AVS Forum banner

Massive Color Drift Correction for i1 Display Colorimeter

13K views 94 replies 15 participants last post by  sotti 
#1 ·
I recently tried measuring both a new Westinghouse VR-3730 LCD HDTV and my older Panasonic 42PZ85 plasma HDTV. I used my i1 Display LT colorimeter with HCFR software, only to find a large RED color shift. The subective PQ looks OK but the i1 is reading RED at about 50-60% lower than what I would be expecting.


I directly compared an older data file of my Panasonic TV with a new one at the same TV settings. The luminance and gamma charts are quite similar but the RGB levels, color temperature and CIE diagrams are different or shifted, especially WRT red values.


Is this magnitude of drift error that others have experienced with i1Display colorimeters? Or is there something else that is the problem?


Can HCFR software null this error out? Is this feature within it's capabilities?
 
#52 ·

Quote:
Originally Posted by airscapes /forum/post/21927443


Nothing sold by Harbor Freight is a serious quality product, I own many of them. I know they are inexpensive replicas of good stuff and they will fail but if I get 2 or 3 uses out of them I have gotten my money's worth.


Think of you i1D2 in that way. You got a couple of calibrations out of it, and you learned how to calibrate. You would have paid a pro more than you spent to do several calibrations. After all isn't that why we buy our own tools and do things ourselves, to save money and have that feeling of accomplishment?

On the other hand, I have many Sears Craftsman tools which were inexpensive from many years ago and still work fine after some being abused on auto repairs from times when funds were limited and repairs were done "under the shade tree".
Given that analogy, it is not unreasonable to expect the same kind of return on investment on a $200 (original price) meter.
 
#53 ·

Quote:
Originally Posted by Phase700B /forum/post/21928323


On the other hand, I have many Sears Craftsman tools which were inexpensive from many years ago and still work fine after some being abused on auto repairs from times when funds were limited and repairs were done "under the shade tree".
Given that analogy, it is not unreasonable to expect the same kind of return on investment on a $200 (original price) meter.

The durability of inexpensive, forged/hardened tools in no way relates to the long-term accuracy of an inexpensively made sensor designed to measure certain qualities of light. As someone who has supposedly worked with "pro-grade" calibration equipment before, I'd think you'd understand and respect that massive difference, as well as the massive difference between a $200 colorimeter and something several orders of magnitude more expensive that is used by professionals.
 
#54 ·

Quote:
Originally Posted by deandob /forum/post/21928005


Yes, this is a similar result I got when both my i1 LT's drifted.

So you saw about a dE=40-50 shift in color error in 12-18 months. I trying to quantify what people have actually measured and over a definite time period.


In my case, the (see above) reference plots were 26 months apart and have massive RED filter error with normal storage.
 
#55 ·
I am still confused as to why the luminance and gamma plots look so similar. The luminance plots are all most on top of each other. The gamma has some differences with the RED data darker (gamma higher) for most of the chart with the exception of 90% point.


Why are is this data reasonably good, when the RGB Levels, color temperature,and CIE diagram data clearly indicate a big problem?


What am I missing?


How would you adjust or correct for this in HCFR? Can an Excel spreadsheet be used to adjust things?
 
#56 ·

Quote:
Originally Posted by HogPilot /forum/post/21929307


The durability of inexpensive, forged/hardened tools in no way relates to the long-term accuracy of an inexpensively made sensor designed to measure certain qualities of light. As someone who has supposedly worked with "pro-grade" calibration equipment before, I'd think you'd understand and respect that massive difference, as well as the massive difference between a $200 colorimeter and something several orders of magnitude more expensive that is used by professionals.

Just following up on Airscapes' comment about Harbor Freight tools. Interesting you would seek out and have to comment ONLY on my post.



Typical though.
 
#57 ·
I don't know exactly the magnitude of the shift but the red decrease was very visible to the naked eye making the picture look green / turquoise (the green was more obvious than the blue when the picture was red deficient). Sorry I don't have actual dE values but I would guess 15+
 
#58 ·

Quote:
Originally Posted by Plasma54321 /forum/post/21929593


I am still confused as to why the luminance and gamma plots look so similar. The luminance plots are all most on top of each other. The gamma has some differences with the RED data darker (gamma higher) for most of the chart with the exception of 90% point.


