flat frames for DSLR

[lucchett]

Hi,
I am trying to help a friend with flats and calibration.

I am used to ccd, and determined the linearity curve of my detector, so I know the perfect spot in terms of adu counts, in the acquisition software.

I wonder how to do this with a dslr.

when do we need to measure the average lightness? (original raw, fit converted, other?)

I don't use Batch Preprocessing script with my images, but I assume it would be needed to flag the CFA option in case of DSLR, right?

Many thnaks,
Andrea

[pfile]

i think you can use the same method for determining where the sensor goes nonlinear. you can use SplitCFA to get a mono image for each of the 4 filter 'positions' in a CR2, or you can use superpixel debayering, but this will average the two green filters together.

note that since CR2 data is only 14 bits deep, and the 14bit data gets put into 16-bit integer images, a fully overexposed flat would read 0.25 (16383/65535)

rob

[lucchett]

Thank you Rob.
If I understand correctly, your advice is to ensure linearity for the single pixels in the matrix.
that said, the simple average in maxim in the cr2 file doesn't work (or at least I should be a bit conservative on the max adu count)

thank you also for the note, I didn't realized that and now can explain the weird flats of my friend (fully overexposed :-))


Andrea

[gvanhau]

Hello

When taking flats with my DSLR's I normally use AV setting.
This way the Camera evalutes the correct expo for flat frames.

Regards
Geert

[lucchett]

Hi Geert,

It seems very easy, it is worth to try.
Thank you,
Andrea

[msmythers]

I do the same thing as Geert with my Sony. There is an added bonus for me since I shot diffused sky flats at dusk or dawn. With the camera making the adjustments during the changing sky brightness my flats have roughly the same exposure level from start to finish.


Mike

[pfile]

Thank you Rob.
If I understand correctly, your advice is to ensure linearity for the single pixels in the matrix.
that said, the simple average in maxim in the cr2 file doesn't work (or at least I should be a bit conservative on the max adu count)

thank you also for the note, I didn't realized that and now can explain the weird flats of my friend (fully overexposed :-))


Andrea

yes, i was just saying that you can make the same graph that you would with a CCD - ADU count vs. exposure time, and see where the sensor stops behaving linearly. i've never bothered to do this with my DSLR - i've simply aimed for 1/2 well depth (peak of histogram at 0.125).

what do you mean by "the simple average does not work"?

another note about OSCs in general is that the flat does not necessarily have to be grey, but it helps if the channels are of similar ADU count. when using light pollution filters sometimes you get a cast in the flat that results in a weak red channel, so your flat has poor SNR in the red. i solved this problem by using slightly pink light (slightly pink T-shirt).

you can use AV but the camera will be very conservative about the exposure, since its computer assumes that the data is going to be stretched with the default curves. but of course flats are never stretched, and the brighter the better (while still linear of course.) you can do +1EV or +2EV with AV - it's probably safe.

rob

[hcostache]

When taking flats with my DSLR's I normally use AV setting.
This way the Camera evalutes the correct expo for flat frames.

Trust but verify. The metering system in most cameras is calibrated for a middle gray level of about 18%. You might end up with kind of underexposed flats.

Regards,
Horia

[lucchett]

Thank you.
Rob, I meant that the pixel of the Bayern matrix could respond differently, depending on the light source used. This is not a real issue using a good panel, I think.

Nice trick the pink shirt :-)

Small under exposure shouldn't be an issue if we increase the number of flats.
But I will try to validate the EV method numerically.

Thank you all, this has been very informative for me.
Andrea