48 replies to this topic

[PokeyMinch5236]

I took a picture of the North America Nebula last night and I initially stacked it with just biases and flats. However, I get frustrated as usually when I post process I see little small pixels of reds, greens, and blues making up the black parts of the sky. Naturally I decided to take dark frames and see if they work to reduce this noise level, but I feel like it made the image worse. I did take this shot under a full moon, so is it just the full moon's light infesting the picture or is it the lack of dark frames? After doing nearly identical post processing methods on one with darks and one without it seems to me that the one without darks looks better. I made sure everything is properly set up, all calibration and light frames are the same gain and temperature ratings. I made my flats 1 second 250 gain by using my smartphone and white t-shirt to reach about 26k ADU peak on the histograms. All frames are at -12 degrees Celsius including the new darks. 

Here's the one with darks:

 

 

Here's the one without darks:

 

PS i just realized that I forgot to use noise reduction on the picture with darks but even with noise reduction you can still see specks of color where it should be black

[Spaceman 56]

Darks should make a OSC image better.  

 

are the darks you used the same exposure and gain settings as your lights. ?

 

meaning if you took a 60 second exposure for the Lights, the Darks would be 60 seconds exposure as well ?

 

and if you shot the lights at gain 100, the darks would be gain 100 as well ?

 

good to check the basics before adding suggestions. 

[PokeyMinch5236]

Darks should make a OSC image better.  

 

are the darks you used the same exposure and gain settings as your lights. ?

 

meaning if you took a 60 second exposure for the Lights, the Darks would be 60 seconds exposure as well ?

 

and if you shot the lights at gain 100, the darks would be gain 100 as well ?

 

good to check the basics before adding suggestions. 

Yes like I mentioned everything is identical. My lights are at 250 gain 60 sec exposure and so are the darks.

[danny1976]

Hot pixels should also be removed by your stacking software. Siril's OSC_preprocessing script takes care of that. I don't see much difference with or without darks on my ASI2600MC. But I use them anyway, 60 to 100 darks.

[bobzeq25]

People differ on this, and it depends on the camera.  I don't use them with my 2600s, they were absolutely necessary to eliminate amp glow on my 183s.

 

I also only cool my 2600s to 0C.  This explains why.

 

https://www.youtube....h?v=3RH93UvP358

[soojooko]

I have an ASI533 and don't use darks.

 

I had a recent project that had the clumps of R, G & B pixels. They are usually removed in stacking but in this case they were not. Most of the advice I got was to apply darks. I knew this wouldn't help, but I took them and restacked anyway. Sure enough, the hot pixels were still there. I saw no benefit to applying the darks to be honest. I really think they are a waste of time with some sensors, if there has been enough integration.

 

In my opinion, dithering and stacking is the best way to remove hot pixels. Aside from the above anomoly, its worked for me consistently. 

[PokeyMinch5236]

I have an ASI533 and don't use darks.

 

I had a recent project that had the clumps of R, G & B pixels. They are usually removed in stacking but in this case they were not. Most of the advice I got was to apply darks. I knew this wouldn't help, but I took them and restacked anyway. Sure enough, the hot pixels were still there. I saw no benefit to applying the darks to be honest. I really think they are a waste of time with some sensors, if there has been enough integration.

 

In my opinion, dithering and stacking is the best way to remove hot pixels. Aside from the above anomoly, its worked for me consistently. 

I see, so the cooling really just eliminates the need for dark frames. I guess both of us had the same issue here. For my shots on the Lagoon and Eagle Nebulae there wasn't much of those hot pixels there either so maybe its just the NA Nebula.

Another question I have is that if I decided to take dark flats/flat darks, could I stack those with flats and biases? Or would biases be completely uneccessary or detrimental to the image?

[drd715]

The cooling reduces read noise. You still have to deal with hot pixels. Darks are the first step or dithering will move the location of the hot pixels position in the image so stacking can then also reduce the effects of the hot pixels in your stacked image.

[F.Meiresonne]

People differ on this, and it depends on the camera.  I don't use them with my 2600s, they were absolutely necessary to eliminate amp glow on my 183s.

 

I also only cool my 2600s to 0C.  This explains why.

 

https://www.youtube....h?v=3RH93UvP358

Why not deeper?