Why are is this data reasonably good, when the RGB Levels, color temperature,and CIE diagram data clearly indicate a big problem?


What am I missing?


How would you adjust or correct for this in HCFR? Can an Excel spreadsheet be used to adjust things?

Luminance and RGB are 2 different things. The meter can read the proper brightness (amount of light or luminance) correctly but not the proper color.

Gamma is the difference in luminance form 0 -100% and has nothing to do with color, so it can be accurate and the color reading be very wrong. You can not fix this without using a known good reference spectror meter, and if you had such a meter, you would probably just toss the D2 in the trash and use the spectro since the speed difference would be of no great issue.


If you are not familiar with meter profiling this is something that is done in the software if it support it. I have no idea what HCFR support.

To profile a meter:

The software tell you to ready your reference meter and you place it so it is looking at a known spot on the screen, hopefully taking into account Field of view difference between the 2 types of meters.


Then you display a Red, Green, Blue and white pattern and take readings of each.


Next you set up the target meter that you want to make better so it sees the same spot on the screen.


You do the same readings on the target meter.


The software makes a table of the difference between what the reference meter sees vs what the target meter sees.


This difference table is then applied to all readings made by the target meter during the calibration.


This fix is only accurate on this one display where the reading were taken.


Hope that clears it up.
 
#59 ·

Quote:
Originally Posted by HogPilot /forum/post/21929307


The durability of inexpensive, forged/hardened tools in no way relates to the long-term accuracy of an inexpensively made sensor designed to measure certain qualities of light. As someone who has supposedly worked with "pro-grade" calibration equipment before, I'd think you'd understand and respect that massive difference, as well as the massive difference between a $200 colorimeter and something several orders of magnitude more expensive that is used by professionals.

+1, well said

Quote:
Originally Posted by Phase700B /forum/post/21929605


Just following up on Plasma54321 comment about Harbor Freight tools. Interesting you would seek out and have to comment ONLY on my post.



Typical though.

so is your reaction to HogPilot's post; this is an open discussion thread, not private messaging
 
#60 ·
I'll say it again, proper storage of these meters is essential to maximizing their lifespan (especially for those meters with completely exposed filter arrays). I'm not sure the OP's storage method was 100% optimal.
 
#61 ·

Quote:
Originally Posted by airscapes /forum/post/21929947


Luminance and RGB are 2 different things. The meter can read the proper brightness (amount of light or luminance) correctly but not the proper color.


Gamma is the difference in luminance form 0 -100% and has nothing to do with color, so it can be accurate and the color reading be very wrong. You can not fix this without using a known good reference spectror meter, and if you had such a meter, you would probably just toss the D2 in the trash and use the spectro since the speed difference would be of no great issue.


Then you display a Red, Green, Blue and white pattern and take readings of each.


The software makes a table of the difference between what the reference meter sees vs what the target meter sees.

Well, the gamma data is based on relative brightness levels so if the absolute level is off that would not affect the delta % values. Therefore, if everything is still linear, then the gamma before and after should look the same.


However, I'm having trouble with the luminance vs RGB levels. The luminance plot I have (see above) has all 3 colors displayed - not the luminance only plot. Therefore, if RED is low shouldn't it show up as low on the RED color luminance chart?


BTW does the i1 Display 2/LT have 3 or four sensors in it? I see the 3 RGB filters, but what is in the 4th hole? Is that a luminance sensor?



The profile procedure then only uses a single white point and the 3 primary colors RGB at a single stimulus (like 75%) to calculate a matrix to stretch or push the meter's (to be profiled) color space to match the reference meter. If things are linear this would appear to correct things. However, if things are not linear is this sufficient. Does the software require a full meter comparison after these 4 measurments to ensure accuracy across the board from low to high saturations and luminance values? The 2 meters should match for all readings not just the initial white and primary values.
 
#62 ·

Quote:
Originally Posted by PlasmaPZ80U /forum/post/21930562


I'll say it again, proper storage of these meters is essential to maximizing their lifespan (especially for those meters with completely exposed filter arrays). I'm not sure the OP's storage method was 100% optimal.

I believe you had/have an i1 Display 2 meter. I think you even stored it in a hermetically sealed (with desiccant) container. How much useable meter life did you get and what was the error that you experienced?


My feeling is that if temperature and humidity kill these meters the only safe place is the refrigerator in a sealed container. Would have been nice if the manufacturer said that! Cool and dry place is what just about everything made says on the label.
 
