Calibrate w/out BlueRay - Page 2

That's what we use. Supports 4:2:2, 4:4:4, RGB and can read off HDCP encrypted data from the players without any modifications. It is also the worlds loudest and slowest pattern generator.

Yep. That's why I'm askin him this stuff, cuz I know he's got the good toys to check these things out.

My YCbCr green push tests on the S185, S380 & S390 could be wrong, because I only used one display for those, and had no accurate source to cross-reference with (yet?). Chris's S590 results make that a very likely possibility.

I'm sure you can also understand some of my skepticism though about this, given the rash of green clipping and push errors I've found in the RGB modes on the recent Sony BD players, which have been confirmed on 2 different displays in my tests, and also re-affirmed by Chris's RGB mode tests on the S590. To briefly recap, here's what we seem to agree on so far...

a Green is being pushed in the RGB HDMI mode with all three PQ Settings (Standard, Bright & Theater).

a The Standard PQ mode clips green at around 92-93% stimulus in the RGB HDMI mode.

a The Bright Room PQ mode clips green at around 91-92% stimulus in the RGB HDMI mode. (I'm assuming Chris's results on the S590 were similar, or he would have disagreed with this.)

a The Theater Room PQ mode shows no evidence of green clipping near white in the standard 16-235 video range in the RGB HDMI mode. (Chris shows some green clipping above video white on the S590 in this mode. But not below video white in the standard 16-235 video range. So our results also jibe on this.) However, neither of us recommend using the Theater Room mode because we're seeing other weird stuff with the colors, levels or gamma in this setting.

a There is no clipping near video white in the standard 16-235 video range in the YCbCr HDMI modes, with any of the PQ mode settings.

aYCbCr HDMI modes in Standard Picture Quality mode are generally the best way to go on these players.

And here's what we don't agree on...

r Green is being pushed in the YCbCr HDMI mode.

Chris says the YCbCr HDMI modes on the S590 are "bit-perfect", which means no sign of green push in those modes. My tests with a Gateway LCD (which has made the correct call on everything else above) seems to suggest otherwise.

If Chris's equipment says the YCbCR modes are 5x5 though, I believe him, which means either...

a) there's a problem with YCbCr processing on my LCD, or...

b) my PNG and DVD-Video tests are not as accurate predictors of HD Blu-ray video results in YCbCr HDMI mode as they are in the RGB mode, or...

c) the other players (S185, S380 & S390) are doing something different in YCbCr modes than the S590.

At the moment, I don't know which of these answers is correct. But I plan to pick up an S590 to see if I can reproduce his results with that player. If the color output looks the same in YCbCr HDMI modes as the other players on my Gateway display, that would tend to make me think the problem is in the display's YCbCr processing.


Agreed.

Thank you (sincerely) for clarifying that, Chris.

If you don't mind me askin, what is the software version installed on the BDP-S590 that you're using for your tests?

And also what is the 2-digit manufacturer's code on the back of the unit? (It should be something like 1C, 1D, 2A or 2B, probably located near the wattage rating.)

Using a 16-235 pattern, it clips in the 89-90% range. If 89 and 90 are different would be very iffy, but 91 and 92 would not be different than 90, that's certain, as it clips below 90.


Theater compressed the range, in that you get 236 in green with an input of 229 instead of 236. 16 = 16, but in there it's clipping 7-8 values in the range.


Theater mode doesn't clip because the range is so reduced that it's impossible for it to clip. It has a very small, limited range.


As covered, it's perfect. 100% perfect.


I'm voting #2 at the main reason, #1 as a secondary reason. As far as testing is concerned, the display is a big black box that does who-knows-what with the data given it. It could do a lot of processing, it could do no processing, the only way to know is to remove it from the chain.

I haven't read all the results on the Sony, I've had too much to do to do more than run the benchmarks, process it, and create the charts, but I haven't done in-depth analysis on it. I have to do that this weekend so I can get this review out the door.

Close enough. I only looked briefly at the Bright Room PQ mode, so my 91-92% values could be off by a percent or so.


Makes sense.


Fair enough. If you could include the software rev. and 2-digit mfr. code in the review (or post em here when you're done), that'd be very helpful. I'd like to make sure my S590 hardware and software is as close a match to your test unit as possible, if that's feasible.

