[karoloydi]

So, I finished my calibration. Was quite happy with it.

Next day I tested again to see if I ll get the same results. The difference was not huge, but big enough. For example instead of gamma 2.2 I got gamma 2.7. Instead of deltas being lower than 3 for the grayscale they were around 5.

So, I decided to calibrate again. After an hour of fiddling with the settings, I decided to test my original (yesterday's) settings again.

Perfect results again! Like the way they were the previous day.

So, I am thinking that the process of calibration produces results that dont reflect the real world.

Maybe we should harry up and finish the calibration within 10 minutes? But that would be unreasonable.

 

[PlasmaPZ80U]



Quote:
Originally Posted by karoloydi /forum/post/18244044


So, I finished my calibration. Was quite happy with it.

Next day I tested again to see if I ll get the same results. The difference was not huge, but big enough. For example instead of gamma 2.2 I got gamma 2.7. Instead of deltas being lower than 3 for the grayscale they were around 5.

So, I decided to calibrate again. After an hour of fiddling with the settings, I decided to test my original (yesterday's) settings again.

Perfect results again! Like the way they were the previous day.

So, I am thinking that the process of calibration produces results that dont reflect the real world.

Maybe we should harry up and finish the calibration within 10 minutes? But that would be unreasonable.

What display and what meter are you using? I have an i1 Display 2 that must be warmed up on the powered up display for at least 30 mins to give consistent results. A full hour or even 90 mins might be even better. I have noticed this on my Samsung LCD and I'm sure it applies to my Panasonic Plasma as well (although it's been a while since I've used it on that). The reason for this is that my meter is not temperature compensated and so it must stabilize on the powered up TV (and the TV must also stabilize as well) before you can get repeatable and reliable readings.


Also, my meter gives wrong readings during the first 1-3 runs after the sensor is calibrated. So, I simply take several readings and discard the first 1-3. After those initial runs, the meter is reliable assuming it's properly warmed up and the display's properly warmed up (to the same temperature).

 

[jkcheng122]

For plasmas you definitely want to wait at least 30 min, 60 to 90min preferred, before you start calibrating. And by waiting I mean TV on with meter in calibrating position on the screen for the duration.

 

[karoloydi]

I have a Samsung LE46A656A. And using an eye 1 LT colorimeter.

What you said about the i1 Lt needing to get warmed up makes sence.

I never thought if it.

I ll do a test. Warm up the i1 LT on my tv while showing test patterns. Then measure. Then leave it there and watch tv for like 30 minutes. Then test again. And compare.

But I am thinking that the problem could be that during the calibration there is only a window in the middle of the screen that has light.

During normal viewing conditions the whole tv has light, not just a window in the middle. This could change the temperature of the tv and alter the results.

Would it be ok if I used patterns that took the whole screen instead of windowed patterns? I think I ve read somewhere that its ok for lcds, but not ok for projectors. Is that correct?

 

[PlasmaPZ80U]

Full fields work fine on LCDs. In fact, they can work better than windows if auto-dimming is enabled (called CE dimming on Samsungs).

 

[orion456]



Quote:
Originally Posted by karoloydi /forum/post/18244044


Maybe we should harry up and finish the calibration within 10 minutes? But that would be unreasonable.

You have to wait for the lamp and projector to reach operating temperature which means at least 30 minutes.


With an LCD projector, where you place your meter is important because LCDs don't have uniform gray fields; there is always some variation in color and intensity. If you place the meter to the right or left, you will get different grayscale results.

 

[karoloydi]



Quote:
Originally Posted by orion456 /forum/post/18244635


You have to wait for the lamp and projector to reach operating temperature which means at least 30 minutes.


With an LCD projector, where you place your meter is important because LCDs don't have uniform gray fields; there is always some variation in color and intensity. If you place the meter to the right or left, you will get different grayscale results.

This is happening even if I have the tv on for hours.

 

[sotti]



Quote:
Originally Posted by karoloydi /forum/post/18244800


This is happening even if I have the tv on for hours.

Yes. LCD's aren't very even, and warm up can take 90 minutes.


I was just doing this with my Toshiba LCD yesterday. It can take a good 1hr for the lamps to stabalize. So when I started the 100% Y was only 98cd/m or so. This was only 15minutes of warm up. Another 15 minutes it was up t0 118cd/m or so. By the time I was done Y was up 145cd/m.


I wasn't changing any settings, just measuring some patterns.


Then once we are full warmed up, I can see variation if I move the probe around.

 

[sotti]



Quote:
Originally Posted by karoloydi /forum/post/18244563


I have a Samsung LE46A656A. And using an eye 1 LT colorimeter.

What you said about the i1 Lt needing to get warmed up makes sence.

I never thought if it.

I ll do a test. Warm up the i1 LT on my tv while showing test patterns. Then measure. Then leave it there and watch tv for like 30 minutes. Then test again. And compare.

But I am thinking that the problem could be that during the calibration there is only a window in the middle of the screen that has light.

During normal viewing conditions the whole tv has light, not just a window in the middle. This could change the temperature of the tv and alter the results.

Would it be ok if I used patterns that took the whole screen instead of windowed patterns? I think I ve read somewhere that its ok for lcds, but not ok for projectors. Is that correct?

That's correct LCD's typically aren't affected by using full feild patterns.


LCD's typically have 3-7 lamps behind the screen that are always fully lit up. So the whole screen warms up fairly evenly. Also the ouput of one area of the screen isn't any brighter/darker depending on what the rest of the screen is doing.


