[Alan Sh]

The XBR910 is much better at displaying deep shadow detail(RGB 1-5 levels) than my RCA MM36100 or my Sony PFM-32C1 plasma. However it is easy to set brightness at a level in which shadow detail is missed without knowing it.


Each time I recalibrate the XBR910, I check the two scenes below to determine the range of brightness settings needed to see all shadow detail. Dark scenes can have a lower brightness setting than bright scenes with deep shadow detail.


Detail in the left pant leg of Dark City suddenly appears in the 4-2 RGB range. The two fingers in Punch Drunk Love have similar characteristics. The threshold command divides the picture into white and black at any RGB level. The fact that most of the hand and leg inserts are black at the 4 and 5 levels means that most of the detail is below that level.


PowerDVD should work on most PCs with a DVD player and standard video card. Download a trial version and click on the capture button anywhere in the movie. If it works it will create a one megabyte file representing a 24 bit RGB value for each pixel. You can specify the directory in which pictures should be kept.


Open the picture in Photoshop Elements. You can use the commands under Image to examine the RGB values in the picture. The equalize command gives you an idea of what detail is on the DVD. The histogram command tells you exactly how many pixels are at each RGB level if you place the cursor over the value you want. By changing the values of the threshold command you can watch the pixels change from white to black at each RGB level.

 

[ADU]

Hmm.... perhaps I'm missing your point, but I don't think you should actually be able to distinguish different gradations of brightness between RGB values of 0-16, at least not on a CRT. If you can, then I think the black level on the TV may be set too high, and you're loosing some of the contrast in the image. This is what most of the CRT calibration tools I've run across recommend anyway.

 

[Alan Sh]

Poynton on page 23 says that "Eight-bit studio standards have 219 steps between reference black and reference white." and shows a 16 offset meaning 16-235. However, when PowerDVD does the conversion from YUV to RGB, it converts to a 0-255 RGB colorspace.


A PC outputs different voltage levels for 0-255 but many displays cannot accurately display the lower ranges. The photo above is a good test for low range capabilities of a PC monitor or tv with direct RGB input. You should be able to see the left pant leg and right hand in the original portions of the photo. My LCD monitors can't. My Sony plasma accept direct RGB input and I can see both. My RCA crt also accepts direct RGB input and can see both on similar photos.


On my RCA being able to see the shadow detail comes at the expense of deep blacks. In order to see all the shadow detail above the NITS level is higher on the RCA than on the Sony plasma.


With the XBR910 I get a much lower black level (about .06 to .34 NITS) and yet can see all the shadow detail.


You can get a good estimate of roughly how much shadow detail you may be missing by stopping the DVD and adjusting the brightness. In my ISF calibration both my sets were fine adjusted by using scenes from DVDs and balancing washed out blacks vs shadow detail. In the end when a compromise must be made it is strictly a personal preference.


What Photoshop does is tell you exactly how much detail is contained on the DVD and precisely at what levels.


An example of how I used this recently is The Two Towers. When the credits were rolling, the XBR910 background had a faintly red tinge while the plasma showed black. I captured a frame of the credits, read it into Photoshop and applied the equalize command to the top half. This reveals that the background is indeed not precisely black while the border area is.

 

[NTN1]

Alan Sh,


Can you post the service menu items, and data that you changed to bring the color temperature of your 910 to 6500K?


I don't have an optical color analyzer. I was going to eyeball the thing, but any help from the specialist is always welcome. Thanks.

 

[ADU]

I'm a bit fuzzy on some of this, but I don't think the NTSC palette offset is the reason why CRTs are calibrated so that their RGB values from 0-16 appear black. I think it has more to do with the inability of human vision to distinguish minute gradations of near "true" black on a 2.2/2.5 gamma display.


Regardless of what content you're viewing, if you can make out any differences in gradation between RGB 0-16, then I don't think you'll have "true" blacks. Contrast will be diminished, and all imagery (the desktop, Photoshop, etc.) will sit on Poynton's dreaded pedestal of grey.


If NVDVD remaps the color space of DVDs to 0-255, then it's probably doing this to delibrately force the darkest colors in the video into this 0-16 twilight zone to ensure that blacks in the video appear "truly" black, and whites appear "truly" white on a properly calibrated computer monitor, so the full range of contrast is used. That would be my guess anyway. If you try to adjust the Brightness (black level) on your monitor so that you can make out gradations between RGB 0-16, I think you'll be destroying the contrast in images.


