Originally Posted by *UFO*
If you are referring to using that specific image for testing black crush, the image itself is crushing blacks. On all my properly calibrated displays, including a laptop, half the squares are completely missing. That image is designed to not show squares on displays set to limited. That is what you want to see.
because i am finding that using a HTPC is so temperamental and unreliable to produce stable findings, and because of some of these conflicting observations a number of us have reported here, i have reinstalled my AMD CCC driver (to make a clean start) and then made a comparison of what effect the different settings had (either stand alone, or in combination with other settings). i believe 16 possible different main combinations of settings are involved here, and i wanted to see how they affected the DVE black/pluge calibration setting and if the black-crush problem could be reliably reproduced. the end result of this comparative testing are listed below.
for those settings where DVE calibration for black levels worked correctly i then used 2 additional images to determine if any "black crush" was present as a final test (one of which is specifically designed for a VA type panel, which our sony tv uses). For the testing process i also tried to make sure no multiple gfx or video programs were open at the same time, or were present on the sony desktop (which could corrupt obtaining reliable findings). i also used a good reliable image viewer (irfanview) and good dvd video player program (MPC-HT with madVr).
the 2 main variables in AMD CCC (but probably similar settings exist for Nvidia)
a) "video ---> color ---> dynamic range"--> "full (0 - 255) versus
Limited (16 - 235)" or
b) "digital flat panel ----> pixel format ----> sony TV ------> "RGB 444 (Limited RGB) versus
RGB (full RGB)"
the 2 main variables for the sony tv are
a) HDMI port set to limited or
b) scene-select mode of GAME (444 chroma) or Cinema (422 chroma)
the results are a bit long to type out, and would need some excel type spreadsheet, however the results were very clear and logical (imho). a summary is:
1) if the CCC video color is set to full (0 - 255) combined with any other of the above CCC or sony tv setting
, then there is severe black level clipping
and it is impossible to correctly calibrate with DVE on the sony tv screen
- that means that 8 of the 16 possible combinations of those 4 variables (see above) are completely ruled out as viable
for good quality video (eg 4 for Game mode, and 4 for Cinema mode)
2) when using CCC video color setting of "Limited (16 - 235)" combined with the setting
CCC pixel format "RGB (full RGB)"
- then i was able to correctly calibrate black-level/pluge with DVE (so good !)
- but both "black crush" test images always failed, showed severe black crush. eg for the VA "test" image the text was invisible, and for the second image it had 80% of the black blocks invisible. the only way to correct for this was to turn the brightness up significantly by 10 points (eg 10% of the 100 range) but doing this immediately significantly distorted the previously identified black level setting in the DVE calibration image, and increasing brighness by that 10% also completely "washed out" the colors in normal video
- so that rules out another 4 of the possible 16 combinations of those 4 variables
(eg 2 for Game mode, and 2 for Cinema mode)
thus far we just ruled out 12 of the 16 possible settings
of the last 4 possible combinations of those 16 options, 2 gave correct DVE black level calibration results for a sony tv brightness setting of 41, and 2 gave a correct DVE calibration setting for a brightness setting of 51. eg, all 4 of those last options were able to correctly calibrate black-level/pluge with DVE and also at the same time cause no "black crush"
on the other 2 test images. the main variable that altered this white/black level by 10% (while maintaining correct DVE calibration for black level, AND did not cause any black crush) was the HDMI limited/full setting on the sony tv HDMI input port.
- i suspect out of those 4 last options, the 2 results that correctly calibrated around 50/51 brightness level are likely to be the "better" ones, but the end result of the quality of the image is the same.
- the only variable that affected this brightness setting was HDMI full/limited on the sony input. the "HDMI full" setting caused the DVE black-level/pluge to drop to 41in DVE (for both game and cinema mode). with the same settings of the other variables being the same, setting the HDMI port of the sony tv to "limited" brought the brightness calibration level back to 50/51 in DVE (which is closer to what is expected). but both of those HDMI port settings were ok for DVE calibration, and neither caused a problem on the black crush image tests, so it is a matter of personal preference untill we find out the exact reason for that observed difference in brightness. so what are these correct CCC and sony settings ?
Conclusion: For the 16 possible options (of the 4 variables involved) in the CCC and sony tv settings listed, there are only 2 possible correct settings identified in either game or cinema mode that will correctly show the right blacklevel/pluge calibration settings with DVE, combined with no visible sign of the black crush problem
(the findings are applicable for both Game mode, and Cinema mode) these correct settings are:
- 1) CCC: color ---> dynamic range ---- > "Limited (16 - 235)
................ this is the critical one to prevent black level clipping on DVE. setting it to full 0-255 is a big image killer (with the other settings used in this comparison)
- 2) CCC: digital flat panel ----> pixel format ----> sony TV --"RGB 444 (Limited RGB)
............... this is the critical one to prevent black crush in both test images.
- 3) the sony TV HDMI full versus limited setting:
neither setting affected black crush or black level clipping (as long as the CCC settings were correct obviously), but setting it to "full" dropped the brightness level by 10 points (in both game and cinema scene-select modes). there are a couple of indications that "limited" might be the more appropriate choice because it brings the brightness level at the expected normal level of 50/51 , and with both CCC settings already at "limited (for RGB and for 16-235) it might make sense to also have limited for the HDMI port. i had also found some clever technical reason why somebody recommended this, but i still cant find the notes on this.
-4) There were no differences for Game vs Cinema mode (on the sony tv)
and their relationship to the different CCC settings and finding the correct blacklevel/pluge settings ( or their effect on black crush). both scene settings had the same results for brightness settings for these purposes.
- after using the above found "best CCC settings" in my setup and then comparing afterwards the video quality in a good blu-ray dvd video for both the the Game and Cinema mode settings on the w900A, the Game mode (444 chroma) seemed to produce a significantly richer image with fuller colors (a subjective observation). i am hoping to locate some test video/images or calibration process that can quantify this more objectively, and will investigate this further.
- the visual "end result" observed in the viewed video quality was also dependent on the video player used ( for eg MPC using MadVr gave for me the best results, where i have already done some basic settings), compared to other video players that i just used with default/standard settings. if anybody uses a HTPC and want to get consistent and good quality video, choosing a reliable good video player without major bugs is also important, and there are several reasons why some bad or incorrectly setup video player can corrupt an otherwise well setup system (like the known bug in WMP for ex).
- I suspect most people that dont want to waste lots of time constantly tinkering with their HTPC are better off using "cinema" mode (with the above CCC and HDMI settings we found to work best), and for those keen on squeezing more performance out of their displays, using game mode 444 chroma might be worth while
thats all folks !
these are 2 good images to use for quick "black crush" testing after the black-level have been correctly set first with a good calibration image
(without first setting the black-level correctly, looking for black crush is meaningless)
1) the http://www.lagom.nl/lcd-test/black.php
image, where the upper 3 rows show BTB from 15 to 1 (eg everything darker then 16 in the 0-255 range). after having correctly set the black level with a black-level calibration image, you should still be able to see the full rows of all black squares on that image (except maybe the last 2 darkest blocks on the top left). download the image and view full screen in a good image viewer (not web browser etc)
2) a simple purpose made black background image (1,1,1) with the word TEST written out in a slightly lighter shade of black (9,9,9). viewing it full screen should instantly give an impression of how hard/easy it is to read. with black levels correctly set first, the TEST image should stand out against the darker background