2600MC Pro and Flats - Better without?

[smr]

Ok so last night was quite a decent night for imaging, no Moon and the HH is rising in alt. 

I've just done a calibration test with, and without flats.

This image was taken with 3hours exposure time, the first one is just a stack of the Lights, no other calibration frames. It looks pretty good to me, all I have done is a few level adjustments to do some linear stretching.

Light Frames only:

 

This second image is calibrated this time with Flats and Dark Flats which I have taken this morning (everything in the imaging train stayed the same between last night and this morning). I used the T-Shirt method (doubled over and held taut with an elastic band)

The Target ADU in APT is around 32,000. It does reach this but with a very short exposure (0.12813s) is what it has do to reach the target ADU. I have tried an LED panel and it's the same with that as well, always an extremely short exposure time), this is also calibrated with Dark Flats. As you can see the flats aren't calibrating the image properly at all. Any ideas as to why? I'm wondering if I even need bother take flats - I can't see any obvious gradient problems or vignette in the first image or dust motes.

The same processing has been done to both, a few linear level adjustments and one gradient exterminator run on each (medium and low) balanced background, inverted with a lasso over the middle of the image.

 

Light Frames, Flats and Dark Flats

 

[vlaiv]

5 minutes ago, smr said:

As you can see the flats aren't calibrating the image properly at all. Any ideas as to why?

You did not include regular darks in your calibration. This is important if you want proper flat calibration.

This way - you are "flattening" - both light signal as well as dark signal that is not removed since dark calibration was not performed. However - dark signal should not be flattened - as it is not subject to optical train and does not vignette or have dust on it (it is not related to light entering the telescope).

[smr]

1 minute ago, vlaiv said:

You did not include regular darks in your calibration. This is important if you want proper flat calibration.

This way - you are "flattening" - both light signal as well as dark signal that is not removed since dark calibration was not performed. However - dark signal should not be flattened - as it is not subject to optical train and does not vignette or have dust on it (it is not related to light entering the telescope).

Ok, I'll try with Darks on both stacks and see what the result is.

[smr]

Ok, looks a lot better with the Dark frame. I didn't realise that Dark frames were so necessary with flat frames. I'm sure I've read a few things where some people say they don't bother with dark frames and just dither etc.?

 

[The Lazy Astronomer]

33 minutes ago, smr said:

Ok, looks a lot better with the Dark frame. I didn't realise that Dark frames were so necessary with flat frames. I'm sure I've read a few things where some people say they don't bother with dark frames and just dither etc.?

 

That advice is for DSLRs or other cameras where you don't have control over the sensor temperature. If you can't temperature matched darks, they may end up introducing more problems than they solve. Obviously you have set point cooling, so the full suite of calibration frames should net you the best result. 

Edit: but do still dither!

Edited January 28, 2022 by The Lazy Astronomer

[smr]

Ok thanks guys, and wrt darks, how often do they need renewing?

The ones I am using are from Feb last year.

[vlaiv]

I usually do it twice a year, there is summer and winter set of darks that I do at different temperature (I often can't cool camera down to -20°C in summer).

Do it when you change settings - like driver update or offset/gain.

[The Lazy Astronomer]

32 minutes ago, smr said:

Ok thanks guys, and wrt darks, how often do they need renewing?

The ones I am using are from Feb last year.

I don't really know what the generally accepted convention is, mine are between 6 - 12 months old currently, and I'll probably just keep using them until I encounter some kind of calibration issue. I suppose if it was cloudy for a prolonged period and I was bored one evening, I may take some new ones.

[Stuart1971]

Flats are not good with CMOS cameras if under around 1 second, best to do between 3 and 5 second flats, so you need a dimmer light source…👍🏼

[vlaiv]

19 minutes ago, Stuart1971 said:

Flats are not good with CMOS cameras if under around 1 second, best to do between 3 and 5 second flats, so you need a dimmer light source…👍🏼

I use milliseconds for my flats with ASI1600. They work just fine.

[ONIKKINEN]

1 hour ago, Stuart1971 said:

Flats are not good with CMOS cameras if under around 1 second, best to do between 3 and 5 second flats, so you need a dimmer light source…👍🏼

I too found that 50ms flats calibrated my lights well with my Rising cam OSC. They were so fast because i was experimenting with different gain settings, i dont usually use so fast exposures for flats. But usually mine are somewhere around 130ms and also work well.

[Stuart1971]

7 minutes ago, ONIKKINEN said:

I too found that 50ms flats calibrated my lights well with my Rising cam OSC. They were so fast because i was experimenting with different gain settings, i dont usually use so fast exposures for flats. But usually mine are somewhere around 130ms and also work well.

