how RAW is PEF

Hi,
occaisional I want to make astrophoto's.

Normally the noise reducing of the istD and K10D do the dark frame subtraction in camera.
This means that exposure time is doubled.

So, when you shoot a 2 minute exposure, you have to wait 2 minutes extra for the dark frame exposure.

Well, I thought, lets speed it up, I will switch off noise reduction, and take that darkframe exposure myself, shoot several 2 minute exposures, and subtract this dark frame later.
But it seems not working like that.

Close examening the raw files ( dcraw -v -D -4 -T filename.pef) reveals that some values in the 0..4095 range are missing.

And the interresting part is that they are at regular intervals, as if the raw ADC values have been scaled up to fill the 0..4095 rage as good as possible.

This of course makes the dark frame substraction outside the camera non functional.

Anybody else noticed this ?
Or better, anyone knows how to switch this off ?

George

What ISO value is this?

Earlier tries with the DS showed that the anything from ISO800, IIRC, was just scaled values.

--
Jonas

Jonas B said:

What ISO value is this?

Earlier tries with the DS showed that the anything from ISO800, IIRC,
was just scaled values.

Click to expand...

ISO 200, for both the istD and K10D

I see the scaling in both the istD and K10D PEF files, the K10D shows it to a lesser extend, probably because it comes from a higher resolution ADC.

George

I wrote a java program that takes a number of "white wall" images and draws a tone curve based on them. I take images of white sheet of paper starting at 5 or 6 stops below the meter reading and stepping by 1/2 stops up to 4 stops above the meter reading.

I used this to determine exactly where the saturarion point of the first color channel is (3 - stops above meter reading).

I get a smooth curve when I run this test. The program examines the RAW 12-bit integers with no processing.

I would expect considerable distortion if thee camera were spreding values.

--
Ed Hannon
http://www.pbase.com/edhannon

edhannon said:

I get a smooth curve when I run this test. The program examines the
RAW 12-bit integers with no processing.

I would expect considerable distortion if thee camera were spreding
values.

Click to expand...

I am not sure

Just take a histogram of all 12 bit values, if these shows a regular pattern of values with zero occurence, you know something has been done to the raw adc values.

George

My latest rev of the java reports min, max, avg and std deviation. I will take a sequence of "white wall" shots - being careful with the lighting to make sure it is even across field of view.

I will also add some code to capture a histogram of each image.

Will report the results when I get it done.

That should show any spreading or shifting.

--
Ed Hannon
http://www.pbase.com/edhannon