I go to -5°C , some people seem to believe it does matter. Myself i don't know.(yet)

 

As for the OP, my last session i made with and without darks. When i viewed it in PI with a light stretch, the one with the darks seems to show the nebulae better, like the signal got better. But i will have to test this further in time to see if it really matters. I use a Touptek 2600 color cam

[Sheridan]

I used darks with my 2600 Risingcam. I just got used to making them because of my 183MC. You had to with that camera to get rid of the amp glow.

[soojooko]

I see, so the cooling really just eliminates the need for dark frames. I guess both of us had the same issue here. For my shots on the Lagoon and Eagle Nebulae there wasn't much of those hot pixels there either so maybe its just the NA Nebula.

Another question I have is that if I decided to take dark flats/flat darks, could I stack those with flats and biases? Or would biases be completely uneccessary or detrimental to the image?

Bias are for the flat frames. If you take flats then you need bias or flat darks.

 

To be clear - my point was related to the 533/2600 sensor cameras. Cant speak for the rest. But I assume all new cameras will have excellent noise handling going forward.

 

As far as hot pixels go - Han, the author of ASTAP had some interesting insights. He suggested that the background level of the session has a bearing on the stacking software's ability to pick out the hot pixels. He made some adjustments to ASTAP in this regard and the problem was improved. But the point still stands: hot pixels 'should' be removed in stacking if you dithered your subs.

Edited by soojooko, 01 October 2023 - 06:58 AM.

[Oort Cloud]

On cameras that do not have amp glow, it is possible to use the master bias for both the flats and lights, rather than using master bias for flats and master dark for the lights. But you can not just omit the master dark; you have to use the master bias in its place, or you will not get correct flat calibration.

When doing this method, you must also dither your lights, or any hot pixels will stack along with the target, becoming brighter the deeper the stack gets. With dithering, the hot pixels will move around relative to the target once the frames are registered, so they get removed during stacking, especially if a rejection algorithm is used.

So it's either a master dark -or- a master bias in place of the master dark, along with a large set of lights (50-100 minimum), dithered at least every 2-3 frames, and a camera with no amp glow.

In the end, it's sometimes easier to just take darks. But we also say you should dither unless your mount has issues settling, because the improvement it provides is usually worth the time spent dithering and settling.

I just bought some parts so I can mount my guidescope to my DSLRs hot shoe. I got a fantastic image of Cygnus, but the shadows are riddled with walking noise because I wasn't able to guide or dither since I had no way to mount the guider when using the camera with a lens instead of a telescope.

[unimatrix0]

Your question is worded wrong.  The appropriate question would be "Do I need darks for my asi 533mc pro?" 

 

 

Cameras vary substantially and most will require darks no matter what.  With the 533 and 2600 sensors you can get away with, but take an example when I tried at higher focal lengths and the remaining hot pixels confused the stacking program, mistaken for stars and I did need darks with the 533 mc pro.   Without the darks, I needed to do a "cosmetic corretion" in Pixinsight, which picked up on the hot pixels and I had to include it in the stacking process to get my image stacked. That was on 1300mm focal length. I can't stack anything at that focal length if I don't do darks or I don't do cosmetic correction. Either or both, but darks were better, because the CC was more of a bandaid. 

 

 

Also, the images you posted are good, but as an example is not good.  We can't zoom in to point out the issues.  When you go down and become more nitpicky and do pixel peeping and checking your image at 1:1 or a drizzled image, you gonna start noticing all the hot pixels that left in your image. Looks good on social media with all the compression and small window to see, but put it on astrobin as a full blown image and you start noticing the issues.  Later on  as you want a more high resolution and clarity and clean background and so on, whatever noise left over due to lacking darks will start to bother you. 

If you are 100% satisfied with whatever you get without darks, and you achieved everything you aimed for then please ignore my comment. 

Edited by unimatrix0, 01 October 2023 - 09:27 AM.

[Dan Finnerty]

The cooling reduces read noise. You still have to deal with hot pixels. Darks are the first step or dithering will move the location of the hot pixels position in the image so stacking can then also reduce the effects of the hot pixels in your stacked image.

Minor nit to pick. Read noise is mostly independent of temperature. Thermal noise is dependent on temperature (that's a bit oxymoronic...), which is why you must match temperature of darks with lights.

