Minolta

On Thu, Jun 25, 2009 at 4:29 PM, Chuck Cole<[EMAIL PROTECTED]> wrote:
>> -----Original Message-----
>> From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf
>> Of Aaron Bredon
>>
>> >> On Thu, Jun 25, 2009 at 2:47 PM, Chuck Cole <[EMAIL PROTECTED]>
>> >> wrote:
>> >> > High contrast copy is the opposite: it has a very abrupt change from
>> >> > recording nothing (no matter the time) to the negative going all
>> >> > black.
>> >>
>> >> This is trivial to simulate with Photoshop or almost any other image
>> >> editor.
>> >
>> > What you are describing certainly is, but what I was discussing isn't.
>> > The
>> > transition is not a pure step,
>> Which is why I also suggested Curves. Curves can add contrast in a
>> controlled manner to a specific range of shades of the image.
>>
>> > and that is the essential
>> > image physics needed. Photoshop cannot add image that simply was not
>> > recorded.
>>
>> Unless the change in tones is <1/1000 of a stop, a good DSLR will record
>> it
>> when you take the picture in RAW mode and expose properly.
>>
>> > Without the actual sensor or film level physics, the tiny shades of gray
>> > in
>> > this case are not captured at all. Maybe a grossly
>> > underexposed DSLR image would suffice, but this is very hard because the
>> > full moon is only a few stops down from the sun.
>>
>> Actually you would want an almost-OVERexposed DSLR RAW image (a DSLR will
>> record 2048 distinct values in the highest stop recorded, 1024 values in
>> the
>> next highest stop, 512 values in the 3rd highest stop, and so forth).
>> If you are trying to distinguish subtle shadings, you want to put them
>> in the highest
>> stop that the camera records.
>
> What you seem to be missing is that the function needed is to apply those
> 2048 levels over about 1/2 stop range in the initial
> exposure. Having a dynamic range that is larger is totally unrelated to this
> case.
>
> .. or am I missing what you are saying about using that 2048 bit range?

Those 2048 values are over a single stop range.
(DSLRs record a total of 4096 values, but in a linear fashion -
2048 values are in the brigtest stop alone,
1024 in the 2nd brightest stop, 512 in the third brightest, etc - there is
effectively no detail recorded in the 11th brightest stop.)

If you are trying to enhance contrast in a 1/2 stop range,
and you can place that just under the brightest that
the sensor can capture, you should be able to distinguish
more than 1000 shades in that 1/2 stop after processing.
(there might be some reduction caused by noise,
but you should still be able to expand the 1/2 stop to
a high quality 8-bit grayscale with room to spare.)

The advantage that an astrophotography CCD provides
compared to a DSLR is vastly reduced noise due to
sensor construction, quality control, and cooling
among other factors.

-- 
Aaron Bredon
http://thecompressionist.blogspot.com