If this doesn't sound correct to anyone, then please educate me, cuz I'm not entirely sure about all this.

 

[Alan Sh]

NTN1


I will shortly try to post some of my calibration results in your thread on calibration. They are currently rapidly changing. I am receiving the DVE DVD today and want to check it using Photoshop against my 1080i DVE and Avia. In addition I did not like the color differences I saw on last night's movie on the plasma vs. the XBR910. Calibrating white makes some colors more inaccurate than they were before calibration.


ADU


I just read something in the WSR Samsung 931 review which tied up a few loose ends for me and might help you:


"The standard DVI-PC levels are 0-255, and the standard DVI-video levels are 16-235. "....which provides headroom for digitally processed signals to occasionally go below black(16) or above 100 IRE white (235)." "Pluge test patterns routinely make use of the below black signal capability to make display calibration of black level easier and more accurate. The HD931 produces DVI-PC signal levels (0-255) and doesn't transport below-black pluge signals across its DVI interface."


What I have been talking about is clearly the PC-levels.


It seems to me that anyone who calibrates using 16-235 and leaves room for below black and above white will have less contrast and greyer blacks than someone who calibrates using 0-255 and uses the full range for normal output.

 

[ADU]

Maybe I'm wrong about the 0-16 RGB thing. I thought the black level on all monitors was supposed to be set so those values appear black, regardless of the applications involved, but perhaps I need to explore that further.


This cross-over from analog YUV to digital RGB seems to have alot of unpleasant little wrinkles. For example, at first, I was concerned about using a full 24bit palette from a computer via DVI on my XBR800, because I thought it might force my TV to display colors it wasn't designed for. So far though (knock wood), I haven't noticed any problems from this. And the TV seems to display all 256 levels of grey just fine with the Brightness at 50%, the way a computer monitor would. So my tendency is just to treat it like a computer monitor.


If these DVI ports are forward compatible with uncompressed digital RGB, then I wonder if that will also just use a full 0-255 palette. Once we reach that point we should be solidly in the digital domain, so I'm not sure why a 16-235 palette would be needed.
Quote:

The HD931 produces DVI-PC signal levels (0-255) and doesn't transport below-black pluge signals across its DVI interface."

I noticed the 931 did not show the drop shadow on the THX black level test, which would seem to confirm this. Perhaps this is also a sign that it's remapping the DVD's palette from 16-235 to 0-255. Very interesting stuff, Alan.

 

[ADU]

I just checked the black level settings on my XBR800 with an HTPC via DVI again, and if I set the Brightness so that gradations are visible betwen RGB 0-16, I get a noticeable amount of light from 0 RGB parts of the screen, IOW a "pedestal of grey". If I lower the Brightness until RGB values from 0-16 are all black, then this pedestal disappears and 0 RGB parts of the screen don't seem to eminate any light. IOW I get true blacks, which are about as deep and dark as this TV allows.


This 0-16 rule of thumb seems to work on my computer monitor as well.


Adjusting Brightness until RGB 16 matches RGB 0 seems to insure that you have solid undiluted blacks in this arrangement.


I guess the next question is which type of palette do you use on your player: the one which leaves DVDs at 16-235, or the one which remaps them to 0-255? If you use a 16-235 palette, then you should be able to distinguish pretty much all of the darkest gradations in the video with Brightness calibrated as above. I'm not sure if you are supposed to be able to see all that detail in the shadows though. Frankly, I don't know which approach is correct. Is the 0-255 remapping approach the standard method that's recommended for playing DVDs on PCs?

 

[ADU]

The odd thing is when I use the THX Optimizer with the Samsung 931 DVI player, it yields a Brightness setting which is almost twice as high as my PC. So it seems to be doing things differently. (??)


I get similar results if I scale the image on the TV horizontally so I can see non-voltage areas, and then adjust the Brightness on both device to match it. The computer seems to be putting through a noticeably higher base level of brightness than the Samsung player. If both are putting through digital RGB values of 0-255, I don't see why they should be different.

 

[ADU]

Nevermind. I think I may have figured out the difference. The computer is scanning progressively at 540p, while the player is scanning interlaced at 1080i. Perhaps this might explain why the computer's a bit brighter.