How many flats, darks, bias?

thegableguy (Chris) · #1 08-11-2016, 11:47 AM

Simple question. How many darks, flats and bias shots should one use when stacking in DSS? I got about 50 good 3-min subs last night, got 30 darks before the camera battery died, and will be taking new flats and bias today (I don't have them for the ISO I used last night).

Is there an ideal ratio, or same number as lights, or just as many as you can get?

Atmos (Colin) · #2 08-11-2016, 12:10 PM

I usually do some 25 bias simply because it takes a few seconds to do so. If possible, all calibration frames should be taken at roughly the same temperature.

LightningNZ (Cam) · #3 08-11-2016, 12:22 PM

I take 100 bias frames because it's such a small issue to take so many. I usually go for 50 flats and 50 darks too. I think most people consider 20 an absolute minimum or else you're just adding noise.

RB (Andrew) · #4 08-11-2016, 12:44 PM

If using a DSLR you don't need bias frames.
Do same amount of darks as lights if possible, at the same ambient temp.
Do an odd number of flats, not an even number and at least 25 frames.
Even better if you can do more.

thegableguy (Chris) · #5 08-11-2016, 01:32 PM

Okay cool. I'll aim for 50 lights, 50 darks (which were taken at same temp as lights), 49 flats, won't bother with bias.

Is the ignoring bias thing because of what I've heard before; DSS can glean the bias data from the other shots?

Anyway I'll start stacking it all together tonight. With a bit of luck it should turn out to be my best galaxy image yet.

Thanks folks

RB (Andrew) · #6 08-11-2016, 01:53 PM

Quote:

Originally Posted by thegableguy

Is the ignoring bias thing because of what I've heard before; DSS can glean the bias data from the other shots?

Yes. Included in the Darks, unless your darks are different length to your lights.

sharptrack2 (Kevin) · #7 08-11-2016, 02:28 PM

Please clarify that in your experiences, the flats do not have to have the same exposure length as the lights and darks, but should be at the same ISO and focus as the lights. It is all about establishing a baseline of the optical train and sensor "capacity" (please forgive for using the term capacity VERY loosely).

RB (Andrew) · #8 08-11-2016, 02:50 PM

For Flats, only the focus and orientation has to be the same.
Not ISO or length.

I put the DSLR in P (Program Mode).
Set ISO to 100.
Take an exposure and check the histogram is 1/3 rd of the way along from the left.
Take a lot and an odd number of flats.
Also I move the camera/scope slightly around the Flats Screen/Light Panel every few shots to avoid any hot spots.

I use P mode and stick to 100 ISO (giving lower noise, better signal) because it initially gives me a histogram close to what I'm after, 1/3 to 1/2 way across.
If it's not close to that I can switch to manual and tweak the setting I got from the P settings to get the ideal histogram.

I use an odd number because I 'median combine' the flats and the 'math algorithm' used in 'median combine' works better with an odd number of frames.

RB

thegableguy (Chris) · #9 08-11-2016, 08:03 PM

Thanks folks. Stacking now. Unfortunately my data from last night is less great than I thought, so only about 23 usable 3-min subs from nearly four hours. Focus moved at some point. Oh well...!

I'm starting to realise that there's vastly more to stacking and processing than I initially suspected (much like every other aspect of this hobby), and I'm going to have to learn a lot more about how it works, what the myriad different options are, what benefits they each have and how best to wrench the best possible results from a cheap unmodified Nikon. I've been a photographer for years but nothing could have prepared me for this astonishingly convoluted, technically challenging process!

Cheers all, thanks for the help - results will be posted in the Beginners' section later tonight if they're not too embarrassing.

Atmos (Colin) · #10 08-11-2016, 08:23 PM

I read a paper a few years ago that suggested the optimal number of subs for flats what was whatever was required to hit 1,000,000 electrons. For my QHY22 that magic number would be ~125 subs per channel.

As doing the above, 125 subs in Ha, OIII & SII would take a bit over two hours... I usually do 25

sharptrack2 (Kevin) · #11 08-11-2016, 08:50 PM

Thanks for the replies,

I have got to make it to an astrocamp.... SOON! I'm having the hardest time with processing and need to see someone work through a set and be able to ask questions. YouTube videos just aren't cutting it for me.

But disregarding that, would it generally be acceptable to have at least half as many flats and darks, as lights? Or is that a fallacy, given that a night of DSO imaging might only generate 10-20 subs, and half is not enough to average out any anomalies.