The YCbCr HDMI performance is really academic though in my case because my main Sony CRT display where the player will be used only supports the RGB HDMI modes. So unless Sony can repair the issues in the RGB mode on one of their players, I'll either have to make do with some green push and clipping, or get a different brand, which I'd rather not do.

The S580 is still a possibility as well. But I haven't been able to retest one of those yet, and I'm not that keen on players with fans.

I have mixed feelings about firmware fixes as well, because if the RGB HDMI issues can be fixed in a software update, that probably also means they can be broken again later on by another software revision. Such is life though in the wide, wonderful world of Blu-ray high-def home video.

S590 is software version M12.R.0320. The back panel seems to indicate Revision 2A.

^ Thanks for this info, Chris. M12.R.0320 is the latest update, so that should be no problem. The "2A" on the back should mean your unit was built in the first quarter (Q1) of 2012. (2 = 2012, A = first quarter.)

I picked up an S590 (coded 1D) yesterday, and ran some tests on it this morning with the default software installed (M12.R.0097), and so far it's 1080i/p color output looks identical to the S185, S380, and S390 with all my PNG and DVD tests.

I also ran the same tests at 480p on both the S390 and S590, and there are no obvious indications of green push or clipping in the RGB or YCbCr HDMI modes on either player at that resolution. So it looks like both the players and my displays are working pretty ok at SD resolutions.

So far, all of my tests still seem to point to an issue with the HD output on the players. So if there is a problem with the YCbCr processing on my Gateway LCD display, it appears to be limited only to HD resolutions.

The green push looks the same in both the RGB and YCbCr modes at 1080p on my LCD display with these players. The only difference is the RGB HDMI mode clips green at 92-93% stimulus (in Standard PQ mode), and the YCbCr HDMI modes don't.

However, neither of my players (S390 or S590) have been updated with the latest software yet. And they were both mfr'd in Q4 of 2011 (code 1D on the back), which was probably the first quarter that these models were built. So they should make pretty good "control units" to see if any hardware or software updates effect the HD output.

I've downloaded the latest software for both players, and will install it this weekend to see if it makes any difference on my 1080i/p results. If not, then I'll try to find a "2A" S590 to test as well.

Where to get adu's color clipping test?

Right here. Since it was uploaded as an attachment, you must be logged in to see and DL it. Use at your own risk. If you own a plasma display, don't leave it on screen too long, or it may cause burn-in.



There should be no signs of clipping on any of the colors if both the TV and player are working correctly. If you do see clipping on any of the colors, try lowering the contrast on the display and turning off any dynamic contrast features to see if that fixes it.

If not, then the player could be the source of the issue, and you'll need to look at the HDMI color settings and/or Video/Picture Quality settings on the player for a potential solution. Deviating from the "standard" or recommended settings may have other detrimental effects on picture quality though, on some players (and the video/PQ settings may have not apply to still image files like this). So be advised.

The "clip point" is where the slightly darker shades of color (labeled 90 through 99) begin to blend into the 100% stimulus background color.

Installed the latest firmware (M12.R.0320) on my code 1D BDP-S590 Sony player over the weekend, and there was no apparent change to the HDMI color output. My LCD display is still showing green push in both the RGB and YCbCr HDMI modes, and green clipping in the RGB HDMI mode with the Standard PQ setting.

Maybe switching to a slightly newer player built in Q1 2012 (code 2A) will make some difference.

I finally had some time to do some testing on this, so here's what I did. I made a PNG file, 1920x1080, with ever RGBW value from 0-255 in there. I loaded this onto the Sony BDP-S590 and then checked to see if the output from it over RGB matches the output of the RGB test patterns in RGB mode. They aren't even close. Green clips at 218 from a Blu-ray disc, and it clips at 238 from the PNG. Red has a max value of 234 from Blu-ray, and a max value of 215 from the PNG. Blue has a max value of 244 from Blu-ray, and 225 from the PNG. Peak white is 234/234/232 from the test patterns, which isn't close to what it would be off the player.

I'll see if I can test the Oppo tonight, but it's getting late for me, but hopefully this will put to rest testing with an image file and why the results really have no bearing at all on what compressed YCbCr from a Blu-ray will do.

Even the Oppo, which is known to be perfect for Blu-ray, isn't quite right here. With RGB Video selected the values are way off, and with RGB PC they still aren't 100% accurate. So using a PNG to evaluate the player will lead to incorrect results, with no correlation to the accuracy of Blu-ray content, is what it looks like.