#63 ·

Quote:
Originally Posted by Plasma54321 /forum/post/21931134


Well, the gamma data is based on relative brightness levels so if the absolute level is off that would not affect the delta % values. Therefore, if everything is still linear, then the gamma before and after should look the same.


However, I'm having trouble with the luminance vs RGB levels. The luminance plot I have (see above) has all 3 colors displayed - not the luminance only plot. Therefore, if RED is low shouldn't it show up as low on the RED color luminance chart?


BTW does the i1 Display 2/LT have 3 or four sensors in it? I see the 3 RGB filters, but what is in the 4th hole? Is that a luminance sensor?



The profile procedure then only uses a single white point and the 3 primary colors RGB at a single stimulus (like 75%) to calculate a matrix to stretch or push the meter's (to be profiled) color space to match the reference meter. If things are linear this would appear to correct things. However, if things are not linear is this sufficient. Does the software require a full meter comparison after these 4 measurments to ensure accuracy across the board from low to high saturations and luminance values? The 2 meters should match for all readings not just the initial white and primary values.

That is the process to profile a meter with Calman. What the software does I do not know, all I can say is it works. My C6 reads very close to the same as the i1Pro once it is profiled. The difference is in the bottom end 10 and 20% stimulus, the i1pro does not work well at that low light level and the C6 readings are more accurate.


You really should read this article about Tri-Stim meters and how they work and why they don't always work well.
http://www.tlvexp.ca/2012/04/do-cali...-stim-devices/
 
#64 ·

Quote:
Originally Posted by Plasma54321 /forum/post/21931134


BTW does the i1 Display 2/LT have 3 or four sensors in it? I see the 3 RGB filters, but what is in the 4th hole? Is that a luminance sensor?


The meters actually measure in XYZ data. The filters are suppose to match the XYZ curves for the 1931 Standard observer color matching function.


A color matching function is the process that converts raw spectral data from watts per sterdian per nm into XYZ in candelas. X roughly equates to red, Y roughly equates got green and Z roughly equates to blue. The X curve has a little hump near around violet, the 4th sensor I believe it to pick up that data that would otherwise just be binned into the Z sensor data.


The Y value is what is used as luminance, obviously an actual green color has some X and Z components as well. As you can see any wavelength in the visible spectrum has some Y component.


 
#65 ·

Quote:
Originally Posted by Plasma54321 /forum/post/21931217


I believe you had/have an i1 Display 2 meter. I think you even stored it in a hermetically sealed (with desiccant) container. How much useable meter life did you get and what was the error that you experienced?


My feeling is that if temperature and humidity kill these meters the only safe place is the refrigerator in a sealed container. Would have been nice if the manufacturer said that! Cool and dry place is what just about everything made says on the label.

The first one I had I used briefly and returned it after calibrating my displays (got it from Amazon just to try out for the first time... had never calibrated with a meter before).


I bought another later on to keep and use for periodic calibrations and after 6 months or so I thought it had drifted since it profiled my PC monitors with pinkish grays. I got a free brand new replacement from X-Rite after explaining my issue and to my surprise it also left a pink tint to all grays when profiling my PC monitors. So, the 'defective' one I sent in was actually normal and even brand new ones made the grayscale too pink post-cal. So, at this point I had used three of these (non-enhanced meters) with ColorHCFR and they all made the overall image have a pink tint.


The last one I bought was a X2 with CalMAN, an enhanced D2 that only works with CalMAN. This was more accurate from day one and it stayed consistent for the whole time I owned it (as did the other ones I owned previously). Looking back, though, I didn't own any of these for longer than a year and so that may be why I never experienced any drift.


Conclusion: all the standard D2's/LT's I used made the overall image too red and no drift was experienced within the first 6 months to a year of use with proper storage. I kept them in two sealed ziploc bags with a dessicant inside. The enhanced X2 was more accurate initially and maintained its accuracy for as long as I owned it. I've sold the D2 I got as a replacement from X-Rite and my SpectraCal X2 as well.


Since, I never kept any of these past a year I cannot say much about drift over longer periods but I remember reading posts from other D2 owners who had their D2s last for up to 5 years. I'd say these can last anywhere from 2 to 5 years if stored in ideal conditions based on what I read on this forum. They can also go bad in six months or less if directly exposed to the elements (left out in the open without airtight bags or case and no dessicant... also exposure to light and heat).
 
