4

I have read many places that for the best calibrated astrophotographs, you should be taking bias frames, dark frames, and flat frames.

But on a forum I read the following statement

You usually don't need to. DSLRs with CMOS sensors (as opposed to CCD) automatically take and subtract a bias frame every time an exposure is made. If they didn't, the noise would be horrendous.

Almost all current models of DSLR use CMOS sensors; some older Nikons used CCD.

Is this true? Does this mean that all the astrophotography literature I have been reading is either outdated, or just wrong?

Is there no point taking bias shots to subtract from my flat frames and light shots?

EDIT: I should clarify that a bias frame is not a dark frame. The quoted person went on to indicate that all CMOS sensors take a bias frame afterwards, and SOME have the option for take a dark frame, which is taken with the same shutterspeed as the light frame (bias frame is always taken with fastest shutter).

Scorb
  • 1,048
  • 11
  • 22
  • I think the forum poster should have said, "SOME cameras have the OPTION of taking a dark frame automatically." This isn't a default option, and where it is, it isn't on be default either. So, your literature is correct. – OnBreak. Nov 29 '17 at 23:00
  • Actually further discussions went on to make the distinction that bias frame is NOT a dark frame. And this optional dark frame is in addition to the light and the bias. I will update main question. – Scorb Nov 29 '17 at 23:01
  • There are probably some cameras somewhere that do not measure bias from the sensor with each exposed frame. So until you specify a specific camera the question probably can not be definitively answered. – Michael C Nov 30 '17 at 00:24
  • The question does not pertain to a specific model. The question is reduced to "based on the statement that all CMOS sensors take a bias frame, do I still need to take bias frames for astrophotography on CMOS sensors"? The question is totally valid and independant from a specific model. – Scorb Nov 30 '17 at 00:58
  • AFAIK CMOS sensors have a occluded strip at the side which gets read out with every exposure to obtain the readout noise level, which then gets subtracted from the exposure. i believe this is what the bias frame of CCD's is about, but i haven't found a reference for that now, so not posting as answer. – ths Nov 30 '17 at 10:06
  • 1
    It'd be good form to include a link to the forum where the quoted text comes from. – Caleb Dec 03 '17 at 07:23
  • I have been told including links is bad form, as they can go dead in the future. – Scorb Dec 04 '17 at 14:27
  • You can find some info here: http://www.clarkvision.com/articles/astrophotography.image.processing/ – FarO Dec 04 '17 at 16:24

1 Answers1

2

I've had a couple of tries with Siril and found it very complex and probably designed for the astronomy enthusiast who takes photograph's rather than the photographer who takes the occasional astro picture. That said, results are possible.

https://free-astro.org/index.php/Siril:Tutorial_preprocessing states as follows:

WARNING: Remember that dark frames are always composed from real dark signal and bias signal. If you don't apply dark optimization, you can leave the bias signal and your masterDark will be in fact . In consequence subtracting this master to the light frames will remove both signals. However, applying dark optimization makes things different by multiplying masterDark by a coefficient factor not equal to 1. In this case, you must subtract masterBias from each dark frame.

http://www.rawastrodata.com/pages/typesofimages.html states it a bit more clearly (it also describes the different types of images used quite nicely):

Note: Dark frames also contain the bias signal which is why I can get away without subtracting bias frames. I have read that subtracting a bias from the darks can help improve DeepSkyStacker’s dark optimization feature.

It seems that if you are using the Dark Frames made necessary by long exposures, you don't need Bias Frames as the information is already in the Dark Frame.

dmkonlinux
  • 1,003
  • 1
  • 8
  • 24