Oh well I read that CMOS don’t work as well with fast flats, but maybe it’s just the QHY268c then….my bad…👍🏼

[Adam J]

5 hours ago, Stuart1971 said:

Oh well I read that CMOS don’t work as well with fast flats, but maybe it’s just the QHY268c then….my bad…👍🏼

Depends on the chip, some perform internal calibration for short exposures and have difference noise characteristics in short exposures. Or just don't take consistent length exposures below 2 seconds. 

Edited January 28, 2022 by Adam J

[teoria_del_big_bang]

1 hour ago, vlaiv said:

I use milliseconds for my flats with ASI1600. They work just fine.

I had issues with short flats on my ASI1600 and also was advised on SGL to use exposures over 1 second. Now without looking back at my notes I am not sure how short they were but I would thing still in the 100mS + I would have thought.
The issue I got was intermittent banding and after I thought about it afterwards the issue was probably more to do with my old flat panel flickering than the camera.

But, out of habit I have sort of kept up with the flats at 3 or 4 seconds, as much as anything because with my new flat panel I can get good 3 or 4 second flats for all my filters, which with much shorter exposures would not suit my NB filters, meaning I can use the same dark for calibration without needing PI to use scaling.
They seem to work okay, just take a bit longer to take.

Is there any advantage then (apart from time) to taking fast flats as opposed to 3 or 4 seconds ?

Steve

Edited January 28, 2022 by teoria_del_big_bang

[Adam1234]

I've found short flats less than 1s don't work so well with my asi1600mm, they tend to overcorrect (or undercorrect, can't remember which)

I get longer flats by increasing the number of layers between the light source and the scope if I can't dim the light source enough. I use a similar method to the t-shirt method, but I use sheets of white paper instead, reduces the risk of creases and loose fibres falling onto the lens.

My flats are generally around 5 - 7s. The advantage of longer flats is it averages out any flickering of the light source that you can't see with the eye but that can be captured using short exposures

[Clarkey]

6 hours ago, smr said:

wrt darks, how often do they need renewing?

Mine are at least 18 months old and still seem to calibrate pretty well. Probably should renew them though.

To be honest I have never seen much difference between long and dark flats on my ASI1600.

 

6 hours ago, vlaiv said:

I usually do it twice a year, there is summer and winter set of darks that I do at different temperature (I often can't cool camera down to -20°C in summer).

Summer, winter - same temperature here. Just more rain in the summer😁

[vlaiv]

4 hours ago, teoria_del_big_bang said:

Is there any advantage then (apart from time) to taking fast flats as opposed to 3 or 4 seconds ?

No except for time dedicated to flats. I take a lot of them - like 256 (divisible with 2 actually power of two - I had couple of beers - that is divisible by 2 ), and that takes about 10 or so minutes mostly because SGP takes a bit to download sub for some reason - although camera is capable of 10+ FPS on full ROI with 16 bit readout.

Flats are like any other exposure - if you have dim source you need long exposure - if not - you can use short exposure.

My flat box is good and does not produce flickering. I actually did have issue once - but it had to do with power connector - it was a bit loose and did not form proper connection to power source and I had flickering because of that, but once I solved the power connector issue - I don't have issues anymore.

4 hours ago, Adam1234 said:

I've found short flats less than 1s don't work so well with my asi1600mm, they tend to overcorrect (or undercorrect, can't remember which)

It could be down to calibration. If you use bias instead of flat darks - then this could be issue. It seems that some cameras, ASI1600 included (at least with old driver version) has two different regimes - up to 1s and above 1s. In first, timing is kept by camera itself, while in second, it is up to driver (computer) to do exposure timing.

Bias/offset values differ between the two. For this reason, bias is not usable for long exposure as average ADU is higher than exposure of say 30s or 1 minute (which should not happen as darks have dark current signal which bias does not have).

In any case - since bias is "Strange" - calibrating with bias can end up in over / under correction.

Proper way to do it is to do dark flats - which is basically same flat exposure with scope covered instead of using flat panel.

Edited January 28, 2022 by vlaiv

[The Lazy Astronomer]

6 hours ago, Stuart1971 said:

Flats are not good with CMOS cameras if under around 1 second, best to do between 3 and 5 second flats, so you need a dimmer light source…👍🏼

 I've read that the 294 cameras don't do so well with flats under 1s, so l stick to flats above 3s for my 294MM, but thinking about it, ultra short flats are about the only thing I haven't tried to try and fix my Oiii calibration woes, so maybe I'll give it go tomorrow. 

[Chefgage]

28 minutes ago, vlaiv said:

No except for time dedicated to flats. I take a lot of them - like 256 (divisible with 2 actually power of two - I had couple of beers - that is divisible by 2 ), and that takes about 10 or so minutes mostly because SGP takes a bit to download sub for some reason - although camera is capable of 10+ FPS on full ROI with 16 bit readout.

