Crescent Nebula DSLR Woes

[imakebeer]

They say you can't polish a t**d but you can roll it in glitter.....

Fresh from my recent Veil Nebula Fail I've moved on to making bad images of the Crescent Nebula..... and I think I might have answered my own question.....

@simmo39 posted a cracking pic of the same recently so how come my image sucks, especially when I've got roughly double the integration time?

Well I did some searching and found this old post by @Richard_

 

From what he says, I would hazard a guess Simmo39 used a dedicated astrocam (goodness, I really hope so!) whereas I'm using a modified Canon 450D - if I'm understanding right from Richard_'s post above DSLR's are OK for galaxies* but fundamentally less sensitive to H-Alpha (whatever that is!) that you get from these nebula.....

.....right?

( * which explains why I've got some half decent pics of galaxies before)

So, is there much I can do such as even more integration time, different dithering etc, or am I fundamentally stuffed for these targets and need to pony up for a proper astrocam if I want to image them? (Or stick to galaxies!). Is it just a question of understanding the limitations of DSLR's? (I have a vague idea what H-alpha is really). On the flip side from what I read it seems a lot of dedicated astrocams have much smaller sensors so I'd end up with a much smaller image (I think???).

FWIW even though the nebula sucks I thought the stars had some lovely colours! And on the plus side I've had lots of fun manually stacking in Siril and learning scripting, so not a total waste of time! 🤣

For the record:
SW150PDS + Canon 450D
Stacked 76 x 3min subs (guided & dithered)
20 flats (but not sure why they aren't removing the dust bunnies!)
Siril, Graxpert, more Siril and GIMP (no need for darks or biases with a 450D, so I am reliably informed 😉)

[Elp]

Hydrogen alpha is hydrogen whereby electrons move within their atomic structure and the event can be seen which gives off red light in the 656nm bandpass of light. Better explained online.

The issue with standard daytime cameras is they block infra red of which HA infringes, so an unmodded camera won't respond to imaging it so well. A mod removes one filter so it can pass through to the sensor (basic mod, not a full spectrum conversion).

The other issue with DSLRs especially older ones is that their quantum efficiency is low compared to an astro camera, so their pixel response won't be as good.

I don't think there's fundamentally an issue imaging with a camera body, people have been doing it for years.

What will help you imaging emission nebulae rich in hydrogen alpha and oxygen iii is a dual narrowband OSC filter, it'll help separate the emission from the background sky and light pollution.

Flats likely haven't worked due to over or under exposure, temperature can make a difference, focus shouldnt change or shift from when taking your lights.

Smaller sensors generally mean a less resolution image (depends on pixel sizes too), so smaller file size and a crop IF compared to a larger sensor image (from the perspective of an unknowing they would perceive the smaller sensor image as zoomed in, it's not, it has simply imaged what the optics have allowed when projected onto the surface area of the sensor).

Edited September 18 by Elp

[simmo39]

36 minutes ago, imakebeer said:

They say you can't polish a t**d but you can roll it in glitter.....

Fresh from my recent Veil Nebula Fail I've moved on to making bad images of the Crescent Nebula..... and I think I might have answered my own question.....

@simmo39 posted a cracking pic of the same recently so how come my image sucks, especially when I've got roughly double the integration time?

Well I did some searching and found this old post by @Richard_

 

From what he says, I would hazard a guess Simmo39 used a dedicated astrocam (goodness, I really hope so!) whereas I'm using a modified Canon 450D - if I'm understanding right from Richard_'s post above DSLR's are OK for galaxies* but fundamentally less sensitive to H-Alpha (whatever that is!) that you get from these nebula.....

.....right?

( * which explains why I've got some half decent pics of galaxies before)

So, is there much I can do such as even more integration time, different dithering etc, or am I fundamentally stuffed for these targets and need to pony up for a proper astrocam if I want to image them? (Or stick to galaxies!). Is it just a question of understanding the limitations of DSLR's? (I have a vague idea what H-alpha is really). On the flip side from what I read it seems a lot of dedicated astrocams have much smaller sensors so I'd end up with a much smaller image (I think???).

FWIW even though the nebula sucks I thought the stars had some lovely colours! And on the plus side I've had lots of fun manually stacking in Siril and learning scripting, so not a total waste of time! 🤣

For the record:
SW150PDS + Canon 450D
Stacked 76 x 3min subs (guided & dithered)
20 flats (but not sure why they aren't removing the dust bunnies!)
Siril, Graxpert, more Siril and GIMP (no need for darks or biases with a 450D, so I am reliably informed 😉)

Yes im using an ASI 2400mc. a mighty fine camera. I think with still more time on target you willmpull out more data.

