[OM] Re: olympus Digest V1 #128

Subject: [OM] Re: olympus Digest V1 #128
From: "C.H.Ling" <chling@xxxxxxxxxxxxx>
Date: Fri, 4 Jun 2004 23:05:19 +0800
----- Original Message ----- 
From: "Stephen Scharf" <scharfsj@xxxxxxxxxxxxx>

> I guess if this were correctable by using C1, then the data was
> actually there and being recorded by the camera; what you're saying
> if I take it correctly is that you feel the JPEG conversion was
> dropping shadow detail to black too quickly?

It didn't totally dropped, the data is still there but level was not right
and not a simple (or natural) curve can correct.

> ><SNIP>

> The correction was done simply by evaluating the R/G/B numbers with
> the sampler tool and correcting with Curves what is obviously  a
> color cast, according what to the numbers say. A levels examination
> showed me where the black and white points were, and I adjusted the
> levels black and white input sliders very slightly to match that
> according to the photo histogram. For the curves adjustment, I used
> 16 as the black point and 245 as the white point, and 128 as the
> midtone point. I applied the color correction to each color channel
> independantly.

First of all I will never assume there is a common black and white point on
RGB for this low contrast scene, without seeing the actual scene it is hard
to judge what the original color was, you just adjust the color based on
what you "think" (just like using auto white balance in PS). BTW, good color
doesn't mean it must be always accurate. I will not try to force a sunset
protrait to look like a day time shot.

> But these details are really besides the point- the point was to show
> by example that all image capture devices have inaccuracies in their
> recording of color, whether it's an E-1 or a 10D. I just didn't agree
> that one could take the un-corrected image out of one camera and
> another and say that the color from one is good and the color from
> another is bad, when they both are inaccurate; they are just
> inaccurate in different ways.

Yes, but the most important is how it looks, some image can shift to super
natural look but some shift naturally, which already make big different.

> As for your monitor, you can set it to whatever you like, however,
> most color management professionals recommend D65 (6500K) for
> photography work; especially if the images are veiwed via computer
> monitor or printed via photo (RGB-emulting) printers, and viewing of
> the print at 5000K in a controlled lighting booth or daylight. 5000K
> is usually regarded as too yellow for accurate color output for these
> applications, and really intended for prepress color management types
> who go to CMYK press using paper stocks that have a rather yellowish
> base, as is frequently the case in the U.S.
> -Stephen.

This is the first time I heard about this theory, all professional tools I
have seen is 5000K. May be you are right but anyway using my workflow I can
get what I wanted. Just like now I mainly do Fuji Frontier output, with a
proper profile in Photoshop I always get closely matched output.


List usage info:     http://www.zuikoholic.com
List nannies:        olympusadmin@xxxxxxxxxx

<Prev in Thread] Current Thread [Next in Thread>