25 replies to this topic

[Whereisclearsky]

Hi,

 

I usually take around 30 flats, I've found some thread on CN where someone says they take 5 flats if using dual band filter.

I'm using L-Extreme.

 

I wonder what using a filter has to do with anything?

 

Thanks

[Daniel Dance]

I usually take number of flats based on squares of integers.  Like 16 or 25.  Sometimes i don't use any.

 

But don't get caught up in the whole "how many" is best.

 

Try 16, Try 25, Try 30.   Use your eyes and see what looks best in your final image.   After all, that is all that matters - what your final image looks like.

[idclimber]

I personally take 20 flats with all my filters. I do not recall reading anything about a dual band filter needing less. It makes no senses to me why it would. 20 to 30 is pretty common around here.

[bobzeq25]

Hi,
 
I usually take around 30 flats, I've found some thread on CN where someone says they take 5 flats if using dual band filter.
I'm using L-Extreme.
 
I wonder what using a filter has to do with anything?
 
Thanks

People say a lot of strange things about DSO AP. Ignore the silly stuff.

I take 30 also. I've never seen anyone serious seriously recommend 5.

I'd just ignore it. It's a waste of time to even consider it.

Edited by bobzeq25, 05 December 2024 - 07:44 PM.

[daveco2]

I've only ever taken 10 flats.  Why would you need more?  To reduce noise?  In a flat?

[Brian Carter]

30 is probably just fine.  I personally take 50 of all my calibration frames, darks, flats, bias, whatever.  Doesn't matter the scope, camera, or filter.  My images are calibrated nicely, so I haven't seen a reason to change anything.  I do recognize that it is probably a bit overkill.

 

More is always going to be better, but you do hit a point of diminishing returns.  30 is probably around that point where you aren't adding a huge amount of value for a few more subs.  Stick with it if it is working for you.

[idclimber]

I've only ever taken 10 flats.  Why would you need more?  To reduce noise?  In a flat?

Every frame (light, dark, bias or flat) has two components signal and noise. In the case of flats the signal we want is the differences in pixel response to a light source. Some of this is pixel dependent and some is based on the optics. The rest in an individual exposure is noise. 

 

We primary reduce the random noise by using more frames or longer total integration times. In other words more subs. 30 frames being better than 20. The math is simple to double SNR you need about four times the data. 

 

The other thing is that 20 frames is about where better rejection algorithms start working best. 

[Whereisclearsky]

People say a lot of strange things about DSO AP. Ignore the silly stuff.

It did seem silly to me, just wanted to make sure, lol.

[PIEJr]

It depends.

I see you, too, have an ASI2600MC Pro.

I began doing Bias, Flats, Darks to stack with my Lights.

I use an Antila Quad LP filter alone because of the light pollution around me, for full disclosure.

I have my libraries of Bias, Flats and Darks. But don't use them. Don't need to.

 

If I see an airplane track, I just delete those before stacking.

But I just use ASI Studio to look at and stack my files with.

And for a while now, I just take my lights and stack them. They get aligned and stacked and are ready for light finish processing.

 

Try it. You don't need to goober up the 2600MC's images.

They are pretty dang good.

[Whereisclearsky]

It depends.

I see you, too, have an ASI2600MC Pro.

I began doing Bias, Flats, Darks to stack with my Lights.

I use an Antila Quad LP filter alone because of the light pollution around me, for full disclosure.

I have my libraries of Bias, Flats and Darks. But don't use them. Don't need to.

 

If I see an airplane track, I just delete those before stacking.

But I just use ASI Studio to look at and stack my files with.

And for a while now, I just take my lights and stack them. They get aligned and stacked and are ready for light finish processing.

 

Try it. You don't need to goober up the 2600MC's images.

They are pretty dang good.

I've a lot of dust spots and vignette with Redcat 51 and this filter, so it's not an option.

Also it doesn't take a lot of time so why not improve the image? imo

[PIEJr]

I've a lot of dust spots and vignette with Redcat 51 and this filter, so it's not an option.

Also it doesn't take a lot of time so why not improve the image? imo

Tally Ho!

My point was, "You don't NEED flats".

Or, I've discovered I can do without them.

 

But by all means do it your way.

[idclimber]

Tally Ho!

My point was, "You don't NEED flats".