Thanks for takin the time to give this a try, Chris. I was hopin you'd try my PNG pattern so you could see visually what I'm talkin about. But it's interesting to hear what's goin on digitally on your tests as well.
 

(*My emphasis.)

Your figures above on the Sony S590 are consistent with my PNG and DVD-Video tests on the Sony S185, S380, S390 and (1D) S590. All of these players are currently clipping green at around the same stimulus level (about 92-93%) in the Standard/RGB HDMI mode. The reason we're not seeing eye-to-eye on this is that I'm thinking about the clip point in terms of % stimulus, and you're looking just at the "raw" or "naked" RGB codes (and you're also having some difficulty accepting that the Sony players could treat still images the same way as video files).

Both types of information are useful. However, if you really want a better understanding of why the player maps and clips different files at different codes, IMO, the % stimulus can offer some greater insight into that. If, for example, I say that green is being clipped at 92-93% stimulus, and the green color information above that is missing in both still images and video files, I think most people here will have a pretty good idea what I mean. And many will also be able to infer the approximate RGB codes from that.

However, if you look just at the RGB codes in a vacuum, without regard to what they represent (as you seem to be doing), then it'll appear as if the PNG still files are clipped at "different levels" than the video files (because the codes are different). This really isn't the case though, because the raw RGB codes don't take into account the difference in head and foot room between the different file types.

To try to make this a bit clearer, here's how I calculate the stimulus level for still and video files from the RGB codes...
 

The formula is really the same, regardless of the file type. What changes are the number of bits or codes, and the black and white points. Here are some common examples, for comparison...
 

File Type:	Bits Per Color Component:	Total # of Codes Per Color Component:	Black Code:	White Code:	Normal Color Coding Range:	Stimulus:
"24-bit" Still Images (PNG, JPEG, BMP, etc.)	8	256 (0 to 255)	0	255	0-255	(RGB Color or Clip Code - 0) / (255 - 0) or simply... RGB Color or Clip Code / 255
"8-bit" Consumer Video (Blu-ray HD video & DVD-Video)	8	256 (0 to 255)	16	235	16-235	(RGB Color or Clip Code - 16) / (235 - 16) or... (RGB Color or Clip Code - 16) / 219
"10-bit" "Full-Swing" Professional Video	10	1024 (0 to 1023)	0	1023	0-1023	(RGB Color or Clip Code - 0) / (1023 - 0) or simply... RGB Color or Clip Code / 1023
"10-bit" "Studio-Swing" Professional Video	10	1024 (0 to 1023)	64	940	64-940	(RGB Color or Clip Code - 64) / (940 - 64) or... (RGB Color or Clip Code - 64) / 876

This probably goes without saying, but to get percent stimulus from the equations above, you simply multiply the end result by 100, in all cases. Note also that the number of color codes (excluding black) in 10-bit studio-swing pro video is exactly 4 times as many as consumer video (219*4=876), which simplifies the downsampling/conversion process.

It's probably also worth mentioning that the consumer video RGB codes discussed here are really equivalents, since Blu-ray and DVD video are actually stored in 8-bit Y'CbCr format, rather than RGB. In 8-bit Y'CbCr, the Y' or luma component represents the brightness of a color and is normally coded to a range of 16-235. The Cb and Cr chroma components contain the color information, and are coded to a range of 16-240. Since Y' represents brightness, the nearest 8-bit RGB equivalents are generally limited to it's 16-235 range. In practice though, the values will frequently drift slightly higher and lower than that range due to overshoot. (More about Y'CbCr, for anyone interested. Still JPEG/JFIF files also uses Y'CbCr btw.)

Also, I'm not sure why the desktop imaging world refers to 8-bits per color component files as "24-bit" (or "32-bit", if there's also an alpha channel), and the video world refers to 8-bits per color component video files as "8-bit", but those are the conventions. And they really mean the same thing: 8 bits per color component (as shown on the table above), in both cases.

Anyway, if I plug your Blu-ray and PNG clip codes into the appropriate equations above, the result is about 92-93% stimulus in both cases...
 

