Crescent Nebula DSLR Woes

[TiffsAndAstro]

4 minutes ago, Elp said:

The issue with using screens is that their colour balance or natural tone and even refresh rate may cause issues, some don't even display a uniform white from edge to edge.

im guessing my tablet screen is a decent ips and with the tshirt cloth its at least diffuse. what its possible replacement tracing panel is like i dunno, but ill give it a try

[imakebeer]

1 hour ago, TiffsAndAstro said:

No need for credit card for flats

 

1 hour ago, TiffsAndAstro said:

a tablet

I think a credit card might be needed if I am to acquire a tablet big enough to stretch across a SW150PDS 😜

Also, having just ditched the laptop and cables for an ASIAIR I'm reluctant to rush to use NINA or more softwares than absolutely necessary (notwithstanding the very reasonable argument that the ASIAIR alone is not exactly working out for me at the moment! 😖😂)

[Elp]

3 minutes ago, imakebeer said:

think a credit card might be needed if I am to acquire a tablet big enough to stretch across a SW150PDS

Maybe not, my A4 led tracing panel fits over my C6 perfectly fine. Less than 10 it cost.

Edited October 10 by Elp

[imakebeer]

5 minutes ago, Elp said:

Maybe not, my A4 led tracing panel fits over my C6 perfectly fine. Less than 10 it cost.

When @TiffsAndAstro said tablet I understood large Android device or iPad type thing. But LED tracing panel sounds somewhat more cost effective 👍

[Elp]

Just now, imakebeer said:

When @TiffsAndAstro said tablet I understood large Android device or iPad type thing. But LED tracing panel sounds somewhat more cost effective 👍

Think they meant if you already had one at hand, worth trying out.

[TiffsAndAstro]

43 minutes ago, Elp said:

Think they meant if you already had one at hand, worth trying out.

Yeah definitely do not rush out and buy a tablet just to use as a light for flats  But if you have one lying around that can cover your big scope try it.

i also have an A4 sized tracing backlight thing on the way. fingers crossed its good enough. 

[TiffsAndAstro]

1 hour ago, imakebeer said:

 

I think a credit card might be needed if I am to acquire a tablet big enough to stretch across a SW150PDS 😜

Also, having just ditched the laptop and cables for an ASIAIR I'm reluctant to rush to use NINA or more softwares than absolutely necessary (notwithstanding the very reasonable argument that the ASIAIR alone is not exactly working out for me at the moment! 😖😂)

Also I'd want to stick with asiair there must be a solution as plenty of people use it to take great images.

[imakebeer]

2 hours ago, TiffsAndAstro said:

i also have an A4 sized tracing backlight thing on the way. fingers crossed its good enough

Do you have a link? I assume one might find such a curio on Amazon? 🤔

[ONIKKINEN]

Tracing panels can be hit and miss, but at their cost its a worthy gamble.

I have 2 A4 sized panels, both of which prouduced decent flats at first, but the first one developed some fault in some of the LEDs and started producing uneven illumination. The second one has a terrible USB port and only works if i hold the cable in. I also have an A3 sized tracing panel that wont produce good flats due to uneven illumination (not obvious by eye, but flats are bad anyway). So just FYI its a bit of a lottery with them and you might just get a bad copy.

[TiffsAndAstro]

2 hours ago, imakebeer said:

Do you have a link? I assume one might find such a curio on Amazon? 🤔

er i just found one on ebay that looked ok and was in the uk for faster delivery. 8.95 earth pounds 

once i can try it out ill post results. how good can it be for 8.95 lol? 

https://www.ebay.co.uk/itm/404690054580

 

Edited October 10 by TiffsAndAstro

[imakebeer]

JFYI, this is what I managed to get so far for M31 - don't laugh! (it's ok really, you can laugh a little bit)

This is my first ever attempt at a mosaic, and I realise a NB filter isn't ideal for a broadband target!

Only 2 panels, 40 x 3min = 2hrs per panel

I took some new light frames (x20) this morning but had to use an offset of 4096 in Siril as 1024 just doesn't seem to work with this target so far.

Not dreadful but clearly needs work - through inexperience I just manually blended the two frames together in GIMP and did my best to try and match the colours. Didn't really stretch it much, just adjusted the black point and did a bit of layer masking to tidy up the noisy background.

I feel like maybe the flat frames are all over the place so perhaps an LED tracing panel is needed to try and bring some consistency to proceedings.

[TiffsAndAstro]

Very cool. No sign of the join and nice colour

I'm trying same thing trying to get up to 6 on each panel

Also you seem to have multiple nebula which looks great. I don't have any

Edited October 10 by TiffsAndAstro

[imakebeer]

OK, strap in and get comfy, I think I might have finally figured what is happening with my light frames, flats and offsets in Siril - even if I can't entirely explain why!

Fundamentally I think it has to do with conversion from 14-bit to 16-bits (which involves multiplying by 2^2 = 4)

I ran some tests last night using Siril scripts to ensure my steps are repeatable. I looked at all 8 possible combinations of the following 3 pairs of variables:

1. The RAW 14-bit *.cr2 files from my Canon 450D vs. the 16-bit FIT files from my ASIAIR Mini
2. A hardcoded offset in Siril of 1024 vs. 4096 (i.e. 4 x 1024)
3. Flats with 15s vs. 2s exposures (now using an LED panel!)

