Flat DSLR ?

[Herbert_W]

Hi,

When one opens DSLR Images, its necessary to check before the DSLR Raw Preferences in FormatExplorer.
See the image - all settings are shown (flat from canon 20Da).

My questions are:
1)
Why is there a big difference in the median values for the left and right image?
Median left image (b/w) = 0,01749, median right image (color) =  0.0(!) 0.00449 0.0(!)
2)
What should be a correct value in statistics for a DSLR flat?



Maybe the questions are stupid, but ...

Best regards.
Herbert, Austria

PI 1.8RC1, win7 64bit

[georg.viehoever]

If you dont activate "No Blackpoint correction", dcraw (the library underlying PI for reading RAWs) will fix raw values based on some camera specific adjustments. When working in PI, you should always work with "No Blackpoint correction" activated. Only then, PI sees the true values as contained in the CR2 file.

For my CanonEOS40D (14 bit DAC): Typical values for dark pixels in bias frames are around 1024/2^16. Flats- if the peak is about central in Canons preview, is around 2^12/2^16. Saturated pixels are around 2^14/2^16=0.25.

Georg

[dzso.bacsi]

Hi Georg,

I am very new to astrophotos, PI and to a modded Canon 600D too. I "fight" to find out somehow how to make good flats. So I just want to understand your reply. 
You say

If you dont activate "No Blackpoint correction",...

and

When working in PI, you should always work with "No Blackpoint correction" activated.

In Herbert's post it can be clearly seen that he did activate "No blackpoint correction" in both screen. Do I misunderstand your answer?
I think that the numbers may vary from camera to camera. Is it a good approach to good flats to get a strong peek in the middle range of the histogram?

Cheers

[georg.viehoever]

Sorry, I did not see the right half of the image fully in the other browser.

- when you select "Create RAW Bayer image", each pixel in PI is either red, green or blue. For a red pixel, the other two values are 0. Now if you compute the median of an image where 3/4 of all pixels have R values of 0, the median is also 0. The same holds for B. For green, 50% of all pixels are 0, so the median may be !=0 under certain conditions. Thats what we see in the right half of the image.

-  I usually do my flats with a strong peak approximately in the center of the histogram as shown by the Camera. Its good to have a strong signal that at the same time is not to bright (i.e. has pixel values that are overexposed).

Georg

[dzso.bacsi]

Thank you Georg!