#66 ·

Quote:
Originally Posted by Phase700B /forum/post/21929605


Just following up on Plasma54321 comment about Harbor Freight tools. Interesting you would seek out and have to comment ONLY on my post.

You are incorrect again - Plasma54321 made no such comments. Only airscapes and PlasmaPZ80U did, and I agree with airscapes' assessment - you get what you pay for. Your analogy was 180° off of his, and I pointed that out. No need to get accusatory or attack me over this, even though personal attacks and irrelevant insinuations have become the modus operandi of a few here who cannot participate on a rational, contributory level.

Quote:
Originally Posted by Phase700B /forum/post/21929605


Typical though.

Indeed, as we continue to wait for any pertinent points to be raised in response to my previous post.
 
#67 ·

Quote:
Originally Posted by airscapes /forum/post/21931271


That is the process to profile a meter with Calman. What the software does I do not know, all I can say is it works. My C6 reads very close to the same as the i1Pro once it is profiled. The difference is in the bottom end 10 and 20% stimulus, the i1pro does not work well at that low light level and the C6 readings are more accurate.


You really should read this article about Tri-Stim meters and how they work and why they don't always work well.
http://www.tlvexp.ca/2012/04/do-cali...-stim-devices/

That is a very good article and I also read it. . .. too late after I purchased my i1 D2 that was inaccurate and unacceptable in performance.


Some of us may have been more understanding had the vendors been more forthright in declaring they had sold marginal products and also ones that would not stand up to time and required such special storage and handling.
 
#68 ·
... And yet I've had zero issues with mine (so far) ... of course it's only ~8 months old ... no complaints ... no "oddly colored" greys.


I don't know of any colorimeters of this design vintage that were meant to "last" more than a year without being recalibrated.


Plus, I didn't spend $600 to calibrate a $524 display ... or spend 8 months trying to "eyeball" something that might remotely resemble a "flat" greyscale.



Sometimes a product is *exactly* good enough (for the price) for a particular task.
 
#69 ·

Quote:
Originally Posted by HDTVChallenged /forum/post/21933315


... And yet I've had zero issues with mine (so far) ... of course it's only ~8 months old ... no complaints ... no "oddly colored" greys.


I don't know of any colorimeters of this design vintage that were meant to "last" more than a year without being recalibrated.


Plus, I didn't spend $600 to calibrate a $524 display ... or spend 8 months trying to "eyeball" something that might remotely resemble a "flat" greyscale.



.


Hmmm. . . I wouldn't spend 8 months doing that either.



Nevertheless, I wouldn't spend another $300 to calibrate a $500 display and have it "age" while in storage so the next time I wanted to use it, it was no longer viable.


Good thing there are discs such as AVS HD709 and Disney WOW to help owners perform Media Assisted Settings.
 
#70 ·
Also, why should it be too much to expect even a $200-$300 meter to last and be viable?


In the audio world I have two vintage Technics SL-1700 MK2 Direct Drive turntables. They are loaded with sophisticated electronics and even optics for the tone arm functions in semi automatic operation. And they remain fully functional. These devices are elctromechanical with servo feedback drives to maintain speed accuracy and after 35 years still work just fine.


Then there are the 4 Dual turntable I have starting with a 1229 from 1972. Since then, I've added a 1229Q and two Dual 1219s. All of these are intricate mechanical and electrical devices that to this day work as intended as the day they were made. Many that were neglected are even restored to full operation with a little repair and clean up.


And do we even need to go into vintage phono cartridges, that are as delicate as feathers and still deliver superb audio delight?


All of the fore mentioned items and were, for the most part, only $30 to $175 when new.


Not to mention 25 year old laser disc players that still function just fine. These devices , utilize very intricate and sensitive electronics and still operate just fine. So, it's very difficult to excuse a meter that can't last 1 to 3 years or remain usable without purchasing another spectro to make "patches" and put band aids on profile tables to make it work properly. And then for only that particular TV or monitor they were profiled on. . . . and for how long??
 
#71 ·

Quote:
Originally Posted by Phase700B /forum/post/21933625


Hmmm. . . I wouldn't spend 8 months doing that either.



Nevertheless, I wouldn't spend another $300 to calibrate a $500 display and have it "age" while in storage so the next time I wanted to use it, it was no longer viable.