Quote:

As doing the above, 125 subs in Ha, OIII & SII would take a bit over two hours... I usually do 25

So what would a Canon 5D MKII work out to be?

mikeyjames (Mick) · #12 08-11-2016, 09:17 PM

Quote:

Originally Posted by RB

For Flats, only the focus and orientation has to be the same.
Not ISO or length.

I put the DSLR in P (Program Mode).
Set ISO to 100.
Take an exposure and check the histogram is 1/3 rd of the way along from the left.
RB

Hi Chris & Rob,
Chris, I hope you don't mind me hijacking your thread with a question for Rob.

Rob, I hope you don't mind a real newb question. I used the white t-shirt method to take some flats the other night. The basically looked like washed out white pictures. Is this what they are supposed to look like? Also, if so, why is focus and orientation so important?

Thanks
Mick

thegableguy (Chris) · #13 08-11-2016, 09:41 PM

Quote:

Originally Posted by mikeyjames

Hi Chris & Rob,
Chris, I hope you don't mind me hijacking your thread with a question for Rob.

Rob, I hope you don't mind a real newb question. I used the white t-shirt method to take some flats the other night. The basically looked like washed out white pictures. Is this what they are supposed to look like? Also, if so, why is focus and orientation so important?

Thanks
Mick

Yes, they're supposed to look like white washed-out pictures, with the peak of the histogram just left of halfway along.

Focus is important because it'll affect the amount of vignetting. (Not a lot, but enough.) Orientation will affect the direction of the vignetting. The vignetting is largely what flats are used to eliminate.

Having said that, I get vastly better results if I don't use flats. I've tried using them repeatedly, following all the instructions I've been told or have found, and they invariably end up ruining my stacks. No idea why but I've vowed that tonight's effort was the last time I tried them until I know better why they screw everything up every time!

Atmos (Colin) · #14 08-11-2016, 09:45 PM

Quote:

Originally Posted by sharptrack2

Thanks for the replies,

I have got to make it to an astrocamp.... SOON! I'm having the hardest time with processing and need to see someone work through a set and be able to ask questions. YouTube videos just aren't cutting it for me.

But disregarding that, would it generally be acceptable to have at least half as many flats and darks, as lights? Or is that a fallacy, given that a night of DSO imaging might only generate 10-20 subs, and half is not enough to average out any anomalies.

So what would a Canon 5D MKII work out to be?

Depending on your ISO probably about the same. In many respects hitting that 1,000,000e- count is a bit of overkill. The paper I was reading was in regards to creating the perfect flat field with professional observatories in mind where they are wanting to get millimag precision. At this level they're not just wanting to correct for vignetting but also pixel to pixel sensitivity variation.

mikeyjames (Mick) · #15 08-11-2016, 09:56 PM

Quote:

Originally Posted by thegableguy

Yes, they're supposed to look like white washed-out pictures, with the peak of the histogram just left of halfway along.

Focus is important because it'll affect the amount of vignetting. (Not a lot, but enough.) Orientation will affect the direction of the vignetting. The vignetting is largely what flats are used to eliminate.

Having said that, I get vastly better results if I don't use flats. I've tried using them repeatedly, following all the instructions I've been told or have found, and they invariably end up ruining my stacks. No idea why but I've vowed that tonight's effort was the last time I tried them until I know better why they screw everything up every time!

Thanks Chris, I may five it a go without the flats and see what it looks like.

BTW, I meant no insult by directing my question to Rob. Still learning the lay of the land around here.

thegableguy (Chris) · #16 08-11-2016, 10:19 PM

Quote:

Originally Posted by mikeyjames

Thanks Chris, I may five it a go without the flats and see what it looks like.

BTW, I meant no insult by directing my question to Rob. Still learning the lay of the land around here.

No insult taken!! It's just rare that I know the answer to a question, couldn't resist answering. This is the newbie corner after all. Most of us are acutely aware we're playing in the shallow end of the pool; for the most part, even at the more advanced end there's a delightful lack of egos in IIS forums.

mikeyjames (Mick) · #17 08-11-2016, 11:08 PM

Quote:

Originally Posted by thegableguy

No insult taken!! It's just rare that I know the answer to a question, couldn't resist answering. This is the newbie corner after all. Most of us are acutely aware we're playing in the shallow end of the pool; for the most part, even at the more advanced end there's a delightful lack of egos in IIS forums.

Thanks Chris. I've been around for nearly three months - asking lots of questions pre-telescope and then over the past two weeks since picking it up. I have made over 100 posts and I am yet to receive a single smart ar%$ reply or any hint of condescension. Everyone has been patient and done their best to help.

This is really a unique forum.