Plasma are completely different.

 

[karoloydi]

I dont think that the warming of the tv or the position of the calibrator on the screen is the problem.


This is why:

I have the tv on for hours before I start. I measure the accuracy of my calibration with calman.

After I have been displaying calibration patterns for 1 hour with same settings, same position of calibrator on screen, I measure again. The results are different.

The position hasnt changed. The tv had already been on for hours before I started the calibration.


The only thing that has changed is:

The calibrator has warmed up by the screen.

I have been displaying calibration patterns.


So the problem is either:

The calibrator needs to warm up before the calibration.

Or the patterns change the temperature of the tv.


But, if the calibrator needs to warm up, whats happening on projectors? There is very little warming up of the projection screen happening when you turn on the projector compared to lcds, right? So, the calibrator remains relatively cold during the calibration process. Does that mean that you cant calibrate projectors accurately then? I think not

Thats why I think the problem is that I am displaying windowed patterns. I should be displaying full screen patterns instead.

 

[sotti]

it's probably the probes temprature that's causing the difference then.


It's not that it's more or less accurate when it's hot, it's just that it's different.


A more precise instrument won't be effected by temprature, or will have temprature correction circuitry.

 

[Kelvin1965S]

Once the sensor has warmed up (or in the case of a projector, stabilised to the ambient room temperature) a black level calibration should be done. If this is done before warm up then thermal drift will effect this reading and make the results skewed.


I went through this exact issue when I first started calibrating my LCD TV as a practise for my projector. The non repeatability of results nearly made me give up, but putting the TV on for at least an hour (preferably longer), leaving the sensor on the screen for this time and then doing a black level caliibration seemed to make a big difference. There is still the issue of uniformity of the LCD to contend with, but I found that if I put up a 100IRE pattern and slowly adjusted the position of the sensor so that I got close to the previous day's 100 IRE reading (no controls adjusted in the meantime of course) then the gamma, etc when remeasured would be very close to the previous run.

 

[thomasl]



Quote:
Originally Posted by karoloydi /forum/post/18244044


Next day I tested again to see if I ll get the same results. The difference was not huge, but big enough. For example instead of gamma 2.2 I got gamma 2.7.

That luminance difference actually seems fairly large. I have a number of Samsung LCDs and I've never experienced such variability due to temperature drift with my Eye-One Display. My LCD screens don't seem to really get much hotter than the room temperature so as long as the meter is warmed to room temperature and the display has been on for an hour or so, it seems to work fairly reliably but a little variation is always expected with a budget meter. The only time I've noticed larger than expected variability in readings was when calibrating the non-heated basement LCD during winter - cooling off definitely seemed to affect the Eye-One but nothing that would have caused a difference of 2.2 to 2.7 for the average gamma.


You'll probably want to try an experiment where you leave the meter on the display while the display is initially on for an hour or so. You could also try simply holding the meter up to the display just when taking readings and not leaving it on the display the entire time during the calibration session.


cheers,



--tom

 

[karoloydi]



Quote:
Originally Posted by Kelvin1965S /forum/post/18247041


Once the sensor has warmed up (or in the case of a projector, stabilised to the ambient room temperature) a black level calibration should be done. If this is done before warm up then thermal drift will effect this reading and make the results skewed.


I went through this exact issue when I first started calibrating my LCD TV as a practise for my projector. The non repeatability of results nearly made me give up, but putting the TV on for at least an hour (preferably longer), leaving the sensor on the screen for this time and then doing a black level caliibration seemed to make a big difference. There is still the issue of uniformity of the LCD to contend with, but I found that if I put up a 100IRE pattern and slowly adjusted the position of the sensor so that I got close to the previous day's 100 IRE reading (no controls adjusted in the meantime of course) then the gamma, etc when remeasured would be very close to the previous run.

What you said about the 100 IRE value is what I have noticed as well. But the way I do it is instead of moving the sensor, I adjust so that the tfL value at 100 IRE is the same as before. Then everything else is same as before.

 

[Kelvin1965S]



Quote:
Originally Posted by karoloydi /forum/post/18247169


What you said about the 100 IRE value is what I have noticed as well. But the way I do it is instead of moving the sensor, I adjust so that the tfL value at 100 IRE is the same as before. Then everything else is same as before.

What do you adjust to match the fL value though? If it's contrast or anyother controls on the set (including the backlight) then that will skew the results. Short of making a clamp for the sensor it's the best I can do and in any case it is still only measuring one point from the screen and there could be considerable differences across the screen, especially in corners.


At least with a projector you could defocus to help average the the image a bit or use the diffuser facing the projector which means the image is shrunk across a smaller area of the sensor and thus averaging the readings a little more.

 

[karoloydi]



Quote:
Originally Posted by Kelvin1965S /forum/post/18247805


What do you adjust to match the fL value though? If it's contrast or anyother controls on the set (including the backlight) then that will skew the results. Short of making a clamp for the sensor it's the best I can do and in any case it is still only measuring one point from the screen and there could be considerable differences across the screen, especially in corners.


At least with a projector you could defocus to help average the the image a bit or use the diffuser facing the projector which means the image is shrunk across a smaller area of the sensor and thus averaging the readings a little more.

I just remembered. What I said before was wrong. I adjust the brightness control on my tv so that the tfL at 10 IRE is the same as before. Not at 100 IRE.