Re: `Lossless' retroactive embedding of color profile (avoid JPEG recompression); basic color profile theory
Re: `Lossless' retroactive embedding of color profile (avoid JPEG recompression); basic color profile theory
- Subject: Re: `Lossless' retroactive embedding of color profile (avoid JPEG recompression); basic color profile theory
- From: Marco Ugolini <email@hidden>
- Date: Tue, 01 Aug 2006 23:53:57 -0700
In a message dated 7/18/06 12:54 AM, Moritz Heckscher wrote:
> In other words: Do you think that the raw data in an image without a
> profile and in the same (hypethetical) one with a profile would be
> the same, except for the added colour profile? Or does adding
> (embedding) a colour profile actually alter the RGB data of the
> pixels in the image? (I think it shouldn't.)
I compare this distinction between assigning and converting to what happens
with a text written in a language unknown to the reader.
- Assigning a profile is like discovering what the language is, and then
making sense of it in the original (after trying other languages that proved
incorrect).
- Converting, instead, is like translating a text from one known language to
another: the words in the text will change (just like the color numbers in a
file will change), but hopefully, if the translation is done well, the
*sense* ought to remain very close to the original (just as in a color file
the *appearance* ought to be as close as possible to the source).
> - If no profile is embedded, is it (more or less) safe to assume that
> embedding the sRGB profile (afterwards) will be the `right' choice?
> Or which profile should I chose? Is there a way to tell, or is the
> information regarding which profile is the `right' one irreversibly
> lost since the profile is not embedded?
Most of the time the "right" profile is a crapshoot anyway, because even
when a camera embeds a profile, it's not necessarily the one that describes
the color of the file in the best manner possible. Specially when you
consider that profiling a camera still remains a daunting task, given the
myriad possible shooting scenarios.
One viable option is to start out with a calibrated and profiled monitor,
open the file in Photoshop, then assign any of a series of RGB work spaces,
then another one, and so on, until what appears on the monitor looks closest
to what you consider "right". Then embed that profile. It's a procedure that
involves the user's visual evaluation, but demands that the equipment be
well-behaved, i.e., properly color-managed.
> I've read somewhere that it
> is usually a good assumption that devices such as cameras use sRGB if
> not configured specially and not embedding any profile.
It's certainly what most people end up suggesting when asked, but I question
its validity as advice that applies to most or even a majority of cases.
> That assumption could then be rephrased as follows:
> - If the camera is set up to embed a profile, it captures the real
> data from the sensor, performs some calculations appropriate for the
> chosen profile and then writes out that data with the profile.
I highy doubt that the camera performs the "calculations appropriate for the
chosen profile." Can anyone else confirm or deny?
> - If the camera is set up not to embed a profile, it does the same as
> above, using the calculations for sRGB, but does not embed the sRGB
> profile.
Again, maybe not.
Regards.
--------------
Marco Ugolini
Mill Valley, CA
_______________________________________________
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