[imakebeer]

26 minutes ago, simmo39 said:

Yes im using an ASI 2400mc. a mighty fine camera. I think with still more time on target you willmpull out more data.

I am very relieved to read this! 😅

[imakebeer]

1 hour ago, Elp said:

What will help you imaging emission nebulae rich in hydrogen alpha and oxygen iii is a dual narrowband OSC filter, it'll help separate the emission from the background sky and light pollution.

Thanks @Elp, that all makes sense. 

I double checked - I had the rear filter removed from my camera, increasing Ha transmission from 25-97%-ish (not a full spectrum mod).

So with this in mind are you suggestions above still the right kind of filter to be looking for?

[Elp]

11 minutes ago, imakebeer said:

So with this in mind are you suggestions above still the right kind of filter to be looking for?

It'll work to a point, but an astro camera will generally be better, mainly because their noise profiles are finer especially if using a cooled one. At present it might be cheaper to try for a used lextreme or lenhance, the former has a tighter bandpass so blocks out more unwanted signal. One of the cheapest which performs well from reviews is the SVbony sv220. You'd need to find a way to integrate it into your imaging train though.

@alacant I believe has good experience with a UHC filter, and has used dslr bodies.

Edited September 18 by Elp

[WolfieGlos]

If you can't incorporate a 1.25" or 2" filter into the imaging train, then some filters are available as DSLR clip-ins that sit directly over the sensor. The l'enhance is one such example.

I have one, and it worked really well with my modded 800D.

[ONIKKINEN]

3 hours ago, imakebeer said:

no need for darks or biases with a 450D, so I am reliably informed

This is sort of a partial statement, and relies on you subtracting offset in calibration for this to be the case. Did you do this in the calibration phase with Siril? If your camera has a black level of 2048, then just write "=2048" in the bias and dark fields in the calibration tab.

If you want a headache to learn why only flats wont work, have a look at this: https://siril.org/tutorials/synthetic-biases/

[wesdon1]

@imakebeer Mate I strongly recommend you buy an L-Enhance or L-Extreme filter. I was astounded by the difference they make when imaging HA rich Emission nebula. I absolutely guarantee you’ll thank me for the tip! 
 

A little note, I currently am using my L-Extreme with my Astro modded 600D DSLR and I’m noticing my DSLR is struggling a little bit getting signal through the much narrower band pass L-Extreme, so go for the L-Enhance to begin with, it’s cheaper and works a dream with DSLR cameras. 
 

Clear Skies! 
 

Wes

Edited September 18 by wesdon1
Missed a bit!

[imakebeer]

6 hours ago, ONIKKINEN said:

This is sort of a partial statement, and relies on you subtracting offset in calibration for this to be the case. Did you do this in the calibration phase with Siril? If your camera has a black level of 2048, then just write "=2048" in the bias and dark fields in the calibration tab.

If you want a headache to learn why only flats wont work, have a look at this: https://siril.org/tutorials/synthetic-biases/

That's almost exactly correct, just that for the 450D it's 1024 (so @alacant tells me).

Thanks for the link, I'll check it out 👍

[imakebeer]

@Elp @WolfieGlos @wesdon1 Thanks for the suggestions, I'll check out those filters.

The SW150PDS has a 2" focuser to start with. I've also got a SW x0.9 Comms Corrector which might take a 2" filter, if not I could go for a Canon one (which might give me more flexibility in case I remove the CC).

Thanks 👍🔭

[michael8554]

Take Flats with the Canon set to AV mode.

The light source needs to be adjusted to give around 1sec exposures, to avoid shutter and Mirror artefacts.

Michael

[alacant]

2 hours ago, imakebeer said:

for the 450D it's 1024

Method for its determination here. I can however confirm 1024.
Note that this is the value for the offset (AKA bias). Do not enter any value or filename into the Use Dark field.
HTH

**EDIT: ... and you MUST remove the edge stacking aretefacts BEFORE attempting any processing.
 

Edited September 19 by alacant

[imakebeer]

2 hours ago, michael8554 said:

Take Flats with the Canon set to AV mode.