[Robert7980]

Darks aren’t really needed for the modern CMOS cameras, but they are so easy to take and apply I can’t see a good reason not to have a dark library… If you’ve gone as far as acquiring expensive special purpose astro gear why not get the most out of it. Takes me about 3 seconds to include the darks in the stack, not really a big deal… 

[soojooko]

The thing about darks as they relate to hot pixels - in my case the hot pixels are not fixed. Ive compared images from 2 different sessions and the hot pixels are in different places. So I dont see how applying a master dark that I took a few weeks ago can help remove the pixels - seeing as said dark has its hot pixels in different places.

 

This really confuses me as I always thought hot pixels on camera sensors are the same as hot pixels on a TV screen; they are basically damaged pixels that always show full brightness in R,G or B. But I think I was wrong in this assumption. Can somebody who knows more than me explain what hot pixels are on camera sensors? And is it normal for them to not be permanently fixed?

[AhBok]

I may have missed it, but is the camera an ASI533MC Pro? If so, a gain of 250 and 60 sec exposures would be too much for my 533 in Bortle 4/5 skies. Under a full moon (if I chose to image then), I’d think something like 15-20 seconds at unity gain (100) on a bright object like the NA would yield far better results.

[WadeH237]

The thing about darks as they relate to hot pixels - in my case the hot pixels are not fixed. Ive compared images from 2 different sessions and the hot pixels are in different places. So I dont see how applying a master dark that I took a few weeks ago can help remove the pixels - seeing as said dark has its hot pixels in different places.

Darks are not just related to hot pixels.  Every pixel has dark current that benefits from proper calibration.

 

It's interesting that your camera has hot pixels that are not fixed.  This is extremely unusual.  On all of my cameras, if I blink dark frames that match my light frame exposure times, the fixed pixels are very obvious, and they do benefit from calibration.

 

In terms of whether you "need" darks, of course you don't need them.  You can clean up the worst offenders with either a bad pixel map, or something like CosmeticCorrection in PixInsight.  A better question is whether your camera benefits from them.  In that regard, every camera I've ever had benefits from proper dark caliabration.

 

I have both an ASI2600MC and ASI2600MM.  The mono camera is fairly recent.  When I first got my one-shot-color camera a few years ago, I skipped dark calibration and just did bias and flats - solely because most users of the camera claimed that it didn't need them.

 

...And then I did my own evaluation of my camera.  What I discovered is that, at least my sample of, the ASI2600MC Pro has very significant fixed pattern noise, and most of it appears to be dark current.  As I mentioned above, there are ways of dealing with this noise, but it all involves replacing the pixels with values that "look" OK.  In my mind, it is better to calibrate those pixels than to replace them with fake data.  So ever since, I have done a dark calibration with my ASI2600MC Pro.

 

When I got my ASI2600MM Pro, I automatically did a check of some long darks even before putting the camera on a telescope.  I found the same characteristics as my older ASI2600MC Pro.  So from day 1, I have used darks with this camera, too.

 

So here is my response to the question in the title:

 

• Are darks needed for osc cooled cameras?  No.
• Will osc cameras produce better data with darks, than skipping them?  Yes!

 

-Wade

[Astro_Art]

Darks aren’t really needed for the modern CMOS cameras, but they are so easy to take and apply I can’t see a good reason not to have a dark library… If you’ve gone as far as acquiring expensive special purpose astro gear why not get the most out of it. Takes me about 3 seconds to include the darks in the stack, not really a big deal… 

 

I have the 533MC Pro which does not have amp glow, and consequently, doesn't require darks. However, I typically apply a handful (5 or 6) darks anyway to go along with 100 bias, 30 flats, and anywhere from 30 to 60 lights. Lights are usually 300 second (5 minute) subs.

 

How many darks do you add? Are 5 or 6 enough or should I take more?

[soojooko]

It's interesting that your camera has hot pixels that are not fixed.  This is extremely unusual.  On all of my cameras, if I blink dark frames that match my light frame exposure times, the fixed pixels are very obvious, and they do benefit from calibration.

Hmm... ok. That's given me the impetus to triple check what I'm seeing.

