"High Definition Benchmark" BD Edition by Stacey Spears and Don Munsil - Page 3

As posted earlier, there is an entire section on film deinterlacing (Source adaptive) and video deinterlacing (edge adaptive). These are available in three flavors: HD VC-1, SD VC-1 and SD MPEG-2.

The MPEG-2 SD is encoded using DVD constraints. This includes GOP length and peak bitrate. The VC-1 SD is encoded using BD SD constraints, which gives you a longer GOP and higher peak.

Couple the above with the hidden SD static patterns and it should allow you to evaluate your BD player, a processor or the display. This should help you find the best way to connect them. It will also help you in the future when it comes time to upgrade.

Stacey or Don,

What is the official release date for this title?

The disc is available now.

I was asking for the purposes of putting it on a database. I need the official date of release.

Let's go with 3/2/09 as the official release date. It shipped with the first batch of OPPO EAP players.

I just used my disc today with the ps3. In general I really liked it, I could actually see that the sharpness control on my xbr4 lcd really does do something, and found a good place to set it!

I have a noob type question that perhaps doesn't belong here but I'll ask anyway . I have used avia II and dve before, I am using the color filters from a DVE HD DVD I once had. When using the blue filter and adjusting the color and hue controls, I can set them so that the test pattern looks as it should (4 blue bars of similar intensity). But then, looking through the red and green filters shows that these are not quite right. If I then adjust the color and hue controls to get green correct, red and blue are wrong etc. Before, I just set the controls in some sort of comprimise position for all three colors, but the instructions for this Spears & Munsil disc never mention the red and green color/filter settings. Is the preferred method just to set the blue and forget how wrong the red and green look? I also mess around with the white balance red, green, and blue gain and bias controls for my TV, but these don't make much difference here. I basically just set the red gain so that the contrast pattern white bars do not look pink.

Anyway, I would appreciate any comments in this area. Thanks.

You want to set color and tint using only the blue filter. A blue filter is not 100% accurate. The only 100% method is for the display to offer a true blue only mode.

The red and green filters are to check for color decoding errors. The red filter in DVE works pretty well, at least on my display. The green filter does not work at all on my dispaly, which is the Samsung SP-A800B. I know this because the Samsung offers a red, green and blue only mode to verify all of the above.

Don may chime in and explain why green filters may never work.

I have a question about deinterlacing test. I tried the race car test on Oppo 83 this weekend and noticed that with some of the more complicated cadence, the pan of the race car was not very smooth. I did not see any moire pattern on the seats, but there was definitely lot of judder as the camera panned from right to left. Is this normal? Should the pass/fail depend on moire pattern in the seats or should it include judder?

The judder is normal. For example, 5-5 and 6-4 are 12 fps and 8-7 is 8 fps. These will have pretty bad judder. 2-2 is probably the smoothest.

2-2 was very smooth. Moire was seen in the seats for ~0.5 sec then the player locked on to correct cadence and solid from then on.

2-2 is one of the more difficuilt cadences to detect. The ABT solution usually drops on loop and then recovers.

It's true that the only sure way is to have a blue-only mode of some kind. But in fact you should be able to use a green-only or red-only mode and do the calibration and get the same results.

The whole point of the color bars is that the bars represent every combination of red, green, and blue at a single level (generally 180). So the white should end up as RGB (180,180,180) and cyan should be RGB (0,180,180), and red is RGB (180,0,0), and so forth.

So if you view through a blue filter, you are just verifying that all of the bars have the same level in the blue channel. If you view through a red filter, you'd be doing the same for the red channel.

Assuming that the display uses the standard conversion matrices for converting between YCbCr and RGB (which is not always a good assumption), then when color and hue are set right, all three channels should be at equal levels. If color or hue isn't set right, all three channels will be wrong. It's not possible to make the blue channel right and the red or green channel wrong, or red right and blue and green wrong, etc.

So blue got chosen to be the traditional channel to check color and hue settings. It could have been red or green. But you only need one, so BVMs have a simple mode switch that puts the monitor in blue-only mode.

