W3C home > Mailing lists > Public > www-style@w3.org > September 2008

Re: [css3-color] ICC implementation

From: Chris Murphy <lists@colorremedies.com>
Date: Mon, 15 Sep 2008 11:15:33 -0400
Cc: www-style@w3.org
Message-Id: <C3F7DDE4-CCB3-4B99-9C08-F8A91558BC09@colorremedies.com>
To: David Woolley <forums@david-woolley.me.uk>

On Aug 31, 2008, at 8:47 AM, David Woolley wrote:

>
> Chris Murphy wrote:
>
>> This is precisely why photographers are encouraged to "expose  
>> right" meaning to the right side of a histogram, which are quarter  
>> tones and highlights. It is preferable to correct in post- 
>> processing what appears to be a slightly overexposed images. This  
>> results in a higher quality capture.
>
> In any system with quantisation, or other noise, I would have  
> thought it was self evident that one should maximise the use of the  
> dynamic range.  However, my experience is that digital media  
> tolerate over-exposure very badly.

I did state "what appears to be" not actually overexposed. A correctly  
exposed image on a digital camera is more exposed than if you were  
shooting chrome, but probably not as much as if you were shooting  
negative.


>  That I think is also self evident, as over-exposed values are hard  
> clipped, whereas underexposure simply results in increased noise.   
> That isn't to say that quantisation noise isn't a problem with  
> typical 8 bit storage formats.  In my experience, for amateur use,  
> one, or even half a, stop over is unacceptable, whereas two stops  
> under is still usable. (Most scenes have some small highlights that  
> can reasonably be clipped, of course.)

And you're determining correct exposure with a light meter? Or you're  
determining this 1/2 stop over, and 2 stop under anecdote with the  
camera metering itself? What is this camera and how old is it?

It could also be that your camera's in-camera JPEG is adept at  
resurrecting dark images, and sucks really badly at dealing with  
slightly overexposed data. If it were possible to access the Raw data  
from such a capture, assuming the scene was metered and correct  
exposure determined, almost every time I would take 1/2 stop over than  
2 stops under because there are simply more bits for tools to recover  
highlight data than shadow detail.

Chris Murphy
Received on Monday, 15 September 2008 15:17:08 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 27 April 2009 13:55:12 GMT