Good thing there are discs such as AVS HD709 and Disney WOW to help owners perform Media Assisted Settings.

This is one of the reasons we came up with the "rent a kit" program you don't have to worry about the meter being off from aging.
 
#72 ·

Quote:
Originally Posted by Phase700B /forum/post/21933625



Good thing there are discs such as AVS HD709 and Disney WOW to help owners perform Media Assisted Settings.

Who is to say Disney's WOW isn't without issues?
Quote:
Originally Posted by Phase700B /forum/post/21933642


Also, why should it be too much to expect even a $200-$300 meter to last and be viable?


In the audio world I have two vintage Technics SL-1700 MK2 Direct Drive turntables. They are loaded with sophisticated electronics and even optics for the tone arm functions in semi automatic operation. And they remain fully functional. These devices are elctromechanical with servo feedback drives to maintain speed accuracy and after 35 years still work just fine.


Then there are the 4 Dual turntable I have starting with a 1229 from 1972. Since then, I've added a 1229Q and two Dual 1219s. All of these are intricate mechanical and electrical devices that to this day work as intended as the day they were made. Many that were neglected are even restored to full operation with a little repair and clean up.


And do we even need to go into vintage phono cartridges, that are as delicate as feathers and still deliver superb audio delight?


All of the fore mentioned items and were, for the most part, only $30 to $175 when new.


Not to mention 25 year old laser disc players that still function just fine. These devices , utilize very intricate and sensitive electronics and still operate just fine. So, it's very difficult to excuse a meter that can't last 1 to 3 years or remain usable without purchasing another spectro to make "patches" and put band aids on profile tables to make it work properly. And then for only that particular TV or monitor they were profiled on. . . . and for how long??

You are going to beat this dead horse until it is pulp. Why do you keep comparing everything that you have that has lasted a long time to your colorimeter? The color filters have a known failure rate. This has been common knowledge for quite some time. Hopefully the sealed filters on the new i1 will last a lot longer. You are correct though that no one knows how long they are going to last, but hopefully it will be a couple of years.


As Derek said, if you don't want to deal with the uncertainty then rent a kit. You can even profile your colorimeter against an i1 Pro
 
#73 ·

Quote:
Originally Posted by Ericglo /forum/post/21934249



You are going to beat this dead horse until it is pulp.

He is in the anger phase.. has not accepted the loss yet.. but it in reality, what is done is done and no amount of crying will fix the problem.. time to get back on the horse and move on!
 
#74 ·

Quote:
Originally Posted by Ericglo /forum/post/21934249


Who is to say Disney's WOW isn't without issues?


You are going to beat this dead horse until it is pulp. Why do you keep comparing everything that you have that has lasted a long time to your colorimeter? The color filters have a known failure rate. This has been common knowledge for quite some time. Hopefully the sealed filters on the new i1 will last a lot longer. You are correct though that no one knows how long they are going to last, but hopefully it will be a couple of years.


As Derek said, if you don't want to deal with the uncertainty then rent a kit. You can even profile your colorimeter against an i1 Pro

+1 - thanks for bringing some rationality back to this discussion.
 
#75 ·

Quote:
Originally Posted by airscapes /forum/post/21934278


He is in the anger phase.. has not accepted the loss yet.. but it in reality, what is done is done and no amount of crying will fix the problem.. time to get back on the horse and move on!

I agree, you don't see me still complaining about my professional calibration that wasn't all it was supposed to be. It's prudent advice to consider such things as a learning experience and move on. Sometimes that's the best you can do.
 
#76 ·

Quote:
Originally Posted by sotti /forum/post/21931339


The meters actually measure in XYZ data. The filters are suppose to match the XYZ curves for the 1931 Standard observer color matching function.


The Y value is what is used as luminance, obviously an actual green color has some X and Z components as well. As you can see any wavelength in the visible spectrum has some Y component.

So if I understand what you are saying, the grayscale measurement only consists of the GREEN or Y component (of the XYZ system), therefore my OLD and NEW plots for luminace and gamma are similar.


Then, why when adjusting the white point, one balances all 3 colors to 100% as the goal, the colorimeter must read the RED and BLUE colors to allow this adjustment? I'm not getting it. Why some of my charts are similar and some are not? It seems like they should all be out of whack.
 
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