quote:
Can the booklet still be found somewhere?

http://www.spearsandmunsil.com/pdf/HDBenchmark2_Booklet_print.pdf
post #286 of 636
Thanks for the link.
Now I have seen that all menue-entreis are on the homepage ...
Looks great - I have just placed my order.
post #287 of 636
If the S&M disk reveals banding in the video processor tests of a display or PJ then what would be the point of adding an external video processor?
td
post #288 of 636
Quote:
If the S&M disk reveals banding in the video processor tests of a display or PJ then what would be the point of adding an external video processor?

Are you asking if a display has banding, why would you insert a video processor into the mix? If so, better scaling and deinterlacing would be one reason. Its CMS in a video processor where you need to be careful. This is where banding may be introduced if you are adjusting the gamut and not enough precision is used or the adjustment is too large.
post #289 of 636
Quote:
Originally Posted by tucsondave View Post

If the S&M disk reveals banding in the video processor tests of a display or PJ then what would be the point of adding an external video processor?
td

A processor could conceivably improve the situation. For example, if the banding is happening in the conversion from Y'CbCr to RGB (which it usually is), and the processor does a better job of converting from Y'CbCr to RGB than the player does (or the player doesn't support RGB output), then a processor could eliminate the banding. And of course a processor could make improvements in other areas: scaling, sharpening, deinterlacing, etc.

However, there's no guarantee that a processor is going to make things better. It all depends on the specific equipment in question.
post #290 of 636
Quote:
Originally Posted by sspears View Post

Converting between YCbCr and RGB is not lossless. ...........

FWIW, my copy of your new calibration disk was despatched last week, but I haven't received it yet.

Until it arrives, I've been reading articles regarding colour space on your website. What caught my eye yesterday was the topic of colour space conversion from media to display.

My relevant kit:
  • Panasonic TX-P50GT50 plasma display
  • Sony BDP-S790
  • Denon AVR-3313

On your website, you state that all displays use RGB to drive display. I currently have BDP set for YCbCr4:4:4 and display is set for "1080P Pure Direct" mode. AVR is configured to pass video signal through unaltered.

Based on information on your website, it would seem that display is converting YCbCr4:4:4 to RGB4:4:4. Should I reconfigure BDP to send RGB4:4:4, thus avoiding any sort of loss and doing the required conversion at one point?
post #291 of 636
I would start with YCbCr since it is usually the best. There is a good chance your display will convert RGB back into YCbCr for processing and then back to RGB for display. It's this step you want to avoid.
post #292 of 636
Quote:
Originally Posted by Iain- View Post

FWIW, my copy of your new calibration disk was despatched last week, but I haven't received it yet.

Until it arrives, I've been reading articles regarding colour space on your website. What caught my eye yesterday was the topic of colour space conversion from media to display.

My relevant kit:
  • Panasonic TX-P50GT50 plasma display
  • Sony BDP-S790
  • Denon AVR-3313

On your website, you state that all displays use RGB to drive display. I currently have BDP set for YCbCr4:4:4 and display is set for "1080P Pure Direct" mode. AVR is configured to pass video signal through unaltered.

Based on information on your website, it would seem that display is converting YCbCr4:4:4 to RGB4:4:4. Should I reconfigure BDP to send RGB4:4:4, thus avoiding any sort of loss and doing the required conversion at one point?

Quote:
Originally Posted by sspears View Post

I would start with YCbCr since it is usually the best. There is a good chance your display will convert RGB back into YCbCr for processing and then back to RGB for display. It's this step you want to avoid.

Thanks for the information.

