Re: Media Testing for maclife.de
Re: Media Testing for maclife.de
- Subject: Re: Media Testing for maclife.de
- From: Eric Chan <email@hidden>
- Date: Thu, 11 Sep 2008 06:10:19 -0700 (PDT)
I agree that calibration is about bringing a device to a known standard state and that profiling is about the measurement of a device while in a particular state. There is still the question of what you're actually measuring (more on this later).
Contrary to a suggestion made (much) earlier in this thread, Camera Raw (and Lightroom) __do__ perform color calibration based on data stored in the raw file, which can vary from unit to unit within a given model. Not all cameras provide such data, but for many of those that do, we do read the relevant calibration data and use it.
With regards to profiling, one must be careful to understand exactly what is being measured. As an analogy, when I typically profile a printer, I print out a target of some number of color patches, measure those patches using an instrument, and feed the data to profile-building software. As we know, "profiling a printer" doesn't tell the full story here. I am profiling much more than just a printer. I am profiling a big system, which has many components: the printer, certainly, but more specifically the exact hardware configuration of that printer (i.e., via the printer panel), also the inks, the paper, the printer driver, and any settings in the printer driver (such as the choice of media type, resolution settings, etc.). Change any one of those things and all bets are off in terms of the validity of the profile.
The same is true when "profiling a camera" by shooting a target (or sequence of targets) and feeding the measurements to profile-building software. Certainly the camera and sensor itself plays a role, but depending on exactly what the measurement data is, you may actually be profiling much more than just the camera. In most cases today, you are grabbing the data that comes out the back end of a raw converter, after it has been rendered using some non-specified algorithm (e.g., tone mapped, usually with some non-linear brightening and contrast-boosting curve). This is output-referred data and was one of the objections that Thomas raised back in the day when he was investigating methods to use to build color profiles: namely, that such profiles were not really measuring the behavior of the camera, but rather the camera + characteristics of raw converter.
Eric
--- On Thu, 9/11/08, edmund ronald <email@hidden> wrote:
> From: edmund ronald <email@hidden>
> Subject: Re: Media Testing for maclife.de
> To: "Bob Frost" <email@hidden>
> Cc: email@hidden
> Date: Thursday, September 11, 2008, 3:56 AM
> I'm not very good on terminology, I'm using the
> words mostly
> intechangeably. Certainly, any operation done on Tiffs and
> not Raw
> should only be called profiling.
>
> Calibration usually refers to a process closer to
> establishing
> characteristics of hardware. In that sense, I'd call
> measuring a
> sensor's reponse a calibration. There are devices out
> there which will
> do this for you.
>
> Of note however, with a modern LCD with LUTs one changes
> the behavior
> of the display by writing in the LUT. That is a certainly a
> calibration operation as it has changed the hardware's
> behavior.
>
> Cameras are a different story. Assuming you're the
> manufacturer, yes
> you will calibrate the unit in that sense too before
> shipping it :)
>
> Edmund
>
>
> On Thu, Sep 11, 2008 at 12:18 PM, Bob Frost
> <email@hidden> wrote:
> > Edmund,
> >
> >> If the CFA is badly matched to human
> >> vision or has spikes then a calibration may work
> on a target but that
> >> camera will make its users unhappy. Has happened.
> >> If the CFA matched (is that possible ?) then
> calibration with almost
> >> any patches would be trivial.
> >
> > I may be showing my ignorance here, so am I wrong, or
> are the words
> > 'calibration' and 'profiling' wrongly
> being used interchangeably in much of
> > this discussion?
> >
> > I always understood, but am subject to correction,
> that they are two
> > entirely separate processes. When I run iMatch, or
> BasicColor, or
> > ColorNavigator on my monitors, it is my understanding
> that they first
> > 'calibrate' the graphics card or monitor
> hardware, and then 'profile' the
> > resulting color, although I can opt to do just the
> latter, with no
> > 'calibration'.
> >
> > What I don't fully understand is the
> 'calibration' of a camera.
> >
> > Uli's work seems to be just concerned with
> 'profiling', not 'calibration'?
> >
> > Bob Frost.
> >
> >
> >
> > ----- Original Message ----- From: "edmund
> ronald" <email@hidden>
> > _______________________________________________
> > Do not post admin requests to the list. They will be
> ignored.
> > Colorsync-users mailing list
> (email@hidden)
> > Help/Unsubscribe/Update your Subscription:
> >
> >
> > This email sent to email@hidden
> >
> _______________________________________________
> Do not post admin requests to the list. They will be
> ignored.
> Colorsync-users mailing list
> (email@hidden)
> Help/Unsubscribe/Update your Subscription:
>
> This email sent to email@hidden
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden