Originally Posted by lovingdvd
The reason I chose to start at the bottom end is that it is much more sensitive to changes than the low end.
Yes, but the current concern is at the top end.
Therefore one should correct this area first.
Second, at the top end -95- the gradation stepping is very low (about 2 notches for 0.1 gamma variation) whereas at 5 IRE it is high (about 6 for the same gamma variation). So endind up the job at the bottom end is easier and the compromise has less impact.
Another trick I use is to save often (for instance exit and save changes after I do a few points, then come back and do some more, then save, etc and also copy these to another preset). This way if I mess up I do not have to start all over.
I abandoned this: the procedure is reliable enough.
I just keep track of the values.
I used to run the full grayscale and correct everything in one go.
Working point after point is possible but switching between HCFR, Excel and the JVC is long and source of mistakes
At the curve end it's different
One possibly difference in our findings is that from the translated version I read it seems you were finding that it changed points both above and below the point you were working on? So if you were working on 60 it made changes to 50 and 70? I did not find that.
Yes it is.
That is why it is possible to apply the same procedure downward and upward.
In some occasions it happens that one next point does not move probably because the JVC does not need to change it.
JVC tries always to smooth the curve.
It also happens that one next point moves in the opposite direction!
This is puzzling.
The result was nearly perfect, except for the high end. At 90 and 95% my 2.2 target gamma falls off to more like 2.1 and 2.0. I concluded there was no way to get this 90-95% range flat, without then messing up 70-80%. So I settled for a perfectly linear 2.2 gamma from 5-80% with the fall off at the top end.
It is not easy. Even when proceeding from the top end.
I envisage another approach...
Also, the low sensitivity and fluctuation readinds of the probes in the low range does not help.
I am still wondering why the gamma changes so much after some bulb aging, compared to the new bulb. As you pointed out, its not just that the gamma becomes too low. But rather it has an inverted "Z"-like shape, with the low end having gamma way to high, the middle about on target, and the high end falling off significantly.
Looking around I notice that there is no rule with ageing.
Some show the trend very early and others later.
The shape is identical
On another hand it seems that the 3 panels shift about the same way in time
It looks like a kind of non-linearity problem at the top end.
The luminance curve seems to be bent as from 90 IRE.
I did an experiment. With the HCFR probe one can get the low level sensor reading. I edited the data and increased the values at 100 IRE by 1.7%.
As a result the gamma curve straightened up nicely.
Also, I reversed the gamma tables and found that they are not designed for a linear gamma at both ends. That's why I'm using the tables as reference in my spreadsheet. Since I changed my mind.
Might be done in order to compensate the native panel behaviour.
I can't imagine that the bulb alone is responsible for that.
I don't mind that much to make this adjustment once. But I haven't remeasured since I made these changes months ago, and have since put another 100-200 hours on the bulb. I certainly hope that it hasn't shifted again! It seems to me that perhaps it would keep shifting.
Giving the complexity and time involved in this gamma calibration, I certainly would not want to have to repeat this every few hundred hours. I am not saying it is necessary, but rather I am questioning if it is. Have you put a few hundred hours on the bulb since you last calibrated and remeasured to see if it held nearly constant?
I'm almost certain that a bulb change will not recover the initial gamma performances, and that the shift has a limit even if it takes more or less time to be achieved. And that the limit value is nearly the same for all the units. Don't ask me why
As Tom who insisted many times, I agree that an accurate grayscale calibration including gamma is THE key factor. Mainly for 3D perception and tonic images.