Filters would be just as good as a blue-only (or red-only or green-only) mode if they perfectly passed one color channel and removed all of the other two. But in practice, no practical filter does that perfectly. There's always a little red and/or green that leaks through the blue filter. If it's a small enough amount, it has a negligible effect on the calibration. But if it's large enough that you can see it, there's no point in even proceeding with the calibration because it won't be right.

When you look through the blue filter, look at the red, yellow, and green bars. All three of them should turn black. If they show even a hint of red, yellow, or green, or show different brightness levels, the filter is passing some green and red, and you shouldn't bother calibrating unless the settings are waaaaay off and you just want to get them closer.

If you have all three filters, whichever one of them makes the black bars the blackest (and the most alike in brightness) is the one to use. The calibration procedure is the same no matter which filter you use.

Once you've calibrated with one filter, you can use the other filters to check that the color decoder is using the standard matrix, but again, since filters are notoriously unreliable, I wouldn't trust the results. If your projector allows you to turn off colors selectively, then those results can be trusted.

As Stacey mentions, the green filter is near-impossible to get right, because the color primaries of any practical display have overlapping spectra, and there is often a tiny little notch of frequencies that are only covered by the green primary and not the red and blue. The red and blue primaries only overlap one other primary each (green, in both cases), and thus have a larger range of frequencies that don't overlap.

Don

Few Qs
1. Is this as good as getting a ISF pro calibration done?
2. Will i need any other gear to help with the calibration ie "Eye One" ect.
3. Is it a nightmare for a novice like me to navigate like DVE for example?
Thanks Stacey

Ahhhhhh!!!!!!! Great! So since I do see NO light for the yellow, green, and red bars when viewed through blue filter, blue filter can be used to calibrate. However, I DO see some light from the magenta (a lot), red (fairly dark), and blue (a lot) bars when looking through green filter, thus I should surmise green filter not good enough (i.e. bandwidth too wide) to calibrate green with, and similarly I DO see light from the cyan (a lot) and green (a lot) bars when looking through the red filter, so red filter is not good either. Therefore blue should be used to set color. I assume with ideal narrow filters (maybe ~10nm bandpass) that 3 color bars should be BLACK when looking through any given primary filter, so would these filters help out with this, or is there just inherently too much overlap between the red green and blue spectra that the tv is outputting? (This is a kind of hypothetical question as I realize that these filters are a little bit expensive.) Anyway, this is so awesome, I never understood this before now--because I'm and idiot--and have probably spent a couple of hours in vain messing with color adjustments! Thanks Don and Stacey!

Edit: A comment about the geometry pattern; I notice that the left and top white border lines are 1 pixel wide, but the right and bottom borders are 2 pixels wide. Is this something my xbr tv is doing or is does the disc do this? This is no big deal, I was just wondering. I guess I could make a special wallpaper in paint for the ps3 to check this if I wasn't so lazy.

Also, I think it is great that the disc makers are responding so diligently to questions in this forum. It's like you buy a $25 cal disc and get FREE tech support. Awesomeness!

Nice find! It looks like we made a mistake on the pattern. Don, didn't you create this pattern? It should have had a border of Y 180, Cb / Cr 128. The top and left are correct, the bottom and right are not, unless this was intended. The outer pixel is Y 213 and the pixel to the left is Y 197.

I posted this on the BDP-83 thread:

Has anyone used the S&M disc to check player/display settings?

I've noticed that the results of the crop tests on that disc differ greatly from those on the DVE disc and HDNET test patterns (stored on DVR).

With S&M, when calibrating for optimal height/width, that will cause serious misplacement of DVD/BD movie source material. The horizontal is waaaaay off, with the image skewed about 30% to the right and vertical shows about 15% less black bar above, than below the image.

Going back to using either DVE or HDNET, I get perfectly aligned image with movie playback after calibration.

I've tried this with all resolutions, including Source Direct. There are variations, as expected, however - they all are visually wrong, when using normal BD/DVD content.

