DSLR Flats Issue

[pfile]

i think the low values probably have something to do with black point correction being turned off in the "raw" format hint in PI. i can see that the levels are very different if you let DCRAW debayer a frame vs. setting DCRAW for "raw" and then using the Debayer process to debayer the raw DSLR frame.

i had earlier brought up the bias because it's very important to calibrate your flats with a bias frame, or else bad things happen when the flat scaling factor is computed. are your flats bias subtracted? or you can make flat darks, same effect.

i think the suggestion of mid-histogram flats when the 'linear' box is checked in Digital Photo Professional is good. letting AV compute the flat exposure will lead to a very underexposed flat.

[Ignacio]

The other recommendation is to use a rather dim source of light for the flats, so that a relatively long exposure is required to achieve middle level histograms. Otherwise, dslr shutters can introduce uneven illumination. I would recommend at least exposures of 1/3 of a second or longer.

Ignacio

[ChrisEyhorn]

That makes sense, I will increase my exposure time and decrease my panel brightness.

Will report back with results.

Chris

[ChrisEyhorn]

So just to be clear, on my STX which has the 16803 chip, I should be shooting for values around 30,000 (16bit chip) and not 50,000 (100k well count)?

[Ignacio]

No experience with the STX, but if you can set the gain, I would do it so that fullwell corresponds to maximum count (ie, 64k), and then shoot for an average count of 32k.

Ignacio

[ChrisEyhorn]

Hey Guys,

Ok I have done some more technical research for flats and here is what I have come up with.

The peak brightness level of a flat should be about one half to two-thirds of the saturation level.

For a given chip, the saturation level is represented by this formula.

Saturation = full well in electrons / (A/D Gain)

For example, for my SBIG STX (16803 chip)

Full Well in electrons = 100,000
A/D Gain = 1.27

Saturation = 78,740

So my flats peak brightness should be between 39,370 and 52,493.

And for my 5D Mark II:

Full Well in electrons at each ISO setting:

   50  =  65700
  100 =  59400
  200 =  29700
  400 =  14800
  800 =  7425
 1600 = 3710
 3200 = 1860
 6400 =  930
12800 = 460
25600 = 230

Since each ISO level represents the gain then the full well capacity above then really is the saturation level.

So my target flat value is then really based on ISO setting and here is the range for each ISO setting.

   50  =  32,850 - 43,800
  100 =  29,700 - 39,600
  200 =  14,850 - 19,800
  400 =  7,400 - 9,867
  800 =  3,712 - 4,959
 1600 = 1,855 - 2,473
 3200 = 930 - 1,240
 6400 =  465 - 620
12800 = 230 - 307
25600 = 115 - 153

Is this inline with what other people have observed at each given ISO level for their flats using a DSLR?

I will give this a test go later tonight and see what I come up with.

Chris