14 replies to this topic

[westphalian]

so I have just watched Robin Glover's video re calculating exposure times

https://www.youtube....3UvP358&t=3079s.

 

can I just put these values by you to make sure they are in ball park and I have understood it.

 

so I am in Bortle 5, using asi533mc pro with zwo duo band filter for emission nebulae....360 seconds. Without the duo filter 60 seconds (I assume IR cut filter has little effect on exp time)

 

for C8 with IR filter 60 seconds also (well 66 to be precise using his calculation)

 

sound about right?

Andrew

[Anomaly_11]

I don't understand, for narrowband according to this video you would need like a +5hour long subs... something is not right, or just useless.

[hyiger]

so I have just watched Robin Glover's video re calculating exposure times

https://www.youtube....3UvP358&t=3079s.

 

can I just put these values by you to make sure they are in ball park and I have understood it.

 

so I am in Bortle 5, using asi533mc pro with zwo duo band filter for emission nebulae....360 seconds. Without the duo filter 60 seconds (I assume IR cut filter has little effect on exp time)

 

for C8 with IR filter 60 seconds also (well 66 to be precise using his calculation)

 

sound about right?

Andrew

It doesn't have to be precise. For me, I just wing it based on the target and the scope. For narrow band with an SCT at f/7 (I'm assuming you are using a reducer, if not you should) I usually take 2-3 min subs. I would sometimes but rarely go up to 10m if the target is weak in a particular band. It really depends on the target though. If it's a super bright planetary then I may only do 1 min subs. The key really is total integration time. Theoretically 60x 1m sub is or should be the same as 10x 6m sub. The advantage of taking many subs is that if a cloud gets in the way or the mount gets bumped you don't lose as much data. The downside is lots of files. 

Edited by hyiger, 13 May 2025 - 06:24 PM.

[DevilDog68]

I don't understand, for narrowband according to this video you would need like a +5hour long subs... something is not right, or just useless.

Okay, I'm glad someone else thought the same I did. My NB subs are 300 sec while my L are 60 and my RGB are 180. They're good enough for me... My goal is not APOD.

[kg7]

I think exposure time depends very much on your target and your mount, not just your skies and your scope/camera.  For example, most people who publish really great globular photos seem to use very short exposures, 15s or less.  People with overloaded mounts or bad mounts for their focal length often have to go shorter to avoid oblong stars.  I use shorter subs on nights with gusty winds, so that I can throw away a bunch and still have something left.  

 

I've imaged up to 10m subs on some targets at f/2, but it virtually guarantees that your stars will bloat and even some areas of nebula will go to full white.  With the right focal length and processing, that can all be managed.

 

My questions are mostly on the opposite end; how short is too short?  If I have the hd space and am willing to deal with the processing time, what do I lose by doing 2000x15s subs instead of 500x60s subs?

[jml79]

There are some really good videos on YouTube by a channel Deep Sky Detail covering this exact topic. Shorter subs will usually mean less SNR in the faintest regions of the image. This may or may not matter. If you have a lot of light pollution then the faintest parts of an image are going to be washed out by light pollution and shorter subs won't do any harm. If you are under dark skies and aiming to gather the faintest details then this will matter much more. If you are in B6 or above then forget about IFN and faint dust unless you are willing to go to extremes of integration and sub length. You need to think about this more for B4 and below.

 

In short, unless faint IFN or dust is part of my goal then you don't loose much by going short, I like 60s but even 15s should work (especially if your scope is fast) but when I am hunting the faint stuff, I go as long as 5m for Lum and 10-15m for NB. 5m Lum's seem extreme but I only max out about 2300 pixels out of 12 million doing this and I usually throw those away and don't use the stars from the Lum channel anyway. Of course this is totally location dependent. That wouldn't work if I didn't have dark skies.

[smiller]

so I have just watched Robin Glover's video re calculating exposure times

https://www.youtube....3UvP358&t=3079s.

 

can I just put these values by you to make sure they are in ball park and I have understood it.

 

so I am in Bortle 5, using asi533mc pro with zwo duo band filter for emission nebulae....360 seconds. Without the duo filter 60 seconds (I assume IR cut filter has little effect on exp time)

 

for C8 with IR filter 60 seconds also (well 66 to be precise using his calculation)

 

sound about right?

Andrew

You’re probably in the ballpark, but to know for sure, what are the focal ratios of the scopes and gain assumption on the camera?