Pattern File Type:	Clip Point Stimulus:	% Stimulus: Stimulus*100
"24-bit" PNG Still Image	238 / 255 = 0.9333	93.33%
"8-bit" Blu-ray HD Consumer Video	(218 - 16) / 219 = 0.9224	92.24%

I have some thoughts/questions on the other PNG/BD data you've collected as well, but I'd like to know if we can agree on this much first-- that green is being clipped at around 92-93% stimulus (as described above) in the Standard/HDMI RGB color mode with both the PNG and video tests.

Edited by ADU - 4/2/13 at 2:18pm

Re: the meaning of "stimulus"...

I think most folks here probably have a pretty good idea what "stimulus", "stimulus level" or "% stimulus" refers to in this context. But in case anyone's gettin hung up on those particular terms, here's what I mean by them...

In reference to the RGB color spaces used in video or still computer images, "stimulus" is the normalized value of red, green, or blue used to stimulate the different color receptors in the eye. "Stimulus" would be the values normalized to a range of 0 to 1, while "% stimulus" would be normalized to a range of 0 to 100.

Based on my reading,... in other color science circles (ie, not home video or video calibration), the terms "stimulus" or "% stimulus" are more frequently associated with the CIE XYZ color space. This is not however, what I am referring to above.

I think it's perfectly acceptable to use these terms with other color spaces though (especially of the tristimulus variety), provided the context is clear, and the functions are more or less analogous. Since it's much more common for people to think in terms of RGB when referencing home video content, hopefully it's clear (without me explicitely stating it) that I'm also referring to the RGB tristimulus values in this context, rather than XYZ.

If someone has a better term though, I'm open to using it. One could, for example, simply replace the terms above with "RGB value" or "RGB %". I prefer the term "stimulus" though because it's more descriptive, and not as easily confused with the coded "RGB values".

A rather long-winded explanation, but hopefully most of it makes sense.

Or there is a simple explanation: Blu-ray content and PNG content is handled differently, and you want your method to work, even though it doesn't appear to work.

You're overlooking many things here:

- Blu-ray is a 0-255 medium for encoded RGB values. You aren't supposed to use that whole range for consumer video, but you have the whole range
- The Sony is perfect for YCbCr data, so it can do 4:2:0 to 4:2:2 and 4:4:4 correctly
- The YCbCr to RGB step is broken
- The Oppo also didn't match up

So in your theory, the Sony is taking the RGB PNG file, converting it to YCbCr and then back to RGB to get the same error, and putting logic into this error that says "Clip at X% stimulus". Also, you have an error that is different by a percent. Not a single rounding error (say, 217.8 vs 238.4 where they are rounding the wrong ways to match up), but a variance enough to show that they are working differently.

The Oppo issue can't be ignored either. It doesn't output the same, since from Blu-ray it is perfect and from a PNG it is off. Why would this be the case, if it doesn't clip at any stimulus. The simple answer is "It handles PNG and RGB differently", and I have proof of that. I can give it the same RGB value from a Blu-ray and from a PNG and get different results, and so you are getting different values, and so they are being handled differently. Suggesting otherwise is now making so many leaps and assumptions that are much harder than this one that I don't see it as a reasonable explanation for it.

Hey Smack, what are the top 3 bluray players under $150 in general that you like overall?

Panasonic DMP-BDT210 / DMP-BDT215 (Costco)

The Panasonic 210 was my favorite last year but I haven't use the 220 yet. The Sony S590 is good but with a couple issues but that review is hopefully up this week. Hopefully I will see the Panasonic and a Samsung here soon. My recommendation last year was the Panasonic 210 for value and the Oppo for the best overall. I also really wanted to try one of the Dune players but they wouldn't provide me one.

In case anyone missed Chris's post in the Sony S390/S590 thread (or elsewhere), the green push/clipping issues in HDMI RGB mode were fixed on these players in a firmware update late last summer. Chris tested the update on an S590. And I confirmed that the problem was also fixed on my S390. There was a firmware update for the S185 around the same time as well. So that player's RGB mode may also be fixed. I haven't retested the S185 since the update to confirm it though.

 

After applying the updates, the S390 and S590 (and probably also S185) should have essentially perfect color output in the HDMI RGB mode... though you may want to re-check it with some of the patterns here, or with your own decoder tests, just to be sure.

 

I have not tested any of the new 2013 Sony Blu-ray players for color accuracy yet. And I still haven't organized any better YCbCr tests for my own players yet. So Chris and Secrets are definitely your best source for info on that (and probably the RGB modes as well).