Pages: 1 [2] 3 4 5   Go Down

Author Topic: The terms "linearization" vs "calibration"  (Read 27286 times)

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com
Re: The terms "linearization" vs "calibration"
« Reply #20 on: May 02, 2016, 08:24:27 pm »

Fair enough.
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

Bart_van_der_Wolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 8911
Re: The terms "linearization" vs "calibration"
« Reply #21 on: May 03, 2016, 04:33:17 am »

From Wikipedia:

The formal definition of calibration by the International Bureau of Weights and Measures is the following: "Operation that, under specified conditions, in a first step, establishes a relation between the quantity values with measurement uncertainties provided by measurement standards and corresponding indications with associated measurement uncertainties (of the calibrated instrument or secondary standard) and, in a second step, uses this information to establish a relation for obtaining a measurement result from an indication."

Does that clarify the definition?  :)

Hi Bill,

You're right, that doesn't really help. I knew about it, and therefore decided not to refer to it.

Maybe the ISO definition of the term "Calibration" is a bit more helpful (although essentially it says the same as the above definition).

This is from the standard: ISO/DIS 13655(en)
Graphic technology — Spectral measurement and colorimetric computation for graphic arts images


Quote
3.3 calibration
set of operations that establish, under specified conditions, the relationship between values of quantities indicated by a measuring instrument or measuring system, or values represented by a material measure or a reference material, and the corresponding values realized by standards
[SOURCE: ISO/IEC Guide 99 (VIM)]
Note 1 to entry: Contrary to a common usage, calibration is not the process of adjusting a measurement system such that it produces values that are believed to be correct. Calibration permits either the assignment of values of measurands to the indications (creating a reference table) or the decision to reset or adjust the device. Following the resetting or adjusting of the device, a calibration needs to be verified to ensure that the new device setting(s) provide indications within the accepted values.

Emphasis in bold is mine.

So Calibration establlshes the relationship between the measured values and the values as realised by standards. It's a reference table between input, and output in specific units.

'Realised by standards' means, calculated by methods described in standards, say a definition of the method of determining luminance, or chrominance, or density, and the units or coordinate system to use for those). Sometimes new (non-SI) units are created, like SFR (Spatial Frequency Response) instead of MTF, because they are not 100% identical, or different ISO speed ratings.

Cheers,
Bart
« Last Edit: May 03, 2016, 04:42:31 am by BartvanderWolf »
Logged
== If you do what you did, you'll get what you got. ==

hjulenissen

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2051
Re: The terms "linearization" vs "calibration"
« Reply #22 on: May 03, 2016, 04:46:49 am »

I’m researching printer calibration. As I pore over books, manuals, and guidelines from various manufacturers of RIP software and printing machines, I’m finding many seem to use the terms “calibration” and “linearization” interchangeably.

Do the experts use these terms as entirely interchangeable nowadays or are they differentiated somehow?

Comments, clarifications anyone?
I don't know this stuff.

Linearization sounds like something one would do to each channel individually (e.g. "gamma" or "gamma compensation"), while calibration might be the inter-channel part or the whole thing?

-h
Logged

GWGill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 600
  • Author of ArgyllCMS & ArgyllPRO ColorMeter
    • ArgyllCMS
Re: The terms "linearization" vs "calibration"
« Reply #23 on: May 03, 2016, 07:54:35 am »

So Calibration establlshes the relationship between the measured values and the values as realised by standards. It's a reference table between input, and output in specific units.
That would be a non-typical usage, more aligned with profiling, although it might come down to whether it is per-channel or the whole color response.

The usual meaning is the second one in your quote "or the decision to reset or adjust the device".
Logged

Bart_van_der_Wolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 8911
Re: The terms "linearization" vs "calibration"
« Reply #24 on: May 03, 2016, 09:52:33 am »

That would be a non-typical usage, more aligned with profiling, although it might come down to whether it is per-channel or the whole color response.