What I found is the following:

1. The result didn't really vary with the different flats. I believe ideally we want a level about 45-50% of full well depth, being careful to note whether we're working in 14 or 16-bits. Best guess is that as long as the histogram peak isn't right up against the min or max then it's not too critical???
2. When I used the RAW Canon *.cr2 files I got good results with an offset of 1024 (but an almost entirely black image with an offset of 4096)
3. But with the 16-bit FIT files from the ASIAIR I got good results with an offset of 4096 (but garbage with an offset of 1024)
4. Both "good results" in (2) & (3) above produced nigh on indistinguishable images!

What I still can't figure out:

1. Before processing I think Siril converts the RAW 14-bit *.cr2 files to a 16-bit FIT file. So how come this works with an offset of 1024 but not 4096 - why don't Siril 16-bit FITs behave the same as ASIAIR 16-bit FITs?
2. Whether any of this behaviour can be configured in the ASIAIR (probably not!) or Siril preferences (maybe???)
3. How come I managed to get a decent image of the Crescent Nebula in spite of all this - maybe I just got lucky???

Next steps:

• More data - in every sense!
• I repeated the above on both panels of my M31 2-panel mosaic and it was consistent, but what about other targets?
• Currently I'm using a L-eNhance filter - what happens if I fit my L-Pro, or no filter at all, is it still consistent?

[TiffsAndAstro]

best of luck trying to resolve the issue.

not very helpful for you, but i used my stock 600d for a few months and it seemed ok via nina stacking lights, flats (manual wasn't using wizard back then) bias and darks in siril. probably not as good as i thought at the time though.

[imakebeer]

48 minutes ago, TiffsAndAstro said:

but i used my stock 600d for a few months and it seemed ok via nina stacking lights, flats (manual wasn't using wizard back then) bias and darks in siril

I think the key point is what the ASIAIR is doing to RAW *.cr2 files before it spits them out as FITS files - so I think one would only run into this issue if using an ASIAIR, and even then (so my hypothesis goes) it's a question of understanding what it's doing and dealing with the output accordingly 🤞

[alacant]

OK.
This should help you eliminate most of the unknowns. I'm using Linux (as in the air) throughout.
Here are three .cr2 files taken in darkness. Immediately afterwards I took the same but checked that EKOS convert them to fits. The latter you have already seen that the offset is ~1024. 
I roll back you your version of Siril and convert the .cr2 to fit. The offset is also ~1024.

Conclusion: the air is doing something to the cr2 after it arrives from the camera.

HTH

Edited Tuesday at 15:00 by alacant

[imakebeer]

8 minutes ago, alacant said:

Conclusion: the air is doing something to the cr2 after it arrives from the camera.

Fundamentally yes, I have the same conclusion. I think I tested that before with bias frames at 1/4000s in darkness:

• The RAW cr2 has an offset of 1024
• The ASIAIR converts it from 14 to 16 bit and the offset is 4096 (x4 the 14-bit value)
• But if we let Siril convert the cr2 to FIT file (which as far as I can tell is also 16-bit) it still has an offset of 1024 (this is the bit I can't figure out)

[alacant]

22 minutes ago, imakebeer said:

But if we let Siril convert the cr2 to FIT file (which as far as I can tell is also 16-bit)

No. Take any individual cr2 file which has been converted to fit IN SIRIL. Siril retains the 14 bit (Max value ~16000). The STACK may well be more than this of course.
Bear in mind I have no air anywhere.

Edited Tuesday at 15:37 by alacant

[Elp]

Have you tried converting the cr2s to Tif using Canons Digital Photo Professional first? I always used to to this prior when using my Canon. But thats not resolving the air output.

[imakebeer]

3 hours ago, alacant said:

Siril retains the 14 bit (Max value ~16000)

Really? The 2 screenshots below are from the preview in Siril's load dialogue - exact same flat file, one is the original CR2, the other was converted by Siril to FIT. But it could be wrong. In any case I think it's fine since I think I've figured out how to manage my data now 🤞

2 hours ago, Elp said:

Have you tried converting the cr2s to Tif using Canons Digital Photo Professional first? I always used to to this prior when using my Canon. But thats not resolving the air output.

Thanks but not really necessary I think dude - now I understand (I think! ) what the ASIAIR is doing there's nothing to resolve as such. Even though I got there by 2 different routes, I was astounded how both methods yielded almost identical results, I mean the difference was really splitting hairs.

[alacant]

So that other air users don't suffer, perhaps you could summarise how to process DSLR images using the same?

Cheers 

[imakebeer]

15 minutes ago, alacant said:

So that other air users don't suffer, perhaps you could summarise how to process DSLR images using the same?

Cheers 

The Canon 450D produces 14-bit RAW *.cr2 files and has a hardcoded offset of 1024. The ASIAIR Mini appears to convert these to 16-bit FIT files.

If you're shooting with an ASIAIR Mini (I would assume other ASIAIR types behave the same) and want to process your images in Siril you have 2 choices:

1. Work with the original *.cr2 files and let Siril convert them to FIT format for processing - in this case use a fixed bias settting of 1024 in Siril. (You also need to apply the setting in the ASIAIR to save the RAW *.cr2 files to the camera's SD card)
2. Work with the FIT files from the ASIAIR - in this case use a fixed bias settting of 4096 in Siril (i.e. 4 x 1024 = 4096, because to convert from 14 to 16-bits you multiply by 2^2=4)

One advantage of the ASIAIR FIT files is that they are sorted into separate directories. On the other hand purists may prefer to work with the original *.cr2 files but they will need to be sorted manually. My experience so far is that provided you use the right bias setting in Siril the result is nigh on identical.