Gain and offset for very dim details - Forums

Topic starter

Mar 13, 2025

Hi everyone,

As you certainly know, the galaxy NGC3628 has got a very dim (I hope that is the right word), tail of dust. Almost all images do not show this dust tail but it exists and some images show it.
I have processed my most recent image of the Leo Triplet, a result of 508x120" subs, taken with my QHY600PH C on my TSRC12". Everything fine, you can see that image in my portfolio here on Astrobin. However, that dim tail does not show up - it also is not existing on the raw stacks.
Now my question - I used - as recommended, a gain setting of 26 and offset 40 and my suspicion is that this offset of 40 was too high to make that dust tail visible. Other user told me to disregard those settings and use gain 60 and offset 0, some said gain 0 and offset 10…an so on.

Have you had a similar situation with some very dim details and how did you resolve them successfully?

You comments and suggestions are highly appreciated!

Thanks and CS,
Georg

andrea tasselli

Frequent contributor

Mar 13, 2025

An offset is just a fixed DN added to the recorded image in order not to clip your black point. To verify it just take bias frame and see that no value is zero with your offset of 40 (which is what I use, by the way). That has nothing to do with not showing the faint dust tail of NGC3628, as long as you are not clipping you black point. I barely, just barely, captured it on this very poor image of mine with very humble equipment (80mm f/7.5 refractor with a DSLR) with a B7 site, so you must certainly be able to pick it up.

Leo Triplet (M65, M66 & NGC3628) RGB (OSC) - AstroBin

Helpful Engaging Supportive

Anderl

Mar 13, 2025

Hey georg,

try to use blurx and noisex (or similar tools) before stretching the data. After applying noisex do an strong autostretch (shift + stretch) within the screen transfer function and you will most likely see the tail. Removing the stars will also increase the visibility

cs
anderl

Concise

Georg N. Nyman

Topic starter

Mar 13, 2025

andrea tasselli:
An offset is just a fixed DN added to the recorded image in order not to clip your black point. To verify it just take bias frame and see that no value is zero with your offset of 40 (which is what I use, by the way). That has nothing to do with not showing the faint dust tail of NGC3628, as long as you are not clipping you black point. I barely, just barely, captured it on this very poor image of mine with very humble equipment (80mm f/7.5 refractor with a DSLR) with a B7 site, so you must certainly be able to pick it up.

Leo Triplet (M65, M66 & NGC3628) RGB (OSC) - AstroBin

Yes Andrea, that is exactly what I thought to be visible - that faint dust tail. And you are right, I am in a B4 site with excellent equipment and very long integration time...it must be somewhere - I do an hunt for it!
Thanks for your feedback!

Georg N. Nyman

Topic starter

Mar 13, 2025

Anderl:
Hey georg,

try to use blurx and noisex (or similar tools) before stretching the data. After applying noisex do an strong autostretch (shift + stretch) within the screen transfer function and you will most likely see the tail. Removing the stars will also increase the visibility

cs
anderl

Hi Anderl,

I try it - and get back with results as soon as I have got some!

CS
Georg

Anderl

Mar 13, 2025

Georg N. Nyman:
Anderl:
Hey georg,

try to use blurx and noisex (or similar tools) before stretching the data. After applying noisex do an strong autostretch (shift + stretch) within the screen transfer function and you will most likely see the tail. Removing the stars will also increase the visibility

cs
anderl

Hi Anderl,

I try it - and get back with results as soon as I have got some!

CS
Georg

The way your picture will look if you do that will be terrible ;) but it will reveal whats inside your data and whatever it is there will be a way to display it in an eye pleasing manner

Rob Kiefer

Mar 13, 2025

Hi Georg,
Looking at you image, my guess is that you clipped the tidal tale. At 16h you should be able to see it. I only did 8h and was able to pick it up. I should have added another 8h of LUM. This would have helped…
Rob

Concise

Nigel A Ball

Mar 13, 2025

Georg

I use Gain 57 and Offset 10 on my QHY600SBFL with good results.

I found using BlurXT and NoiseXT carefully on the linear image is very important. It took me a while to fine tune the latest version of NoiseXT so I didn't lose any detail. There is a definite way to do this. Quite happy to share this procedure with you :-)

Nigel

Helpful Respectful Supportive

andrea tasselli

Frequent contributor

Mar 13, 2025

Rob Kiefer:
Hi Georg,
Looking at you image, my guess is that you clipped the tidal tale. At 16h you should be able to see it. I only did 8h and was able to pick it up. I should have added another 8h of LUM. This would have helped...
Rob

*I concur. You clipped your shadows rather heavily there.