The usual meaning is the second one in your quote "or the decision to reset or adjust the device".

Hi Graeme,

I interpret the latter as an act based on the measurement result being far enough out of tolerance to warrant an adjustment,  a decision. This could be the result of measuring a reference tile which has know characteristics (which change very little over time).

Cheers,
Bart
Logged
== If you do what you did, you'll get what you got. ==

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 19959
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: The terms "linearization" vs "calibration"
« Reply #25 on: May 03, 2016, 12:45:36 pm »

So Calibration establlshes the relationship between the measured values and the values as realised by standards. It's a reference table between input, and output in specific units.
Not buying that either. We calibrate a display right? If there's a 'standard', it is as likely to work as not work because of the huge variables between display technologies, the goal of the end user (match a print? Under what illuminant?). I'd prefer to say calibration is the attempt to put a device into a desirable, well behaved condition and to be able to recalibrate to that aim point should the device alter it's behavior. Whatever the 'standard' for a display, unless it's a WP of CCT 5150K, 150cd/m2, 300:1 contrast ratio, that 'standard' will not produce the display behavior I specifically need to match a print next to my display! Using a specific viewing condition. And I'll often alter the CCT and contrast ratio based on the papers I'm viewing, again, next to that display. That's why one size doesn't fit all needs, and why SpectraView and the very few, similar products that allow us to build multiple calibrations for multiple needs are so darn useful.
Logged
Author “Color Management for Photographers" & "Photoshop CC Color Management" on pluralsight.com

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2177
Re: The terms "linearization" vs "calibration"
« Reply #26 on: May 03, 2016, 01:37:49 pm »

Not buying that either. We calibrate a display right? If there's a 'standard', it is as likely to work as not work because of the huge variables between display technologies, the goal of the end user (match a print? Under what illuminant?). I'd prefer to say calibration is the attempt to put a device into a desirable, well behaved condition and to be able to recalibrate to that aim point should the device alter it's behavior. Whatever the 'standard' for a display, unless it's a WP of CCT 5150K, 150cd/m2, 300:1 contrast ratio, that 'standard' will not produce the display behavior I specifically need to match a print next to my display! Using a specific viewing condition. And I'll often alter the CCT and contrast ratio based on the papers I'm viewing, again, next to that display. That's why one size doesn't fit all needs, and why SpectraView and the very few, similar products that allow us to build multiple calibrations for multiple needs are so darn useful.

Yep. This is the beauty of products like SpectraView and ColorNavigator. They combine a kind of calibration, ie: setting up the monitor so that it is optimized for profiling, and the actual profiling. Lesser software/hardware requires setting up max luminance, CCT, and black point behavior manually or semi-manually then profiling to that. That's adequate if you are only using one setup, one light booth setting, and papers that are sufficiently close to each other in white point and OB behavior.
Logged

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387
Re: The terms "linearization" vs "calibration"
« Reply #27 on: May 03, 2016, 02:03:31 pm »

Hi Bill,

You're right, that doesn't really help. I knew about it, and therefore decided not to refer to it.

Maybe the ISO definition of the term "Calibration" is a bit more helpful (although essentially it says the same as the above definition).

This is from the standard: ISO/DIS 13655(en)
Graphic technology — Spectral measurement and colorimetric computation for graphic arts images


Emphasis in bold is mine.

So Calibration establlshes the relationship between the measured values and the values as realised by standards. It's a reference table between input, and output in specific units.

'Realised by standards' means, calculated by methods described in standards, say a definition of the method of determining luminance, or chrominance, or density, and the units or coordinate system to use for those). Sometimes new (non-SI) units are created, like SFR (Spatial Frequency Response) instead of MTF, because they are not 100% identical, or different ISO speed ratings.