The light source needs to be adjusted to give around 1sec exposures, to avoid shutter and Mirror artefacts.

Michael

Thanks - yes, @alacant told me something similar, in fact he said exposure should actually be >2s (I assume for similar reasons).

I'm controlling everything via an ASIAIR Mini which it seems is unwilling to let me shoot in anything other than M mode. However, it does have an "Auto" setting for flats, which appears to adjust the exposure to get the histogram peak in the middle. In this case I ended up with 21 x 0.8s flats using the white T-shirt method. But I'm really struggling to find a way to lengthen the exposure - I need a thicker "T-shirt", as I found multiple ones left creases which I think showed through in the flats.

[imakebeer]

2 hours ago, alacant said:

**EDIT: ... and you MUST remove the edge stacking aretefacts BEFORE attempting any processing.

I've been having fun writing my own Siril scripts which to the best of my knowledge reproduce the steps you showed in the video you made for me (thank you again!)... For one thing, it's useful, say, if you want to compare all of the stacking algorithms to see if any of them make your rubbish image better (they didn't!)

By the way, a while back @ONIKKINEN recommended 2 pass registration as it selects the best image (rather than simply the first) as the reference image..... but then I stop so I can manually use the registration plot to get rid of the outliers (could be scripted if we had consistent and general rejection rules for a variety of targets) before running a second script to finish off stacking.

Anyway, re. your comment above, once you've selected/rejected the best/worst you need to apply the registration - and in my case at least setting framing=min (in a script or interactively) seemed to do a good job of removing those edge artefacts, e.g.

seqapplyreg sequencename -framing=min

[Richard_]

It sounds like you got the answers in this thread. When using a DSLR, to try and tease out details of emission spectra (particularly Hydrogen alpha), you would need to remove the IR filter (which it sounds like you've done) and use a dual bandpass filter (like the l-extreme and L-enhance mentioned already).

These two actions would allow the H-alpha "pop" out more than light in the background and from stars. I think you may need to look at your calibration and stacking routine (I have not used Siril so can't offer you advice on its usage) as that will give you a good starting point before processing your image.

Any artifacts left in from incorrect calibration (eg dust motes, satellite trails) or artifacts introduced from incorrect calibration (eg stacking artifacts at the edge of your frame which has not been cropped) will make the processing more challenging and frustrating.

Keep at it, your framing and focusing look to be on point! 🙂

[imakebeer]

9 minutes ago, Richard_ said:

It sounds like you got the answers in this thread. When using a DSLR, to try and tease out details of emission spectra (particularly Hydrogen alpha), you would need to remove the IR filter (which it sounds like you've done) and use a dual bandpass filter (like the l-extreme and L-enhance mentioned already).

These two actions would allow the H-alpha "pop" out more than light in the background and from stars. I think you may need to look at your calibration and stacking routine (I have not used Siril so can't offer you advice on its usage) as that will give you a good starting point before processing your image.

Any artifacts left in from incorrect calibration (eg dust motes, satellite trails) or artifacts introduced from incorrect calibration (eg stacking artifacts at the edge of your frame which has not been cropped) will make the processing more challenging and frustrating.

Keep at it, your framing and focusing look to be on point! 🙂

Thanks, yeah, I've already been researching filters..... (though as far as I can work out I don't want/need one for galaxies as they are "broadband"). And of course the clip-in Canon ones have to be the more expensive ones don't they! 🤣

I can't convince myself that calibration is working perfectly, otherwise why aren't the dust motes being removed. I stack the flats then use the master flat to calibrate the lights individually - and even looking at individual light frames at this stage I can still see the dust, which doesn't seem right to me 🤔

I think I got lucky with the framing, I just had the camera in what I would regard as default orientation, then the ASIAIR app just lets me adjust to get the target centred 💪👍I took my time with the focus though - at this stage I'm just relying on Mk.I eyeball, hands and the FWHM plot. I like the idea of the ZWO EAF but at this stage it seems similar money for a filter will give me a bigger bang for my buck. For now I'll settle for a Bahtinov mask which is on order at the moment.

[WolfieGlos]

If you want to quickly check that the flats are working, you could always try shooting some bias frames (shutter as fast as possible, usually 1/4000th sec with the lens cap on) and using a script to stack it all, instead of manually stacking with the flats and "offset=1024". This way you won't make any errors when manually stacking

If the flats are still not removing the dust motes, then you know you have a problem with the flats. Should be a fairly quick experiment to do. Of course you could stack manually using the bias frames, but for tests I like to use the scripts for ease, partially due to laziness haha.

