AVS Forum banner
1 - 20 of 65 Posts

·
Registered
Joined
·
154 Posts
Discussion Starter · #1 ·
Hi,


A couple people in another thread heard that I recently purchased CalMan and asked for a comparison of results between CalMan and HCFR, so I started this thread to share the results.


DISCLAIMER #1: I am totally new to the world of TV calibration - I've been learning as much as I can, but I certainly have much left to learn. I am using both apps the best way I know how, but it's entirely possible, and likely, I am using them incorrectly or have them improperly configured. This thread is part learning experience for me anyway, so if you have any suggestions for other things for me to try out, I am all ears.


DISCLAIMER #2: I am not making any statements about which software is "better". I am no where near qualified to make such statements (see disclaimer #1). I am just showing the results I got. Make your own conslusions. Though I am curious to hear what the more experienced on here have to say.


Equipment:

TV: Pioneer SD-643HD5 RPTV

Colorimeter: Eye-One Display 2

Calibration software: CalMan v3.2 & HCFR 2.0.1

Test Patterns: DVE HD-Basics (BluRay)


How I tested: for HCFR, I used Kal's *excellent* guide here . For CalMan, I used a combination of the built-in help and Kal's guide. I attempted my first grayscale calibration about a week or so ago using CalMan. Last night, to gather data for this thread, I re-ran measurements using CalMan and then ran measurements using HCFR immediately afterwards. The TV had been on for a couple hours before I took measurements. I plugged the data into Excel so I could graph both sets of results using charts that have consistent scales, dimensions, etc. I felt this was the best way to make accurate judgements about the differences. My Excel, Calman, and HCFR files are in the ZIP file attached to this thread.


So, without further ado:




I couldn't figure out how to get the individual RGB values for gamma from CalMan, so they aren't shown here.





I plotted the target curve the way it shows up in the respective app, which is why they look different here. I guess I should have made them consistent.





The color coding indicates how far apart the results are between CalMan and HCFR. The more red, the farther apart the values are.



I will say one thing: the HCFR results leave me thinking I still have quite a bit of calibration left to do, whereas I felt like I was more or less done with CalMan. So this begs the question, which app do I listen to?

 

CalMan & HCFR Results.zip 52.6904296875k . file
 

Attachments

·
Registered
Joined
·
3,351 Posts
Don't take offense, but from the data I have to say that you have misconfigured something.


There have been other comparisons of HCFR and Calman and while the results have been different, they are not as wildly different as what you are showing.
 

·
Registered
Joined
·
3,458 Posts
I'm scratching my head why your xyY measurements vary that much. For the most part your procedure seemed good. Only two questions come to mind:


- Were you in a light controlled room or is there a chance that light in the room could have affected readings? I took my readings with the only light being a bulb behind the TV and the laptop monitor.


- Did you move the meter at all? I know I moved mine a bit because I did CalMAN first and then lifted the meter for the HCFR calibrate the meter step. It probably would have been better if I would have done HCFR and then CalMAN because my SXRD doesn't have stellar uniformity and measurements vary when the meter moves.


Anyway, my measurements are generally within repeatability regardless of the software used. I just did grayscale and attached my readings. I like the new "alt + direction arrow" added to CalMAN and I greatly prefer the non-normalized luminance graph in CalMAN, but as far as readings go I really get no significant differences.


HCFR:



CalMAN:




 

COMPARE.zip 1.7802734375k . file
 

Attachments

·
Registered
Joined
·
132 Posts
Using HCFR, I've found that configuring the sensor (Measures/Sensor/Configure) set to "Average many reads on dark measurements" produces a higher degree of accuracy. Also periodic calibration of sensor offsets is also beneficial( I use 10 min). And, I respect that keeping the room as dark as possible helps too !
 

·
Registered
Joined
·
154 Posts
Discussion Starter · #5 ·

Quote:
Originally Posted by alluringreality /forum/post/15419120


- Were you in a light controlled room or is there a chance that light in the room could have affected readings? I took my readings with the only light being a bulb behind the TV and the laptop monitor.

Same as you, the only light was a lamp behind the RPTV and my laptop screen, which stayed put for the duration of the exercise.

Quote:
- Did you move the meter at all? I know I moved mine a bit because I did CalMAN first and then lifted the meter for the HCFR calibrate the meter step. It probably would have been better if I would have done HCFR and then CalMAN because my SXRD doesn't have stellar uniformity and measurements vary when the meter moves.

Yes, I did move the meter, and for the same reason as you: I did CalMan first and then needed to lift the meter to calibrate HCFR. I put it back in roughly the same spot - it shouldn't have been off by more than, say, three inches. I'll try again tonight with HCFR first to see if that changes anything.
 

·
Registered
Joined
·
154 Posts
Discussion Starter · #8 ·
As per Kal's instructions in his guide I did the following:


- "LCD" calibration mode

- I calibrated the sensor offsets by placing the colorimeters on the inside cover of a black DVD case.