Cheers,
Bart

Bart,

The wording of these definitions is dense, and this indicates that the process of calibration is more complicated than a lay person might think. The process involves comparison with a standard, which might be a primary standard or a secondary standard that is traceable to the primary standard. The decision making process involves consideration of precision and accuracy of both the instrument being calibrated and that of the standard itself.

If I calibrate my NEC monitor with Spectraview,the process is abbreviated. There is no luminance or chromaticity standard to check the calibration of the Spectraview device before each use. Presumably, the device was calibrated at the factory to a specified accuracy and we hope that there has been no drift since the calibration. After the calibration is complete, the instrument reports some delta-Es, but there is no opportunity to address accuracy and precision of the results.

Cheers,

Bill
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 19959
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: The terms "linearization" vs "calibration"
« Reply #28 on: May 03, 2016, 03:38:19 pm »

Presumably, the device was calibrated at the factory to a specified accuracy and we hope that there has been no drift since the calibration. After the calibration is complete, the instrument reports some delta-Es, but there is no opportunity to address accuracy and precision of the results.
Calibrated for what? I hear folks talk about factory calibration and various displays. Again, targeted, calibrated for what? To achieve what goal? Now I can select sRGB emulation in SpectraView, that's clearly calibrating to that 'standard' aim as best it can (without P22 phosphors?) and to be honest, I've never measured what the sRGB calibration provides and how close it really is to sRGB as specified. And accurate in what way? What's the reference values, how many, and what's the measured values and the dE produced? Without those vital pieces of data, the use of the term color accuracy isn't worth the paper it's printed on.
Logged
Author “Color Management for Photographers" & "Photoshop CC Color Management" on pluralsight.com

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387
Re: The terms "linearization" vs "calibration"
« Reply #29 on: May 03, 2016, 05:36:37 pm »

Calibrated for what? I hear folks talk about factory calibration and various displays. Again, targeted, calibrated for what? To achieve what goal? Now I can select sRGB emulation in SpectraView, that's clearly calibrating to that 'standard' aim as best it can (without P22 phosphors?) and to be honest, I've never measured what the sRGB calibration provides and how close it really is to sRGB as specified. And accurate in what way? What's the reference values, how many, and what's the measured values and the dE produced? Without those vital pieces of data, the use of the term color accuracy isn't worth the paper it's printed on.

That was the point of my post and I don't understand your criticism. By calibration, I was referring to the Spectraview puck. It is accurate if the luminance presented to it is the value it reports. In other words, in terms of accuracy, if the standard is 120 cd/m^2, the instrument reports 120 cd/m^2 it is accurate to the nearest integer value. For a proper calibration, one would have to repeat the process 10 or 20 times, comparing the reported value with the standard. The mean reported value difference from the standard would represent the accuracy and the standard deviation indicates the precision.

Bill
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 19959
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: The terms "linearization" vs "calibration"
« Reply #30 on: May 03, 2016, 05:43:11 pm »

That was the point of my post and I don't understand your criticism.
NO criticism, simply a question. Calibrated to what, for what? I thought we were talking about 'factory calibration'. Some displays have this 'promise' hence the question.
Logged
Author “Color Management for Photographers" & "Photoshop CC Color Management" on pluralsight.com

Erland

  • Full Member
  • ***
  • Offline Offline
  • Posts: 129
Re: The terms "linearization" vs "calibration"
« Reply #31 on: May 04, 2016, 04:28:01 pm »

In my world, the digital printing world, linearization and calibration are two different things, were a linearization is something we do every week, but a calibration is to some people linearization, and for some, like me, calibration means profiling in combination with a fresh linearization is done only once per paper.

Linearize is measuring 21 patches of cyan, 21 patches of magenta, 21 yellow and 21 black. This is to see the max and miniumun density, creating a linear curve from zero, to the densest possible patch within each color. This is later put into our Rips, and it calculates how to produce an even as possible output. Meaning CMY' "gray" to be as neutral as possible, and also as consistent from dark to light.