Personally, I would say getting things in focus is more beneficial than shooting with a filter which is out of focus! The tri/dual band filters will darken the background and lower the intensity of the stars, so using a bahtinov mask becomes harder unless you slew and focus on a very bright star. Before I had my EAF, I sometimes had to do this a few times a night - easy if a bright star is nearby. With the filter, your flats will also likely be a longer exposure time.

Correct about galaxies being broadband, so no filter required with a DSLR - at least I never used one. With an astrocam, you might need a UV/IR cut, but I think some come with this built in now? Further down the line, if you are imaging galaxies, spirals generally have Ha in them (see M33) so the dualband filter can be used to bring those out and really make the image "pop", using masking and layered onto the galaxy to create a HaRGB image. But I wouldn't worry about that for now, get the flats and focus right first

[imakebeer]

3 minutes ago, WolfieGlos said:

If you want to quickly check that the flats are working, you could always try shooting some bias frames (shutter as fast as possible, usually 1/4000th sec with the lens cap on) and using a script to stack it all, instead of manually stacking with the flats and "offset=1024". This way you won't make any errors when manually stacking

Not quite following you here - are you saying stack once using bias frames, then repeat using offset=1024 and then compare the two results? Or do you mean something else???

To be clear, I'm using scripts where I can - also very lazy! 😂 Just that I'm writing my own to give me more flexibility than the pre-supplied OSC script and also to incorporate the advice of the folks here. I think I'm at a point now where it's just the registration plot that needs to be investigated, interpreted and decisions made with me in the loop.

And thanks for the comments about filters, focusing and such - sensible and pragmatic advice 👍

[alacant]

1 hour ago, imakebeer said:

seemed to do a good job of removing those edge artefacts,

I know of no automatic method of completely removing all edge artefacts. IMO, best done with a generous manual crop after stacking. It takes only a few seconds.
Cheers

Edited September 19 by alacant
spel

[WolfieGlos]

1 hour ago, imakebeer said:

Not quite following you here - are you saying stack once using bias frames, then repeat using offset=1024 and then compare the two results? Or do you mean something else???

Sorry, I meant take a set of bias frames and use a script to stack the lights, flats and biases (and darks if using).

Then, once it has stacked, compare the stacked file that used the biases (by opening and stretching it), against the manually stacked file using the offset=1024 that you posted in the OP. Essentially, I'm suggesting a method that might rule out whether you've incorrectly manually stacked, and if that's why the flats aren't working  

[wesdon1]

8 hours ago, imakebeer said:

@Elp @WolfieGlos @wesdon1 Thanks for the suggestions, I'll check out those filters.

The SW150PDS has a 2" focuser to start with. I've also got a SW x0.9 Comms Corrector which might take a 2" filter, if not I could go for a Canon one (which might give me more flexibility in case I remove the CC).

Thanks 👍🔭

@imakebeer Right ok, well seeing as you are using the SW 150/6inch aperture, you might not struggle for signal with the L-Extreme due to much more light coming through, my refractor is only 66mm/2.5icnhes of aperture, so that could be why I'm struggling for strong signal. Still though to be safe, I'd go for the L-Enhance, it's brilliant mate.

If you're struggling to get a 2" filter in your imaging train, get a filter drawer, that should solve the problem? Firstly though maybe google the problem and see if anyone else has a workaround, before buying the filter drawer? Money is tight in this hobby! lol.

Best of luck!

Clear Skies, Wes.

[Elp]

3 minutes ago, wesdon1 said:

my refractor is only 66mm/2.5icnhes of aperture, so that could be why I'm struggling for strong signal.

It should still work. I often use my lextreme on my 50mm camera lenses.

[wesdon1]

13 minutes ago, Elp said:

It should still work. I often use my lextreme on my 50mm camera lenses.

@Elp Oohh ok, thanks for clarifying that for me Elp. I just assumed it was my small refractor struggling to gather enough light to penetrate the narrow band passes of L-Extreme. It's likely then due to my old 600D's low quantum efficiency?

[alacant]

3 hours ago, imakebeer said:

using bias frames, then repeat using offset=1024

I just checked a 450d and 1024 is correct. Here is a 1/4000s dark.

@imakebeer are you certain you have the viewfinder covered?

 

Edited September 19 by alacant