- Rec 709 color space


I just noticed that Kal's guide says:
Quote:
If you run into issues such as finding that the software locks up or takes very long (30-60 seconds) to take readings sometimes, you may want to try "CRT" mode instead. We're still trying to figure out the logic behind all this and which is the best mode to use for CRT rear/front projectors - sorry!

HCFR did appear to take a long time making readings (though it was more like 15-20 seconds) so I will try CRT mode tonight to see if that makes a difference.
 

·
Banned
Joined
·
2,421 Posts

Quote:
Originally Posted by jvincent /forum/post/15418793


There have been other comparisons of HCFR and Calman and while the results have been different, they are not as wildly different as what you are showing.

Any difference should scare the hell out of someone.


Looks like a 3rd piece of software is needed to determine "who's right".
 

·
Registered
Joined
·
4,411 Posts

·
Registered
Joined
·
154 Posts
Discussion Starter · #12 ·
Ok, so I re-ran the measurements, this time with HCFR set to "CRT" mode. The results are much more similar this time, so I guess that was the cause. The differences flare up a bit in the lower levels. I'm guessing some of this is to be expected.












 

CalMan & HCFR Results - 2nd Try.zip 52.4326171875k . file
 

Attachments

·
Registered
Joined
·
3,351 Posts

Quote:
Originally Posted by BeachComber /forum/post/15420340


Any difference should scare the hell out of someone.


Looks like a 3rd piece of software is needed to determine "who's right".

Why should it scare somebody?


In the first case he had the meter configured incorrectly so of course they should be a difference.


In other comparisons the differences have been within the error of the meter, so for all intents and purposes they are the same.
 

·
Registered
Joined
·
3,351 Posts

Quote:
Originally Posted by Schmoe /forum/post/15420697


Ok, so I re-ran the measurements, this time with HCFR set to "CRT" mode. The results are much more similar this time, so I guess that was the cause. The differences flare up a bit in the lower levels. I'm guessing some of this is to be expected.

IIRC the Calman team have worked with the meter vendors and have implemented some algorithms in the S/W which improves their accuracy in the low light readings.
 

·
Registered
Joined
·
707 Posts
jvincent

Quote:
IIRC the Calman team have worked with the meter vendors and have implemented some algorithms in the S/W which improves their accuracy in the low light readings.

Simply applying averaging to bad input data only provides you with a median value which is more repeatable however it is still inaccurate data. The i1Pro needs to be re "Dark Measured" every ten minutes as it is not temperature compensated and will drift over time. This makes attempting to characterize the noise of the instrument rather pointless as it is constantly changing. The Dark Measure noise is subtracted from the actual measurement and there is no additional methodology on improving the readings available from X-Rite for this instrument. We have been working very closely with Sequel/GretagMacBeth/X-Rite for over nine years now and can tell you that aside from applying averaging to the instruments data there is little that can be done to improve its performance.


The instruments data becomes unreliable as it approaches 1.0 fL and becomes progressively worse as the light intensity diminishes. Derek has agreed with this in previous posts so I see no reason why you should think differently. The same thing applies as far as averaging the data for any of the filter based instruments which most users currently have.
 

·
Registered
Joined
·
3,351 Posts
The probe in question here was the Display2/LT which I believe has better low light performance than the Pro.


I was going from memory and I thought I remembered that there was something beyond simply averaging that they were doing but I could be wrong. It wouldn't be the first time.


I generally ignore the 0% and 10% reading from my Display2 anyway.
 

·
Registered
Joined
·
1,872 Posts

Quote:
Originally Posted by ghibliss /forum/post/15423051


jvincent




Simply applying averaging to bad input data only provides you with a median value which is more repeatable however it is still inaccurate data. The i1Pro needs to be re "Dark Measured" every ten minutes as it is not temperature compensated and will drift over time. This makes attempting to characterize the noise of the instrument rather pointless as it is constantly changing. The Dark Measure noise is subtracted from the actual measurement and there is no additional methodology on improving the readings available from X-Rite for this instrument. We have been working very closely with Sequel/GretagMacBeth/X-Rite for over nine years now and can tell you that aside from applying averaging to the instruments data there is little that can be done to improve its performance.


The instruments data becomes unreliable as it approaches 1.0 fL and becomes progressively worse as the light intensity diminishes. Derek has agreed with this in previous posts so I see no reason why you should think differently.

Yes the i1Pro does become unreliable under 1fL. This thread is about the Display2 which does not require data modelling to get the best result just a longer integration time.


BTW, our low light handler for the i1Pro is not just a simple average. We take 20 to 30 readings and then run a data modelling algorithm to determine the result. This includes removing noise and any readings suspect of being of low quality.
 

·
Registered
Joined
·
2,341 Posts
First off... thanks Schmo for taking the time to run both programs and making the comparisons.


For the DIY and for the non professional this teaches a very important lesson, that the way the setup/configuration of ALL settings is of the upmost importance.


Very interesting and impressive.
 
1 - 20 of 65 Posts
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top