The S&M disc in my PS-3 is almost exactly the same as DVE and HDNET.

Am I doing something wrong?

Hi Steve,

Some questions for you:

1. How are the players connected to your display?
2. What resolution are the players outputting?
3. Are any pixels cropped on the four sides?
4. Which display?

When using HDMI, the PS3, Pioneer 09 and BDP-83 produce identical results into a Samsung SP-A800B and a JVC RS-2. I get 1:1 mapping and I see all 1920x1080 pixels on the pattern.

For DVE, is this the Blu-ray version? Can you tell me exactly which pattern you are using?

Can you also go to title 127 (have to use title search) and play the SD image cropping pattern there and see if it aligns with the HD version?

If you are using analog outputs, the timing could differ from device to device. Over HDMI they should be the same, assuming no error in the pattern.

Stacey,

I'll answer in more detail and do the tests that you suggest when I get home later today. Here's some answers...



HDMI through Denon 2807 to display.


1080i60, 1080p24 720p and Source Direct


Unsure what you are asking here.


Pioneer Elite PRO940HD (1024x768)


No. It's the SD flavor on the HD disc. I know there are differences.


Geometry and convergence patterns. I'l need to verify exact patterns when I get home


Will do.


100% HDMI here.

Actually, I am leaning toward issues with the BDP-83 as your disc gives results that are in line with the others... when using my PS-3 in the identical configuration. Oppo does not think (at this time) the -83 is the problem.

Steve,

Which pattern on our disc are you using? We have one called Image cropping. While viewing this pattern, press the up arrow and it will display on-screen help to show you how to look for pixels being cropped. Press down arrow to dismiss the help.

What is the native panel resolution of the Pioneer? I am being lazy by not looking. When you switch between the various resolutions, on the OPPO or PS3, do you see the image shift around? You should try this with both the HD Image Cropping AND the SD version from title 127. If you are not seeing any image shifting, lets stick with one resolution to try and debug. I would suggest 1080p24 if at all possible. You may also want to remove the Denon just to have one less variable, assuming its easy to get behind and unplug.

Title 127 is SD, so it should be a good choice to compare with DVE. First test would be does Image Cropping on title 127 look like the Image Cropping from the menu (The HD version under Setup & Evaluation) with regards to centering? If not, this could be a problem with the OPPO. Early on, the alignment of that pattern would change as the output resolution changed. I thought that was fixed. I will double check with the latest FW tonight when I get home.

Are you a beta test or EAP participant? Again, too lazy to check OPPO thread. It moves to fast to keep up! Please tell me the FW on your OPPO.

What about picture control settings in the OPPO? Have you changed anything from their default positions?

Stacey, I added the native res (1024x768) to Steve's post, probably after you read it....

larry

We are looking for feedback. We want to know what is easy and difficult about our disc. We tried hard to make it easy to navigate. It seems many have missed our on-screen help and menuless navigation.

We will probably start publishing a blog and/or articles on our website, which go into greater detail. Probably more than you ever wanted to know. We are using the feedback we receive to try and improve. Your catching the extra row of pixels on our Geometry pattern helps.

Larry, thank you.

No, a professional calibrator will bring extra equipment like a colorimeter or spectrophotometer (or both), and will get into the service menus and adjust individual gains and bias settings, calibrate white level and gamma, and a bunch of other adjustments as needed.

That said, just getting the main settings to the correct levels can have a significant positive effect on picture quality if they're not set correctly at the factory. Some displays have various picture modes with exotic names like "Vibrant" or "Cinema", and our disc can tell you which of those is closest to correct, or help you set up a user mode (if your display allows it) for viewing Blu-ray with the best settings.


No, you don't need anything else. If you have a blue filter, you may be able to use it to check color decoder settings (see my long post about filters in this thread). Otherwise it's just your eyes.


We like to think this disc is easier for the novice, because of the pop-up help (just hit "up" on the remote in any pattern) which tells you what you're looking for and shows examples of right and wrong. We also limited ourselves to tests that are either useful for setting normal user-accessible controls or useful for testing players and displays.