 

As others have noted, the video is about the read noise of the camera, which is one of many factors that are considered for exposure times.  Glover’s primary point is that camera read noise, once the primary driver for ever longer exposure times, often isn’t the driver in most common situations and we are free to consider other factors.  Mount/guiding accuracy, wind, and keeping star over saturation to a minimum are also some of the many factors to consider.   Given all the trade-offs there’s usually a pretty broad range of acceptable exposure times that give similar results… with slightly different tradeoffs.

Edited by smiller, 14 May 2025 - 03:26 PM.

[unimatrix0]

I don't understand, for narrowband according to this video you would need like a +5hour long subs... something is not right, or just useless.

The video and the numbers are outdated. The specs used were specs for poor sensitivity- CCD cameras in the 2010s. 

 

In 2025- with the 2600 or 533 and 585 chips - these values have to be modified, especially the quantum efficiency and other specs of the cameras. 

 

People would be better of seeing the math while considering the age of the video and the obsolete tech in it. 

Edited by unimatrix0, 14 May 2025 - 09:42 AM.

[kathyastro]

Don't stress about exposure times.  Experiment.  It will depend less on your conditions and more on your equipment.

 

Exposure times should be long enough to lift the faint parts of your target out of the read noise.

 

They should be short enough to:

- avoid saturated star centres

- avoid tracking issues.

 

As your alignments and tracking improve, you might be able to increase your exposure times, but they will still be limited by star saturation.

 

So experiment.  See what works for you.

 

My "standard" times at f/4, with an elderly Atik 383L+ camera are:

120s binned 2x2 for R, G, B filters

300s binned 1x1 for L filter

600s binned 1x1 for Ha and Oiii

900s binned 1x1 for Sii

[smiller]

I don't understand, for narrowband according to this video you would need like a +5hour long subs... something is not right, or just useless.

 

 

The video and the numbers are outdated. The specs used were specs for poor sensitivity- CCD cameras in the 2010s. 

 

In 2025- with the 2600 or 533 and 585 chips - these values have to be modified, especially the quantum efficiency and other specs of the cameras. 

 

People would be better of seeing the math while considering the age of the video and the obsolete tech in it. 

Make sure to use his CMOS numbers, not his CCD entries… His CMOS entries are still relevant.  His baseline pixel size is 3.75um which is still the most common pixel size, his system quantum efficiency assumption is 50%, which is actually pretty close because you have not just the camera but losses from the telescope and any loss from a filter you have,  even a camera that has 80% average QE in a scope that is 80% transmission and a UV/IR cut filter that cuts 5% will only have a ~60% system QE.  Put a narrow dual and on there that often peaks at 85-90% and you are at ~50% total system QE.  My reflector with reducer is probably only about 50% total QE.

 

His noise for CMOS is 2.5e which isn’t high gain mode (at about 1.5e) but is better than gain 0 in most cameras.  I tend to use 1.5e as I star with HGC mode even for broadband with my shorter exposures.  So that is one that needs modification, but you can do that I just multiply the results by (yourReadNoise/2.5)^2

 

so I think his tables are completely relevant and you can easily modify the results with any differences in your system.

 

But once again this is all about the read noise which is one element, it can be important, but it’s only one thing.

 

If you run the numbers with a very narrowband filter from dark skies with a slow telescope you can get huge numbers even with CMOS cameras in high gain mode. What that is just telling you is that you’re still gonna have some significant read noise as a contribution because you’re not gonna use 30 minute exposures… or whatever his result shows.  Still, it’s useful to know.  It may encourage someone with an F10 SCT to get a reducer for instance, or to push the gain a bit higher… all useful stuff.

Edited by smiller, 14 May 2025 - 10:00 AM.

[happylimpet]

so I have just watched Robin Glover's video re calculating exposure times

https://www.youtube....3UvP358&t=3079s.

 

can I just put these values by you to make sure they are in ball park and I have understood it.

 

so I am in Bortle 5, using asi533mc pro with zwo duo band filter for emission nebulae....360 seconds. Without the duo filter 60 seconds (I assume IR cut filter has little effect on exp time)

 

for C8 with IR filter 60 seconds also (well 66 to be precise using his calculation)

 

sound about right?

Andrew

I could believe those times - though you havent given a focal ratio.