 

My understanding is: if the darks are the same gain and exposure then I can apply them to lights I took say 2 - 4 weeks ago? I see people talking about dark libraries they use for months so I figured it should be fine. So if it had no effect on the hot pixels - I need to revisit my stacking settings. Still so much I don't know. 

[WadeH237]

How many darks do you add? Are 5 or 6 enough or should I take more?

The thing is, that dark current has shot noise, just like light from the sky.  When you subtract the master dark, this noise will be added to the noise already in the light frame.  So with a small number of darks in the master, you might end up improving fixed pattern noise (hot pixels), but increasing the statistical noise in the image.

 

Taking a larger number of dark frames is just like taking a larger number of light frames.  It increases the signal-to-noise ratio in the integrated result.  This effectively reduces the noise contribution when you do the dark calibration.

 

The question, then, is:  How many darks are enough?

 

I don't think that there is a hard-and-fast rule about this.  There are significantly diminishing returns as you add more frames.  I would say that 5 or 6 is probably too few, though.  Someone did an analysis that I read a whole bunch of years ago that showed a graph on the effect of noise reduction with adding dark frames.  If I remember right, there was a "knee" on the graph right around 15 frames.

 

I used to use 15 as the number, based on the above.  But more recently, I have stopped going for a fixed number.  Instead, I like to take two nights' worth of darks (I actually take them at night...) and go with that.  As long as I get at least 15 frames, I am happy; any more are icing on the cake.  With my exposure times, and with the hours of night near the equinox, that puts me at right around 17 or 18 darks in the master.

[WadeH237]

My understanding is: if the darks are the same gain and exposure then I can apply them to lights I took say 2 - 4 weeks ago? I see people talking about dark libraries they use for months so I figured it should be fine. So if it had no effect on the hot pixels - I need to revisit my stacking settings. Still so much I don't know. 

The darks will be effective at calibrating lights taken under the same conditions.  That means that all of these need to be the same:

• Exposure time
• Gain
• Offset
• Temperature

If any of them are not matching, then they won't calibrate correctly.

[soojooko]

The darks will be effective at calibrating lights taken under the same conditions.  That means that all of these need to be the same:

• Exposure time
• Gain
• Offset
• Temperature

If any of them are not matching, then they won't calibrate correctly.

Yep - they absolutely are the same. I'm gonna do some restacking....

[bobzeq25]

I see, so the cooling really just eliminates the need for dark frames. I guess both of us had the same issue here. For my shots on the Lagoon and Eagle Nebulae there wasn't much of those hot pixels there either so maybe its just the NA Nebula.

Another question I have is that if I decided to take dark flats/flat darks, could I stack those with flats and biases? Or would biases be completely uneccessary or detrimental to the image?

What's completely unnecessary are flat darks.  With cameras other than the 1600s and 294s, bias frames are the same as flat darks.  With my 2600s I've run the numbers and done the experiments.  No difference.

 

"CMOS needs dark flats" is your basic urban legend.  It comes from the old 1600s, a previously popular CMOS camera which does not like very short exposures.

 

What you really don't want to do is use both, that's _asking_ for processing mistakes.  And has NO value at all.  They're the same thing, using both is duplicative, and in a bad way.

 

So pick one.  I use bias.  I see no need for the complications of dark flats.  If I changed my flat exposures, I'd need new dark flats.  With bias I can change flat exposure any way I want to, and use the same bias.  Broadband, narrowband.  Makes no difference.

 

With modern cameras bias and flats are all you need.  Bias is trivial, flats are relatively easy.

Edited by bobzeq25, 01 October 2023 - 08:14 PM.

[Oort Cloud]

I should add, this isn't necessarily _all_ cameras without amp glow. It also has to do with modern sensors having lower thermal noise overall; it doesn't work as well on my MILC as it does on my 533mc. But in the winter, when the MILC is down near 0°C, it will perform a lot better, as I've already seen on warm vs cool nights. The 533 has the benefit of being cooled, and thus, always at 0°C or lower. At that point, a dark frame is basically a bias frame plus a handful of hot pixels that disappear with any significant dithering of the lights during acquisition.

So basically, this applies more to cooled cameras, specifically ones based off the IMX455, IMX571, or IMX533. Probably the newer planetary cams too, like the IMX585.