Brian Puhl

IOTD Staff

Mar 14, 2025

Hey Georg,

Andrea pretty much answered your question, but I did want to add: You should consider shooting longer exposures. You're at F/6.4 according to the image details, which is a bit on the slower side, but not horrible. I'd recommend shooting 5 minute or more. Bortle 4 is a respectable sky condition, I shoot from the same. You should have no issue capturing the faint details. Take a few test exposures to make sure your galaxies aren't blown out, but the depth on a camera like yours should be more than enough to handle 5 or 10 minute. You will gain far more signal in a shorter amount of time. The downside is building up enough images for rejection so plan accordingly.

The topic of exposure length comes up often, too many folks seem to live or die by a certain old 'exposure calculator/spreadsheet' that a certain individual came up with (that shall not be named). I encourage you to push the boundaries. And just to prove to some of those folks, I took a single, 1 hour exposure of Leo Triplet with my QHY268M and Esprit 100 at F/5.5

Some core details are blown, in the lower galaxies, but Hamburger as a whole is not blown at all. And the entire tidal tail was captured in a single exposure. This is an extreme case, for sure, and I only did it for science.... but too many folks kept argueing me that it would be a blown out, washed out image.

Anyways, try longer exposures

Helpful Insightful Engaging Supportive

Dan Kearl

Mar 14, 2025

Brian Puhl:
Hey Georg,

Andrea pretty much answered your question, but I did want to add:    You should consider shooting longer exposures.   You're at F/6.4 according to the image details, which is a bit on the slower side, but not horrible.   I'd recommend shooting 5 minute or more.   Bortle 4 is a respectable sky condition, I shoot from the same.   You should have no issue capturing the faint details.   Take a few test exposures to make sure your galaxies aren't blown out, but the depth on a camera like yours should be more than enough to handle 5 or 10 minute. You will gain far more signal in a shorter amount of time. The downside is building up enough images for rejection so plan accordingly.

The topic of exposure length comes up often, too many folks seem to live or die by a certain old 'exposure calculator/spreadsheet' that a certain individual came up with (that shall not be named).   I encourage you to push the boundaries.    And just to prove to some of those folks, I took a single, 1 hour exposure of Leo Triplet with my QHY268M and Esprit 100 at F/5.5

Some core details are blown, in the lower galaxies, but Hamburger as a whole is not blown at all. And the entire tidal tail was captured in a single exposure.    This is an extreme case, for sure, and I only did it for science....     but too many folks kept argueing me that it would be a blown out, washed out image.

Anyways, try longer exposures

Exposing to the Right of the histogram is the same in any photography discipline. I shoot in Bortle 8 and do 10 minutes routinely. If the whites are not blown, shoot as long as you can.

Tony Hallas

Mar 14, 2025

Hi George,

    I would strongly concur with the above suggestion ... use longer exposures.   I go out to 20 minutes at f/7 when I want to record faint details ... I have made experiments to see if many short exposures are equal to longer exposures ... they're not. Assuming you are doing some sort of averaging of your data with dithered exposures ... you benefit a lot with more depth to each exposure. Conversely ... the average of crap ... is crap.

    I attached my Leo Trio using 2o minute subs RGB through an SVX 180 f/7.

      Best,

         Tony

Helpful Insightful Respectful Engaging Supportive

Georg N. Nyman

Topic starter

Mar 14, 2025

Hi everyone,

thanks a lot for your important and interesting replies - I shall try it again with longer exposure times per sub. I usually use 120 seconds, but I got your point. I read too often in the past, that 1000x120" is better than 100x1200" and there is no difference in what you get - but from what I see here, there is obviously a bid advantage if you shoot longer exposure times.

I shall get back with my results, once I have got some!

CS
Georg

Georg N. Nyman

Topic starter

Mar 14, 2025

I processed the entire stack totally different - not in Siril with the precooked skripts but in PixInsight - here the result - much better, I think and even the faint dust trail is already almost visible...
Well, I learn every day something and regarding the longer exposure times for the subs, I shall definitely do it and get back with some results once the rain is over!
CS
Georg

andrea tasselli

Frequent contributor

Mar 14, 2025

To be sure there are lots of things that would advise against extra-long exposures but in your skies 300s seem just the right way to go.

ScottF

Frequent contributor

Mar 14, 2025

I would be very careful removing gradients. I’ve had gradient removal wipe out faint detail.

Well Written

Brian Puhl

IOTD Staff

Mar 14, 2025

I would be very careful removing gradients. I’ve had gradient removal wipe out faint detail.

You need to be more careful then. Up your smoothing on DBE to something like 0.5 or 0.6 and see how it works for you. You can model a basic gradient with those settings in around 6-8 sample points.

Concise

ScottF

Frequent contributor

Mar 15, 2025

Brian Puhl:
I would be very careful removing gradients. I’ve had gradient removal wipe out faint detail.

You need to be more careful then. Up your smoothing on DBE to something like 0.5 or 0.6 and see how it works for you. You can model a basic gradient with those settings in around 6-8 sample points.

Where I really noticed it was processing the oxygen nebula near M31. I’ll try your suggestion, thanks!

Well Written