 

I tend to go short, because i like sharp images and like to throw away poor subs. People who do 10min subs probably have no idea how much sharper their stack would be if they did 10 sec subs. A bit noisier, but a lot sharper.

 

But at f4, I should be using 2sec subs for some filters with m19 skies. So I actually go longer at 10secs than needed for some (as 2secs makes a collossal amount of data, even though it would be deliciously sharp) and I do 60sec subs for narrowband even though i should go longer. Thats OK, its only adding 20% or so to the noise (as I recall), and i want to separate out the wind gusts without throwing away all my data.

 

Finally, as noted above, I get insanely sharp globular images by using 20msec exposures. Basically planetary lucky imaging style. I just imaged the cores of M13 and M3 on an average night and got 0.65" FWHM.

Edited by happylimpet, 14 May 2025 - 10:11 AM.

[kg7]

You’re probably in the ballpark, but to know for sure, what are the tidal ratios of the scopes and gain assumption on the camera?

This is probably the most overlooked point when I see this discussion on the forum.  The difference between shooting at f/2 and f/10 turns a 15s sub into a 375s sub.  Most of us are thinking about possible sub lengths that vary by a factor of 2 or 3 maybe, 5, but not by 25x.  I don't think anyone can give exposure advice without first asking about focal ratio.

 

 

Glover’s primary point is that camera read noise often isn’t the driver for ever longer exposure times in most common situations and we are free to consider other factors.  Mount/guiding accuracy, wind, and keeping star over saturation to a minimum are also some of the many factors to consider.   Given all the trade-off there’s usually a pretty broad range of acceptable exposure times that give similar results… with slightly different tradeoffs.

I used to rely heavily on Glover's own "smart histogram" feature in sharpcap, which measures your sky brightness and makes recommendations for exposure, gain, offset, and bit depth.  It's a great feature.  Then I realized that it's only looking at your background target, not your stars, and for virtually every image I've used it for, even moderately bright stars are blown out.  Max pixel values over a huge area, which is very hard to deal with while processing.  My early Bubble Nebula attempts had fantastic depth, and then a huge ugly star that's 25% the size of the nebula right in the middle. 

 

I still use the smart histogram to get me in the ballpark, but then I check my pixel counts on the first few subs and almost always end up shortening my subs if there are really any notable stars in the field.

[PIEJr]

Like Kathy said, experiment. I've done that since my beginning.

 

I began with a real dud of a camera. 5 friends volunteered to try and stack the images from it. They all could not do it. Too noisy I think.

Anyway, I tried longer and longer exposures, and finally images began to emerge.

One of my friends told me, "You stack with time."

Years later another approached me with an offer of loaning me an Atik Infinity OSC camera. It was instant success! And he sold me the camera.

 

Now, better everything. But I still like to just experiment with exposure times. I usually just pick round times, and apply to all filters (when I was using filters).

10-20-30-60-90-120-180-240-300-{360-420-480}-600. seconds. I just like round times. But rarely do any 10 minute exposures anymore. Just for curiosity sakes for 600s exposures. Wide eyed wonder.

Mostly, I do 30 to 300 seconds to see what detail emerges. (1/2 minute, to 5 minutes each.)

 

Cuiv has a good video on exposure times. I really like video learning like Cuiv presents. I can stop-backup-rerun- and try the different things presented.

Or just take stabs at it to find something I like. (Lazier than the Lazy Geek.   )

There is no "one size fits all" to this stuff. So experimenting with your exposure times can be a lot of fun. 

[westphalian]

thanks all.

F6 F6.7

gain 100

[smiller]

thanks all.

F6 F6.7

gain 100

At F6, Bortle 4 for his benchmark CMOS camera with 3.75um pixels, mono, read noise 2.5e, and 50% system QE he lists 24 seconds for 5% read noise contribution.

 

For your F6.7, color broadband, gain 100 (red noise 1.5), your 5% read noise is at:

 

24 * 3 * (1.5/2.5)^2 * (6.7/6)^2 = 32 seconds 

 

For narrow band your filter is 15nm for Ha and 35nm for Oiii, let’s call it 25nm overall.  Then you need to multiply the broadband number by 4 (i.e. 100nm/25nm), so you get 128 seconds.

 

So about 30 seconds broadband and 120 seconds narrowband for 5% read noise contribution. 

Edited by smiller, 14 May 2025 - 04:26 PM.