Or, I've discovered I can do without them.

 

But by all means do it your way.

You do not "need" any calibration frames. You can also not brush your teeth, never change the oil on your car, or dozens of other things.

 

The only question is if the addition of Darks, Bias and Flats improves our Lights. IMHO it is without question that it does. You may not find the improvement worth the time, but most of us do. 

[Kevin_A]

In my opinion I would always use flats, as all optics have light falloff at the edges that can cause vignetting on most setups. If the sensor is tiny it will be much less. With narrowband and dualband filters there is also some added colour cast of reds and blues that have abnormal shapes. I do 30-40 and do not use LED light boxes as natural light has always been my best neutral illumination for flats. That is just my preference. 

[Whereisclearsky]

In my opinion I would always use flats, as all optics have light falloff at the edges that can cause vignetting on most setups. If the sensor is tiny it will be much less. With narrowband and dualband filters there is also some added colour cast of reds and blues that have abnormal shapes. I do 30-40 and do not use LED light boxes as natural light has always been my best neutral illumination for flats. That is just my preference. 

I'm using a LED panel and it works fine.

[Kevin_A]

I'm using a LED panel and it works fine.

I sometimes use my Mirrorless cameras and they have focussing pixels (PDAF pixels) and using a LED light panel causes banding with those pixels on my Nikon Z7 flats… so I stopped using it as a light source. I still put the panel in front but without illumination. Supposedly, someone told me also that cheap LED panels are not the best light source due to the wavelengths emitted and that was actually confirmed. I remember now that electroluminescent panels are the best light source but quite expensive. However, I never had problems on my cheap light panel with a zwo cooled camera sensor but just with PDAF pixels, and I could not see much difference or any issues on non PDAF sensors in the past. I just prefer using the panel without using the LED lighting,  as I use fairly long flats and flat darks and I can achieve nice images without illumination. Once again… my preference. Your mileage may vary!

Edited by Kevin_A, 06 December 2024 - 01:31 PM.

[Whereisclearsky]

I sometimes use my Mirrorless cameras and they have focussing pixels (PDAF pixels) and using a LED light panel causes banding with those pixels on my Nikon Z7 flats… so I stopped using it as a light source. I still put the panel in front but without illumination. Supposedly, someone told me also that cheap LED panels are not the best light source due to the wavelengths emitted and that was actually confirmed. I remember now that electroluminescent panels are the best light source but quite expensive. However, I never had problems on my cheap light panel with a zwo cooled camera sensor but just with PDAF pixels, and I could not see much difference or any issues on non PDAF sensors in the past. I just prefer using the panel without using the LED lighting,  as I use fairly long flats and flat darks and I can achieve nice images without illumination. Once again… my preference. Your mileage may vary!

I have struggled with flats forever with my A6400 and Lumix S5.

After switching to 2600mc pro, I just set flats to auto in Asiair and it seems to do the job.

[Kevin_A]

I have struggled with flats forever with my A6400 and Lumix S5.

After switching to 2600mc pro, I just set flats to auto in Asiair and it seems to do the job.

I also use auto flats in asiair but like the results in the longer 2-7s range for flats for my four cooled cameras. The less light the longer my auto flats are. I usually do a preview test first to get my average near 32k at the flat integration time I want.

[bobzeq25]

I've only ever taken 10 flats.  Why would you need more?  To reduce noise?  In a flat?

Basically, yes. 10 may be "good enough" for some imagers. 30 is "good enough" for just about anyone.

5 is ridiculous.

[Whereisclearsky]

I also use auto flats in asiair but like the results in the longer 2-7s range for flats for my four cooled cameras. The less light the longer my auto flats are. I usually do a preview test first to get my average near 32k at the flat integration time I want.

My flats are short because I don't want to overexpose them.

My scope is F4.9 and the LED panel set to minimum brightness.

 

Here is one of my flats if anyone would like to check if it's exposed properly:

https://www.dropbox....t=qioo0733&dl=0

 

My flats are below 1s.

[idclimber]

Basically, yes. 10 may be "good enough" for some imagers. 30 is "good enough" for just about anyone.

5 is ridiculous.

To elaborate on Bobs point here. What is the disadvantage of 30 flats over 5 or even 20? 

