Problems with noise in stack when using flats

[Emiel Kempen]

Pfile and Rüdiger,

I used these RAW format Preferences.

Regards,
Emiel.

[ruediger]

Values from "masterflat-with-t-shirt" look reasonable, also your RAW settings are correct.
Is the final stack with this flat now OK or are you still experiencing noisier results?

[Emiel Kempen]

Hello Rüdiger,

What do you mean with:

Values from "masterflat-with-t-shirt" look reasonable

Could you tell me what there is to improve or make better?

I have generated a stack calibrated with the "T-shirt" flats and compared it with the stack calibrated with the "plastic" flats.

plastic
Calculating noise standard deviation...

* Channel #0
?R = 5.042e-005, N = 576142 (57.61%), J = 4

* Channel #1
?G = 3.095e-005, N = 464134 (46.41%), J = 4

* Channel #2
?B = 1.809e-005, N = 101338 (10.13%), J = 4

tshirt
Calculating noise standard deviation...

* Channel #0
?R = 3.288e-005, N = 194157 (19.42%), J = 4

* Channel #1
?G = 3.055e-005, N = 387924 (38.79%), J = 4

* Channel #2
?B = 3.329e-005, N = 271691 (27.17%), J = 4

The resulting images are attached.

Regards,
Emiel.

[ruediger]

What do you mean with:

Values from "masterflat-with-t-shirt" look reasonable

Could you tell me what there is to improve or make better?

I looked for the median values, which are higher (6000-9000 ADU) than before. Also there is no single channel having much less ADU than the other. The remaining standard deviation (about 40 ADU) as a measure of the photon noise is low enough, so that this master flat can't introduce much additional noise to your light frames.

The master flat is used this way during calibration:

calibrated light = light * ( mean(flat) / flat )

Fo a noisy pixel from the master flat that lies 1*sigma away from the mean, the light ADU will change during calibration by a factor of 9000/9040 = 0.995.

For all master flat pixels, that lie between -1*sigma ... 1*sigma (=68% of all pixels), the additional noise that get added to the light frames is a maximum of 0.5 ADU (assuming a 100 ADU background value)

So I doubt you can enhance your flats much more. The sources for any remaining quality problems must somewhere be hidden in the overall calibration and stacking process.

Rüdiger

[Emiel Kempen]

Hello Rüdiger,

OK, well I learned a lot so far. I will try to find out if there is another source of noise in the workflow.

Thanks!

Regards,
Emiel