Do separate flats for narrowband filters make that much difference?

[irtuk]

Clue in the title I am running off a new set of flats using the flats wizard on NINA and its got options to do flats for each individual filter. Is this worth it?

Edited October 14, 2021 by irtuk

[Skipper Billy]

Yes - there will be different dust particles on each filter. Having said that I am lucky - I make a calibration library at the start of the season and use them for all of that season. If I was setting up and breaking down each time I might think differently!!

 

[vlaiv]

Of course.

I've seen people reuse flats from different filters and I simply can't understand that.

Not only do filter can and have different distribution of dust particles on them - pixels can have different relative sensitivity in different parts of spectrum.

We assume that pixels have same QE curve - but it could be different, and sometimes is - due to manufacturing defects or just nature of material used (and angle of light hitting them).

Look at this:

Three different histograms for three different filters. Or visually:

Yes, dust particles are of different color and that is fine because they are in one filter and not others, but why is background with gradient? How come that pixels on the left part of the image have stronger red response than pixel on the right part of the image, while blue/green response is opposite?

 

[teoria_del_big_bang]

I would say yes for sure regarding taking separate flats for each filter, for the good reasons in the above replies.

Regarding how often I would take them then personally I would say if you break your setup down after a session then should be just before or just after every session, if you do not break it down then you could use a library for some amount of time but dust will move over time so still good to keep updating the library.

Best scenario if you can is to take flats every session for any filters used in that session, but not everybody has the luxury of a flip flat, that makes things so much easier and quicker, or even a flat panel, so that may not be a viable option.

Also check your flats before using them to see if they look right, make a master of them for each filter and stretch them to see if you can see any obvious problems with them. Using bad flats will give you more problems than no flats at all. I like to look at a few stretched flats just as I have taken them, or as soon as I can, in case they are not right then I have time to adjust something and take them again. If you wait until you come to use them in processing it may be too late to take any more. 

Steve

[Adam J]

They do if you have dust on your filters or they centre a little differently when the wheel rotates to them. 

[irtuk]

Sounds like an overwhelming vote for separate then. Its not much of a chore with the NINA wizard.

Thanks all for the input.

[irtuk]

ok so yet another idiot question I think I already know the answer to but here we go...

I am using a tablet flat frame app with the HA filter and the flat looks darker than I expected at first, however this is because my tablet produces very little Hydrogen Alpha signal.

Right?

[vlaiv]

4 minutes ago, irtuk said:

ok so yet another idiot question I think I already know the answer to but here we go...

I am using a tablet flat frame app with the HA filter and the flat looks darker than I expected at first, however this is because my tablet produces very little Hydrogen Alpha signal.

Right?

Yes, some of it is down to strength of Ha in source, and some of it is down to camera sensitivity in Ha, but you can always compensate with exposure length.

Double the exposure length - double the signal.

NB flats often need much longer exposures than broad band filters - since NB filters only pass about 1/15 - 1/30 of light compared to broad band filters (for example red passes all between about 600 and 700nm - so 100nm range, while Ha passes maybe 3.5nm to 7nm depending on make/model).

[irtuk]

Nina is managing it all. and yes its taking between 4 and 6 second exposures on the HA filter, balancing everything out. I am in a meeting

[irtuk]

It seems to be struggling with the S2 filter, needs a very long exposure.

[teoria_del_big_bang]

5 minutes ago, irtuk said:

It seems to be struggling with the S2 filter, needs a very long exposure.

Until I got a brighter flat panel I think SII exposures were in the region of 30 to 40 seconds

Steve

[irtuk]

I went out to have a look and identified that one my daughters apps on the tablet had possibly updated itself and opened, then shut the screen off, so it was struggling to take a flat of Disney Princess Hair Styles, which I can totally understand.

[Richard_]

Yeah there's quite a difference in exposure times with filters. With just an IR/UV cut filter in my OSC camera, a tablet flat requires around 200ms to hit the middle of the histogram whereas my dual bandpass filter (Ha and OIII) requires around 7s to hit the middle of the histogram using the same tablet light source. It sounds like you're in a similar exposure time for your Ha filter. 

If you're not doing so already, make sure that any sort of night/red filter is switched off on your display so that you can correctly use a white light from your tablet display. 

[teoria_del_big_bang]

1 hour ago, irtuk said:

I went out to have a look and identified that one my daughters apps on the tablet had possibly updated itself and opened, then shut the screen off, so it was struggling to take a flat of Disney Princess Hair Styles, which I can totally understand.

That would have made interesting artefacts in the processing stage 🙂 

Steve

[rickwayne]

On 14/10/2021 at 06:51, irtuk said:

I went out to have a look and identified that one my daughters apps on the tablet had possibly updated itself and opened, then shut the screen off, so it was struggling to take a flat of Disney Princess Hair Styles, which I can totally understand.

This is quite possibly my favorite astro fail of all time -- and I speak as someone who has had smoke coming out of their mount. I salute you, sir!

[ollypenrice]

On 14/10/2021 at 10:25, vlaiv said:

Of course.

I've seen people reuse flats from different filters and I simply can't understand that.

My explanation is this: most of the time I find it makes no difference whatever. I fully accept that others might find big differences between flats in different filters but, quite simply, I don't find that. Maybe one image in twenty throws up a filter with its own dust bunny and, when it does, I make a flat for it. I know the theory and if I have a guest who wants to make 'flats per filter' we just do them, but when I'm imaging for myself I only do so if there's a need.

It will also depend on the kit. I find dust bunnies are not produced by dust on the filters but elsewhere, probably the chip window. And I also find that the RASA, at F2, doesn't seem to produce dust bunnies at all, probably because they are so far out of focus. 