 

The hard part of flats is getting a flat panel or the sky to work at the correct exposure. Once the settings are correct it only takes an extra minute to capture 20 to 30 subs. In other words 5 or 30 take about the same amount of time. The only reason I can think of to take less is hard drive space.

 

So how can we take 30 flats and have less long term storage? The answer is simple, save only the Master Flat with the original Lights for long term use. This only requires you subtract the Master Bias and Integrate. This is two manual steps. If you are using WBPP you could simply move those masters into the long term folder after you are done with the project and have a finished image. Then delete the original flats. 

 

I personally don't bother with the above and store original flats along with my lights. 

[TelescopeGreg]

The key to taking good Flats is to take them with exactly the same optical train as the Lights that they will be correcting.  So the filter in its exact position must be included, and by "exact" I mean without moving it.  Don't take things apart and put them back together.  This is a hobby of microns, and microns are very very very small.  You simply can't put things back together with enough precision to have things line up afterward.

 

My workflow is to take my Lights, then roll the scope back into the garage in the state it was when the imaging session ended.  Next morning I point the scope straight up and put the tracing tablet face-down on top.  Fire up the computer, let the camera cool, and take my Flats (20 of them) and Dark-flats (another 20).  Just takes a couple of minutes.

 

You can use Bias, but they presume that the only contribution is the base sensor noise.  My camera (also an ASI2600MC-Pro) has a couple of hot pixels that don't show up until the exposure has some duration, so those pixels end up being under-corrected by applying only Bias.  Is the effect small, sure.  But it's best to develop good habits from the beginning; there are more than enough ways we mess up even so.  Dark-flats always work, they're quick to take, and can be built into libraries for reuse just like Bias.  The only down-side is that they need to be matched to the Flats, so you need a set of Dark-flats for each exposure and sensor temperature combination.  The tracing tablet is pretty consistent, but the filter affects the exposure length, I need one set with the filter exposure, and one set without; for my rig it's 5 seconds and 2 seconds, respectively.  They're just Darks, same as we do with the Lights.  The filter itself doesn't matter with the dark-flats, just the exposure and temp.  I simply take a new set every time since they're so easy to do, and I don't have to remember that Bias is different.  Put the scope cover back on, and push the "go" button one more time.  Lights and Darks (aka Dark-lights), Flats and Dark-flats.  Pair of pairs.  Done.

[Whereisclearsky]

The key to taking good Flats is to take them with exactly the same optical train as the Lights that they will be correcting.  So the filter in its exact position must be included, and by "exact" I mean without moving it.  Don't take things apart and put them back together.  This is a hobby of microns, and microns are very very very small.  You simply can't put things back together with enough precision to have things line up afterward.

 

My workflow is to take my Lights, then roll the scope back into the garage in the state it was when the imaging session ended.  Next morning I point the scope straight up and put the tracing tablet face-down on top.  Fire up the computer, let the camera cool, and take my Flats (20 of them) and Dark-flats (another 20).  Just takes a couple of minutes.

 

You can use Bias, but they presume that the only contribution is the base sensor noise.  My camera (also an ASI2600MC-Pro) has a couple of hot pixels that don't show up until the exposure has some duration, so those pixels end up being under-corrected by applying only Bias.  Is the effect small, sure.  But it's best to develop good habits from the beginning; there are more than enough ways we mess up even so.  Dark-flats always work, they're quick to take, and can be built into libraries for reuse just like Bias.  The only down-side is that they need to be matched to the Flats, so you need a set of Dark-flats for each exposure and sensor temperature combination.  The tracing tablet is pretty consistent, but the filter affects the exposure length, I need one set with the filter exposure, and one set without; for my rig it's 5 seconds and 2 seconds, respectively.  They're just Darks, same as we do with the Lights.  The filter itself doesn't matter with the dark-flats, just the exposure and temp.  I simply take a new set every time since they're so easy to do, and I don't have to remember that Bias is different.  Put the scope cover back on, and push the "go" button one more time.  Lights and Darks (aka Dark-lights), Flats and Dark-flats.  Pair of pairs.  Done.

After taking lights I point my scope at zenith to

be able to lay down the led panel on top.

 

I don’t see the point of taking dark flats anymore, it’s enough to take bias frames for flat calibration.

As for hot pixels, my camera has some too, but you don’t even need calibration frames for them,

WBPP will easily get rid of them.