What's confusing of this is Panasonic infer by their literature that if "1080P Pure Direct" is enabled, then display will be able to take either of the 4:4:4 video feeds direct. rolleyes.gif

Reckon I'll wait till disk arrives and as first step, see which one looks better initially.
post #293 of 636
1080 pure direct could mean any number of things. smile.gif
post #294 of 636
Iain
I have GT50 + Oppo 103 and after lots of testing,measuring with chromapure + i1display3, and using Spears & Munsil disc ver1, (second edition is on the way from Audiocom) GT50 is best working if Oppo is outputting
YCbCr 4:4:4 and 1080p Pure Direct is ON on TV.
post #295 of 636
Quote:
Originally Posted by prsut View Post

Iain
I have GT50 + Oppo 103 and after lots of testing,measuring with chromapure + i1display3, and using Spears & Munsil disc ver1, (second edition is on the way from Audiocom) GT50 is best working if Oppo is outputting
YCbCr 4:4:4 and 1080p Pure Direct is ON on TV.

Thanks for the information. Seems I've been doing it right all along, then. smile.gif

BTW, I'm considering the SpectraCal C3/Calman 5 combination, should I need it. How would that compare to your kit?
post #296 of 636
Quote:
Originally Posted by Iain- View Post

Thanks for the information. Seems I've been doing it right all along, then. smile.gif

BTW, I'm considering the SpectraCal C3/Calman 5 combination, should I need it. How would that compare to your kit?

Just my 2 cents smile.gif

The C3 is a pretty good meter but it does not have sealed optics like the I1d3, consequently it will drift more easily over time.
post #297 of 636
I do not know Calman's products (only from demos), I have learned with HCFR + i1displayLT. Then I bough chromapure + i1display3.
You know, best photographer's friend is trash and your best friend will be articles like well known Kal's article http://www.curtpalme.com/forum/viewtopic.php?t=10457. I believe it is not important if Chromapure or Calman - it is personal preference, but reading and learning and then trying. And then again.
It is joy and fun and benefit is, you will learn a lot about your TV. Do not think, your first calibration will be perfect (although numbers and graphs will be perfect), but then, after lots of reading and asking here in calibration forum you will love your set.
If I put Return To Forever to my Oppo, not only fantastic sound to my Denon's headphones (and Graham Slee headamp) but best possible picture I can imagine. So investment to meter and software is well worth of every cent, at least for me.

And at last , because we are here in Spears & Munsil disc, you need to set equipment for best combination of rgb's, 4:4:4, pure directs and other hard to understand settings. Nothing is better and I got it free with my Oppo smile.gif
post #298 of 636
Quote:
Originally Posted by sspears View Post

Gamut comes in 25%, 50%, 75%, and 100% saturation. The 25, 50, and 75% also include a 100% white.

Stacy, is that correct, or did you mean amplitude/intensity? I sat down for the first time today and calibrated my VT50 with the equal energy windows. When doing the CMS work, it appears to me the 25/50/75/100% options were intensity, because saturation was the same result in each menu (100%). I usually calibrate the VT50 to 75% intensity/75% saturation, and thats why I noticed. Never the less I went into the 75% menu for Gamut since the other option is to do 75% intensity/100% saturation.

I tested the color theory by calibrating to 75%, then went into the 100% menu and read 100% white and 100% red. Red landed spot on the 100% saturation mark, but had a huge luminance error. If Im thinking correctly, that means the patterns vary in intensity not saturation???

I really liked how the windows behave and the quick changes while calibrating, also the Gamma results actually matched the VT50 Gamma menu setting, which you never see on normal 5% and 10% windows. With a gamma setting of 2.4 in the menu, I only had to adjust 10% some to get it flat, 20-90 were just about spot on. In initial viewing I like the result, and the grayscale adjustments were a little different from what Ive seen with 10% windows, which was interesting as well. Hopefully it continues to shine with further evaluation, Ive never used any type APL pattern before now.

Here are the results:





post #299 of 636
Quote:
Originally Posted by gadgtfreek View Post

Stacy, is that correct, or did you mean amplitude/intensity?

It's level, not saturation - you're correct.
post #300 of 636
Thanks!
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Display Calibration
AVS › AVS Forum › Display Devices › Display Calibration › Spears & Munsil HD Benchmark Blu-Ray 2nd Edition