Flats are like any other exposure - if you have dim source you need long exposure - if not - you can use short exposure.

My flat box is good and does not produce flickering. I actually did have issue once - but it had to do with power connector - it was a bit loose and did not form proper connection to power source and I had flickering because of that, but once I solved the power connector issue - I don't have issues anymore.

It could be down to calibration. If you use bias instead of flat darks - then this could be issue. It seems that some cameras, ASI1600 included (at least with old driver version) has two different regimes - up to 1s and above 1s. In first, timing is kept by camera itself, while in second, it is up to driver (computer) to do exposure timing.

Bias/offset values differ between the two. For this reason, bias is not usable for long exposure as average ADU is higher than exposure of say 30s or 1 minute (which should not happen as darks have dark current signal which bias does not have).

In any case - since bias is "Strange" - calibrating with bias can end up in over / under correction.

Proper way to do it is to do dark flats - which is basically same flat exposure with scope covered instead of using flat panel.

What is your reasoning for taking that many flats (256)? Interested as I only take 20 usually.

[vlaiv]

2 minutes ago, Chefgage said:

What is your reasoning for taking that many flats (256)? Interested as I only take 20 usually.

I guess it some sort of OCD - binary type. I'm computer programmer and to me, it makes sense to take power of 2 number of subs because it can be easily divided with integer math (binary shifting).

Real reason would be reduction of noise. ASI1600 is 12bit camera and I use unity gain. That makes single flat kind of noisy with only 4000 ADU levels and say 75% of histogram I'm at 3000e per peak value. Someone using 14 or 16 bit ADC (like in CCDs with 30000-40000 FWC) will have single exposure with x10 signal, so at the start my flat exposure has x3-x4 worse SNR than someone using CCD.

In order to get to same level - I would need x9-x16 more flats in my stack. If person with CCD is using 20-30 flat subs, I'd need to use what, 200+ to compensate - see, it sort of checks out

This is of course exaggeration - since even single exposure of 3000e worth of signal will have SNR of 50+ and since we are dividing with flats - resulting noise polluting the image is very small - but like I said, OCD thing - it is very short, I can do it, it divides nicely by shifting (power of two) - then why not

[smr]

This also explains why my total stack on the Horsehead Nebula came out the same way as the 2nd image in my OP - the image with flats but no darks applied.

In my longer stack of the HH, I've got a blend of 3 and 5 minute exposures, and only a master dark for the 3 minute exposures. I haven't got any darks for the 5 minute subs. So I'm just taking some 20 x 300 second dark frames now, I'll integrate them into DSS tomorrow and see if I can get a nice stacked image. 

[smr]

With regards to taking flats, is it OK to take them indoors against a white wall? It's just that I haven't got a flat panel and sometimes the next morning it might be patchy cloud which I understand isn't ideal for pointing the scope at to take flats (with the t shirt covering the scope)

Edited January 28, 2022 by smr

[vlaiv]

30 minutes ago, smr said:

With regards to taking flats, is it OK to take them indoors against a white wall? It's just that I haven't got a flat panel and sometimes the next morning it might be patchy cloud which I understand isn't ideal for pointing the scope at to take flats (with the t shirt covering the scope)

I think that will be fine, but best thing to do is to try it.

[The Lazy Astronomer]

28 minutes ago, smr said:

With regards to taking flats, is it OK to take them indoors against a white wall? It's just that I haven't got a flat panel and sometimes the next morning it might be patchy cloud which I understand isn't ideal for pointing the scope at to take flats (with the t shirt covering the scope)

Should be, provided the lighting is even. In fact, daylight wall flats is the only way I've managed so far to get passable (but not quite perfect) Oiii flats. 

[Chefgage]

1 hour ago, vlaiv said:

I guess it some sort of OCD - binary type. I'm computer programmer and to me, it makes sense to take power of 2 number of subs because it can be easily divided with integer math (binary shifting).

Real reason would be reduction of noise. ASI1600 is 12bit camera and I use unity gain. That makes single flat kind of noisy with only 4000 ADU levels and say 75% of histogram I'm at 3000e per peak value. Someone using 14 or 16 bit ADC (like in CCDs with 30000-40000 FWC) will have single exposure with x10 signal, so at the start my flat exposure has x3-x4 worse SNR than someone using CCD.

In order to get to same level - I would need x9-x16 more flats in my stack. If person with CCD is using 20-30 flat subs, I'd need to use what, 200+ to compensate - see, it sort of checks out

This is of course exaggeration - since even single exposure of 3000e worth of signal will have SNR of 50+ and since we are dividing with flats - resulting noise polluting the image is very small - but like I said, OCD thing - it is very short, I can do it, it divides nicely by shifting (power of two) - then why not

Ah I see. Thanks for explaining that.