[bobzeq25]

The key to taking good Flats is to take them with exactly the same optical train as the Lights that they will be correcting.  So the filter in its exact position must be included, and by "exact" I mean without moving it.  Don't take things apart and put them back together.  This is a hobby of microns, and microns are very very very small.  You simply can't put things back together with enough precision to have things line up afterward.
 
My workflow is to take my Lights, then roll the scope back into the garage in the state it was when the imaging session ended.  Next morning I point the scope straight up and put the tracing tablet face-down on top.  Fire up the computer, let the camera cool, and take my Flats (20 of them) and Dark-flats (another 20).  Just takes a couple of minutes.
 
You can use Bias, but they presume that the only contribution is the base sensor noise.  My camera (also an ASI2600MC-Pro) has a couple of hot pixels that don't show up until the exposure has some duration, so those pixels end up being under-corrected by applying only Bias.  Is the effect small, sure.  But it's best to develop good habits from the beginning; there are more than enough ways we mess up even so.  Dark-flats always work, they're quick to take, and can be built into libraries for reuse just like Bias.  The only down-side is that they need to be matched to the Flats, so you need a set of Dark-flats for each exposure and sensor temperature combination.  The tracing tablet is pretty consistent, but the filter affects the exposure length, I need one set with the filter exposure, and one set without; for my rig it's 5 seconds and 2 seconds, respectively.  They're just Darks, same as we do with the Lights.  The filter itself doesn't matter with the dark-flats, just the exposure and temp.  I simply take a new set every time since they're so easy to do, and I don't have to remember that Bias is different.  Put the scope cover back on, and push the "go" button one more time.  Lights and Darks (aka Dark-lights), Flats and Dark-flats.  Pair of pairs.  Done.

Any halfway decent number of subs (20?) will have those hot pixels removed _perfectly_ by the stacking algorithm.

There is NO advantage to dark flats over the much simpler bias for any cameras except the 1600s and the 294s, which really don't like very short exposures.

NONE. I've run the numbers and done the experiment. The "hot pixel issue" is simply of NO consequence. You can do one master bias and use it for flats of ANY exposure time. For mono work, that's a BIG deal.

No advantage to dark flats. NONE. The extra work of matching exposures of flats is just wasted time.

By the way I have both 2600s. Use only bias. Anyone want to try spotting any issues with these images? 2600 MC, and 2600 MM. Click on the crummy CN thumbnails to get the real things, and details.

I tried dark flats once. There was NO change in the final image. NONE. I've also done the numeric analysis.

Sure you can see hot pixels in individual frames. But those are EXACTLY what data rejection algorithms are designed to remove, and they simply disappear from the final image.

As far as having to remember any differences, that too is not an issue. Any decent stacking program handles it for you. I don't know of any that doesn't.

The "need" for dark flats is an urban legend. It's origin is due to the fact that the first popular CMOS cameras, the 1600s, needed them because it really didn't like very short exposures.

I have yet to see an argument for dark flats that is anything but rhetoric. Or individual frames. Meaningless.

Edited by bobzeq25, 06 December 2024 - 09:37 PM.

[ratskrad]

I generally as a rule do 20 darks, 20 flats, 20 dark flats and 20 bias. Do I us them when processing, sometimes if I notice

issues with my subs that I know can be fixed with calibration. I always have to use flats when shooting with my 10" SCT

using the Starizona SCT 6.3 reducer because there is too much vignetting and the fact my dark skies are not so dark due

to lights running 24/7. I put them into folders and use them until I make changes in how I am shooting. Filters are a PITA

and I'll usually do flats the next morning to keep the optical path the same so as to move on to the next evenings targets.

I have 3 ASI2600 cameras and a Orion G26 OSC and all are based on the IMX 571 chip and do the same with all of them

as to processing and collecting subs.

[Whereisclearsky]

I tried dark flats once. There was NO change in the final image. NONE. I've also done the numeric analysis.

Do you know how to analyse it in Pixinisght?

I also don’t see any difference but can’t trust my eyes 100%.

 

Also wondering why do I get vignetting with Redcat 51 since it’s supposed to work fine with FF.

Maybe it’s the filter?

Edited by Whereisclearsky, 07 December 2024 - 03:47 AM.