However, linearization does not care about what shade our Cyan is, or any other color for that matter. This is where profiling comes in. After a linearization, we profile the paper, as you do with a "Normal" inkjet printer as well. Here, the shade of cyan, magenta and greens and all the other colors producable count! Cyan in photoshop, fogra 39 colorspace, is not the same shade as our Cyan, and if you would to keep pure primarys in your output, it would differ from what you see on your screen. An often made complaint from old Offset printers is that there is magenta in their "pure" cyan, even though they "calibrated" (meaning linearizing). But when I explain that cyan from our printer does not have the same shade as our competitors or their old ink.
Logged
Service Technician Digital Printers and Peripherals.
Epson Stylus Photo 1400.

GWGill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 600
  • Author of ArgyllCMS & ArgyllPRO ColorMeter
    • ArgyllCMS
Re: The terms "linearization" vs "calibration"
« Reply #32 on: May 04, 2016, 08:05:12 pm »

However, linearization does not care about what shade our Cyan is, or any other color for that matter. This is where profiling comes in.
Profiling is not Calibration though. Calibration changes the device behavior, profiling doesn't.
Logged

Stephen Ray

  • Full Member
  • ***
  • Offline Offline
  • Posts: 213
Re: The terms "linearization" vs "calibration"
« Reply #33 on: May 04, 2016, 11:03:29 pm »

a linearization is something we do every week

Erland,

I'm curious as to what brand of software (RIP?) you are using to do this linearization. Does it have a button that says, "linearize" and / or does it have a button that says "calibrate?"
Logged

Erland

  • Full Member
  • ***
  • Offline Offline
  • Posts: 129
Re: The terms "linearization" vs "calibration"
« Reply #34 on: May 05, 2016, 05:06:59 pm »

I mostly use EFI's Fiery Rips, both using Color Workstation as well as Color Profiler suite. Creo Rips. And Our own newly developed Totalflow rip. The latter you use with X-rites i1Profiler to do a linearization and the former the built in utility.
Logged
Service Technician Digital Printers and Peripherals.
Epson Stylus Photo 1400.

Erland

  • Full Member
  • ***
  • Offline Offline
  • Posts: 129
Re: The terms "linearization" vs "calibration"
« Reply #35 on: May 05, 2016, 05:10:40 pm »

Profiling is not Calibration though. Calibration changes the device behavior, profiling doesn't.

Sure enough. The only thing that actually change our hardwares behaviour is the built in process control. Patches with gradually increasing patches of each toner is placed on the ITB (image transfer belt) and is then read by a sensor, reading the density of each patch. Neither LInearization, calibration, nor profiling is changing the behavior of the hardware.
Logged
Service Technician Digital Printers and Peripherals.
Epson Stylus Photo 1400.

GWGill

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 600
  • Author of ArgyllCMS & ArgyllPRO ColorMeter
    • ArgyllCMS
Re: The terms "linearization" vs "calibration"
« Reply #36 on: May 05, 2016, 07:51:37 pm »

Neither LInearization, calibration, nor profiling is changing the behavior of the hardware.
It's not helpful to say this based on pure technicalities.

Technically few modern devices "modify the hardware" - they all change behavior with electronic control modifications. That's not relevant. What's relevant as a conceptual anchor point is that everything that changes the device behavior in such a way that you must create a new ICC profile (or allows you to avoid creating a new profile in the face of unwanted changes in the output), is calibration.





« Last Edit: May 05, 2016, 08:03:12 pm by GWGill »
Logged

Erland

  • Full Member
  • ***
  • Offline Offline
  • Posts: 129
Re: The terms "linearization" vs "calibration"
« Reply #37 on: May 06, 2016, 02:37:50 am »

While I agree, maybe in today's translations between different languages, calibration in Swedish, actually means something else and thus the different views on what calibration is. I can not develop this further, just wanted to give my sight to this interesting discussion.
Logged
Service Technician Digital Printers and Peripherals.
Epson Stylus Photo 1400.

