Apparently no one reads my pseudo mathematical ramblings. One stop
increase/decrease is a factor of 2, of course, and not the square root
of 2 (which is the area measure increase for an aperture to increase by
1 stop). Don't know where my brain was. 9:23 pm was not that late. :-)
Chuck Norcutt
On 9/17/2010 9:23 PM, Chuck Norcutt wrote:
> From your comment on the E-1 dynamic range test:
> "Now, for the opposite end of the spectrum. The Olympus E-1 sacrifices
> the highlights for preservation of the shadows. This is different than
> the Canon cameras which sacrifice the shadows for preservation of the
> highlights. Where "Expose-to-the-Right" is a recommended procedure for
> the Canons, exposing for the midtones or even the shadows is recommended
> for the E-1."
>
> Me thinks you don't really understand "expose to the right". It has
> nothing to do with sacrificing the shadows for the highlights or
> vice-versa. It's intended to avoid sacrificing anything. It's based
> strictly on the characteristics of the binary representation of the
> brightness of a pixel. Increasing the brightness level of any pixel
> (regardless of the camera that produced it) has a tendency to create
> posterization due to the multiplication and then integer rounding of
> small binary numbers. Decreasing the brightness has no such effect.
> Therefore if the exposure is not perfect it is better that it be
> overexposed and reduced in brightnes in post processing as long as the
> highlights are not blown.
>
> For example: I have 3 pixels with brightness 10, 11, 12. I increase
> their brightness by one stop by multiplying by the square root of 2 =
> 1.4142 which yields (after rounding) 14, 16, 17. But reducing the
> brightness of overexposed pixels 245, 246, 247 by dividing by the square
> root of 2 yields (after rounding) 173, 174, 174. After dividing down I
> end up losing some tonal information by creating two pixels of the same
> value but I have not created any tonality "holes" as in the jump from 14
> to 16 above.
>
> Chris posted a photo the other day that showed exactly this problem.
> The histogram of his image showed a lot of spikes and holes caused by
> brightness manipulation. This is also why you want to do any image
> editing in 16 bits even if the original image is an 8 bit JPEG.
>
> Chuck Norcutt
>
>
> On 9/17/2010 6:19 PM, Ken Norton wrote:
>>> Yeah! Let´s do it and show the charts to everyone. It took me too many hours
>>> to generate them. Norman Koren (The Imatest programmer) told me he was
>>> amazed with KODAK KAF-5101 sensor!
>>> Shadow information is awesome. Yeah!!! Highlight EV is very narrow. I think
>>> is the reason why E-1 underexposes.
>>> I am pretty sure many people thinks E-3 had wider dynamic range compared to
>>> E-1. :)
>>
>> LOL, not in my tests! The E-1 gives what appears to be at least one,
>> if not two stops more recoverable shadow detail than the E-3. I
>> regularily underexpose by a full stop and have to pull it up in
>> conversion. Not a problem at all. There are few cameras you can do
>> that with and still keep your sanity. Noise? Well, no E-1 image is
>> clean at any brightness level anyway, so it's not much of an issue.
>>
>> I think I adequately proved this point with this test:
>>
>> http://zone-10.com/cmsm/index.php?option=com_content&task=view&id=493&Itemid=1
>>
>> AG
--
_________________________________________________________________
Options: http://lists.thomasclausen.net/mailman/listinfo/olympus
Archives: http://lists.thomasclausen.net/mailman/private/olympus/
Themed Olympus Photo Exhibition: http://www.tope.nl/
|