Does exposure time matter? I'd have thought that what matters is that the flat in use be properly exposed. The linearity of modern cameras is what makes flats work in the first place, no?

Some people like to do things by the book, some are pragmatists who base their decision on seeing the results. I'm a pragmatist but I don't insist that this is the best way to be. It's just a welcome shortcut on a night when you have 24 hours of data to calibrate in the morning. (OK, that's a one-off record but we did once manage it with three scopes. )

Olly

Edited October 16, 2021 by ollypenrice

[ollypenrice]

On 14/10/2021 at 10:25, vlaiv said:

Of course.

I've seen people reuse flats from different filters and I simply can't understand that.

Not only do filter can and have different distribution of dust particles on them - pixels can have different relative sensitivity in different parts of spectrum.

We assume that pixels have same QE curve - but it could be different, and sometimes is - due to manufacturing defects or just nature of material used (and angle of light hitting them).

Look at this:

Three different histograms for three different filters. Or visually:

Yes, dust particles are of different color and that is fine because they are in one filter and not others, but why is background with gradient? How come that pixels on the left part of the image have stronger red response than pixel on the right part of the image, while blue/green response is opposite?

 

Sorry, me again! You ask, ' ...why is background with gradient? How come that pixels on the left part of the image have stronger red response than pixel on the right part of the image, while blue/green response is opposite?'

That's a very good question but I don't think it has anything to do with flats. At least, it doesn't in my case because I get this kind of gradient regularly on all sorts of cameras but, most obviously, on the two OSC cameras I've used.  It's extremely pronounced on our ASI 2600/RASA data and, after the flip, it follows the camera and not the sky. However, we have yet to shoot any flats in the RASA so it can't be a product of flats. Nor did flats remove similar gradients from my OSC CCD. I don't expect them to remove the RASA gradients either but perhaps they will.  I'll post on this when I've done so because it's an interresting problem.

Olly

[teoria_del_big_bang]

1 hour ago, ollypenrice said:

It will also depend on the kit. I find dust bunnies are not produced by dust on the filters but elsewhere, probably the chip window. 

I would agree.
Having looked closely at many of my flats, as you should I think because they can go wrong and then make images worse than with no flats, if I overlay all my flats from all filters just about all blobs and donuts align perfectly, and I know there is some dust on my filters. There were a very few odd differences but very minor so probably something to do with the filters but probably most would not be seen in final image and probably would be masked a bit with DBE.

I have even dismantled the FW and cleaned dust off my filters (or at least moved it around 🙂 ) and the flats still look pretty much identical.
I then cleaned the chip window of dust (best I could - dust just never seems to totally go) and then the flats did change.

So I came to the same conclusion that despite there definitely being dust on the filters they do not  show up so much due to being so far out of focus.

Having said that, I still take flats for all filters used, but it doesn't take long and is just part of my sequence in EKOS at end of session with a flip flat so it is easy.

Steve

Edited October 16, 2021 by teoria_del_big_bang

[rnobleeddy]

3 hours ago, ollypenrice said:

My explanation is this: most of the time I find it makes no difference whatever. I fully accept that others might find big differences between flats in different filters but, quite simply, I don't find that. Maybe one image in twenty throws up a filter with its own dust bunny and, when it does, I make a flat for it. I know the theory and if I have a guest who wants to make 'flats per filter' we just do them, but when I'm imaging for myself I only do so if there's a need.

It will also depend on the kit. I find dust bunnies are not produced by dust on the filters but elsewhere, probably the chip window. And I also find that the RASA, at F2, doesn't seem to produce dust bunnies at all, probably because they are so far out of focus. 

Does exposure time matter? I'd have thought that what matters is that the flat in use be properly exposed. The linearity of modern cameras is what makes flats work in the first place, no?

Some people like to do things by the book, some are pragmatists who base their decision on seeing the results. I'm a pragmatist but I don't insist that this is the best way to be. It's just a welcome shortcut on a night when you have 24 hours of data to calibrate in the morning. (OK, that's a one-off record but we did once manage it with three scopes. )

Olly

Out of curiosity, is the consensus that expoure time/exposure does matter a lot?

I've assumed that as long as I have them somewhere in the middle of the histogram then all will be fine - and that I'd rather steer well away from overexposure as that's guaranteed to ruin an image, so I err on the lower side.

I generally find results are solid, although on occasion I can end up with color gradients on LRGB images and I haven't really worked out if that's just present on the image (moon or LP from the local city), or is a problem with the flats.

 

Edited October 16, 2021 by rnobleeddy

[ollypenrice]

41 minutes ago, rnobleeddy said:

Out of curiosity, is the consensus that expoure time/exposure does matter a lot?

I've assumed that as long as I have them somewhere in the middle of the histogram then all will be fine - and that I'd rather steer well away from overexposure as that's guaranteed to ruin an image, so I err on the lower side.

I generally find results are solid, although on occasion I can end up with color gradients on LRGB images and I haven't really worked out if that's just present on the image (moon or LP from the local city), or is a problem with the flats.

 

Well, my skies are sometimes SQM22 and I get one image a year, at most, which has no colour gradient worth correcting. They nearly all do. Who knows where they come from, but since I get them on an OSC and they follow the camera after the flip, I'm going to say that the camera is my prime suspect. It can't be the sky or local light sources if it doesn't follow the flip. ABE or DBE offer a quick and painless fix nearly all the time.

I generally go for 1/3 of full capacity for flats but when I've been higher, up to 2/3, I haven't seen any problems.

The single flat I use when being lazy is the luminance flat. Firstly it nearly always agrees with the others and, secondly, the image will be illuminated by the luminance by definition, so that, too, flattens the other channels to some extent.

Olly