- This topic has 23 replies, 2 voices, and was last updated 7 years, 4 months ago by
.
-
Posts
-
Thanks. Tone response curves look a little bumpy, this seems to indicate ABL might be a problem. If you can, make the pattern area even smaller (5%?) and use 20% APL background instead of 40% to see if this improves things. For testing, reduce the number of patches to a few hundred (e.g. 425).
Hi Florian,
Thanks for all your help so far. I will try a few more experiments tonight with patch sizes and backgrounds. I would really like to try to use Displaycal to set a custom white level, rather than use the TV controls. The reason is, the TV has two ways to control luminance – contrast and ‘panel light’. Panel light is I guess a plasma equivalent to an LCD backlight. I guess it works like a ‘voltage limiter’. The biggest difference though, is lowering it actually increases ABL. So for me it’s not a useful control to control light output. Moving on to the contrast control, it does not seem to work well at anything other than default. So my thinking is if I leave contrast at default, and set panel light at max, the TV will have the maximum voltage that it can draw on hand. Setting the TV like this, white measures an eye-searing 150cd/m2. So my thinking is, if I specify a specific white level (say a more reasonable 120cd/m2), ABL should really be minimized as the TV is able to draw enough voltage for a much much brighter output.
So really in order to test this, to set a custom white level, on the ‘Calibration’ tab, I just change ‘white level’ from ‘as measured’ to ‘custom’, and type a value in there? And for ‘white point’ to have the 3dlut adjust that to 6500k, rather than use the TV controls, I set that to ‘as measured’? (That one is confusing for me, as it reads like setting it that way, whatever the actual TV white point is, would be the output white point of the 3dlut, and not 6500k, but clearly I’m mis-reading something).
Update – there must be something more i need to do other than specify the white level, i put in 120cd/m2 but the result is just as bright as without the 3dlut (154cd/m2)
Ok thanks – think I get it now. I will do some more testing tonight but at this point I think the TVs 2.4 gamma setting is broken. At gamma 2.2 light output is a perfect 120cd/m2 with the panel light at 95%. If I change gamma to 2.4, light output shoots to a crazy 151cd/m2. Something can’t be right there. I’m going to try and calibrate to 2.4 with the TV at 2.2.
Forgot to say, you can set calibration speed to “Fast” to save a bit of time, as the grayscale will be corrected by the 3D LUT anyway.
Hi Florian, after a few more tests I have concluded that there is something broken with 2.4 gamma on my TV.
Could you please check the attached file and let me know if it looks ok? Visually the calibration looks good, in fact incredibly similar to the built in ‘cinema’ preset on the TV. The TV was in ‘Expert 1’, gamma 2.2 and everything set at default. I manually corrected the white point using the TV controls. I used a 4% window with a 36% linear light ARL background (which has an APL around 10.5). I tried using gamma light APL (20-25%)backgrounds but my TV did not respond well to those.
Also, this was a run of 833 patches. Would there be any benefit in this case to run a larger set?
The “bumpiness” that I saw is still there, but this seems to be an inherent property of the TV. If the result is good though, that’s all that matters in the end.
Hi Florian, I’m settled on 4% window with 36% linear light background. This definitely gives the best visual result. A weird anomaly with my latest attempt, however, with the AVSHD white clipping pattern, I have green (and not grey) flashing bars right up to 254 (with the 3dlut off , the bars are clipped at 235 – as it should be with a completely PC levels chain). What could cause this? I also tried the ‘clip WTW’ option for the 3dlut output but no difference. Otherwise, I don’t notice any greenness with regular content and the ‘report on calibrated display tells me the white point is at 6570k, so close enough for me). (Should also add that I used ‘as measured’ so the 3dlut is correcting the white point).