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387
Re: The terms "linearization" vs "calibration"
« Reply #38 on: May 06, 2016, 05:50:02 am »

So Calibration establlshes the relationship between the measured values and the values as realised by standards. It's a reference table between input, and output in specific units.

'Realised by standards' means, calculated by methods described in standards, say a definition of the method of determining luminance, or chrominance, or density, and the units or coordinate system to use for those). Sometimes new (non-SI) units are created, like SFR (Spatial Frequency Response) instead of MTF, because they are not 100% identical, or different ISO speed ratings.

Not buying that either. We calibrate a display right? If there's a 'standard', it is as likely to work as not work because of the huge variables between display technologies, the goal of the end user (match a print? Under what illuminant?). I'd prefer to say calibration is the attempt to put a device into a desirable, well behaved condition and to be able to recalibrate to that aim point should the device alter it's behavior.

The DigitalDog thinks of calibration as applied to digital photography where the output of the device being calibrated can be adjusted electronically. However, in the scientific community there are devices where the output of the device can not be adjusted in the calibration process, and calibration merely establishes the relationship between the measured values and the values as realised by standards as Bart suggests. It's a reference table between input, and output in specific units.

An example of such a calibration exists with high precision mercury thermometers. The readings are taken with reference to the etching marks on the glass, and these can not be changed. The calibration is established by comparing the output of the thermometer with a thermometer traceable to NIST (formerly the NBS, National Bureau of Standards). The calibration may carried out at 4 points on the main scale and the user is provided with a table indicating the results of the calibration. An example is here.

Bill
Logged

Bart_van_der_Wolf

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 8911
Re: The terms "linearization" vs "calibration"
« Reply #39 on: May 06, 2016, 08:33:58 am »

The DigitalDog thinks of calibration as applied to digital photography where the output of the device being calibrated can be adjusted electronically. However, in the scientific community there are devices where the output of the device can not be adjusted in the calibration process, and calibration merely establishes the relationship between the measured values and the values as realised by standards as Bart suggests. It's a reference table between input, and output in specific units.

Absolutely agree with that summary. And even in the case of calibration with an adjustable device, the values are within a given tolerance. In order to increase the reliability, linearization can help and also provide some statistical confidence output.

Quote
An example of such a calibration exists with high precision mercury thermometers.

Yes, a good example. Even with the possibility to calibrate at 0 Celsius / 32 Fahrenheit in melting ice, and at 100 Celsius / 212 Fahrenheit in steam from boiling water at a standard 1013.25 hPa barometric pressure, there may be minute differences in the diameter of the glass column that introduce non-linearities in the intermediate read-outs (interpolations).

So additional precaution is taken (in your example of reference thermometers) to add two more intermediate reference points, thus reducing the interpolation error and also supplying some sense of how (non)linear the known physical expansion actually is, due to external influences (the glass tube diameter). This is the linearization part.

As for spectrometers, they are usually adjustable (by updating either voltages, or lookup tables in firmware), so that the specific output units are within a given tolerance/precision range and in units that can be converted to other units.

Additional issues that affect accuracy, are electronic drift (long term due to aging of electronic components, but also short term due to temperature changes which hopefully is compensated for) and the changes in the lightsource. Multiple short spot measurements with a device that has a halogen bulb as lightsource will e.g. introduce drift due to Tungsten filament deposits on the inside of the quartz bulb because it doesn't become hot enough. Prolonged measurements that heat up the bulb enough will then allow the deposit to redeposit onto the filament, so in a way mechanically resetting the bulb's influence to a known more stable state.

It's an interesting topic.

Cheers,
Bart
« Last Edit: May 06, 2016, 11:24:59 am by BartvanderWolf »
Logged
== If you do what you did, you'll get what you got. ==
Pages: 1 [2] 3 4 5   Go Up