Don

D'oh! Hey, didn't you QC this pattern?

Mea culpa; I wrote the code that generated this pattern, and when I checked it I was checking that the squares were perfectly square and the circles perfectly circular. I can't believe I missed this. But apparently I did. I'm sorry. The pattern is still fine for its intended purpose of checking geometry, but it bugs me to not have it look perfect.

Unless Stacey messed it up in the encoding. Yeah, that's the ticket!

Don

That's the pattern.


1024x768 (thanks Larry )


Yes. It moves considerably when changing between resolutions on the Oppo. It does NOT move on the PS-3. I have only used the BD patterns so far.


Yep - will do tonight. Agreed on variable removing! However, I don't expect any impact - based on PS-3 results.


I hope it's F/W!


EAP-2

Main Version: BDP83-16-0320
Loader Version: BE2100 8572100
Sub Version: MCU83-19-0122
Chip Version: 0B.00.01.00


24P tuned on.
Output Resolution 1080i (toggled around during use)
Standard other than those.

The amount of overlap varies from display to display. There's no rule, but in most practical displays the blue primary has essentially no energy higher than 525nm and the red only goes down to about 575nm. So there's a 50nm window where you could use a filter.

But it's not guaranteed. Some displays might use wider, spikier filters or might have a lamp with a spiky spectrum that necessitates wider filters.

Looking at the Edmund filters spec sheet, they mention the half-energy width as 10nm, but the key is to know the width where the energy falls off to about 0.1% of peak. If that's less than 50nm, I would think it would work on most displays. I'd get one centered at 550nm for green.

Practically speaking for a single display you're better off getting a service manual for it and finding out how to turn off channels for calibration. Sometimes there's a specific service setting, and sometimes you have to write down the gain settings, then set two of them to 0. Make sure you write them down. You may think you're going to remember them, but write them down anyway. I speak from experience.

Don

Steve,

I played the image cropping pattern in HD and SD at all output resolutions on the OPPO and it displays all of the pixels correctly. I would start by performing a factory reset, which will walk you through the wizard again.

A few things have come to my mind.

1. When you switch between output resolutions, on the OPPO, I do expect some image shift may occur since your display is not native 1920x1080. It may have a different memory for each resolution, which might result in a different image position. If this were true, I would expect the same shift on the PS3. e.g. Switch between 1280x720p60 and 1920x1080p60 may shift the image around. Does your display allow you to adjust the position for each resolution and frame rate? If so, you may need to go through each resolution and adjust the image position.

2. With a given output resolution, say 1080p60, playing SD and HD sources should not shift the image around.

Some reasons the image migh shift include auto pillarbox (bars on the left and right) for 4x3 sources. The OPPO has two modes, 16x9 and 16x9/auto. I don't recall the exact name, but it is one of the wizard setup options. One of these will treat all SD as 16x9 and one will add bars to the left and right if it thinks it is 4x3 SD. Be sure to chose the 16x9 version w/o auto for testing.

3. Try turning off the CEC option in the OPPO setup menu.

I can try and locate the Pioneer around here. I may also try another Plasma as well.

Thanks for the research Stacey. Sadly, life got in the way this evening and I did not have time to try the models you suggested. I will have time mid-morning tomorrow.

I was wondering about that 16x9Auto mode for SD. It is currently set to ON. I'll ensure that I turn it off prior to modeling. Good thought!!!

-steve

To the OP: thanks for this disc, and for posting this thread. I have it on order now.

Contrast - "The ramp that goes from black to white and back should not have an extra-wide center white stripe. The ramp that goes from white to black and back should have a wider black stripe in the center."

What exactly does "should not have an extra-wide center white stripe" mean? Do I make this stripe as narrow as I can? Should this white stripe width be about 0.1" wide on a 40" tv or more like 1" wide? Just looking for a more quantifiable set of instructions here.

Thanks.