Carina, 2nd attempt: help required!

[Atmos (Colin)]

Originally Posted by thegableguy

REMINDER TO ANYONE INTERESTED:

Here's the link to all the RAW files. 25 lights, 20 darks, 10 flats, 10 bias.

https://www.dropbox.com/sh/xrk77cpoi...Q39wYZdVa?dl=0

Would really love to see what other people manage with them.

Give me 8 days haha

[rustigsmed (Russell)]

just lights and flats, no darks or bias.

stretched the crap out of this i had a feeling you wanted something with a bit more oomph? this is going a bit too far but why not.
is the dslr stock? i think your version maybe a tad red for the brighter areas.

also unless it was my DSS setting there is either some sky glow or over exposure perhaps on the flats? there was alot of gradient going on in the background to the left of frame which i cropped out (and i think you might have too).

cheers

russ

edit: added something a little less nitro

[thegableguy (Chris)]

Originally Posted by rustigsmed

just lights and flats, no darks or bias.

stretched the crap out of this i had a feeling you wanted something with a bit more oomph? this is going a bit too far but why not.
is the dslr stock? i think your version maybe a tad red for the brighter areas.

also unless it was my DSS setting there is either some sky glow or over exposure perhaps on the flats? there was alot of gradient going on in the background to the left of frame which i cropped out (and i think you might have too).

cheers

russ

edit: added something a little less nitro

Wow!

Love them both but probably prefer the second one. I never even considered not using darks or bias but the difference is enormous. I often wondered how others got their amazing blanket of stars; now I know. Darks sure do take out a lot of the data.

Yep camera is stock. It's the cheapest Nikon DSLR they make, but we've got some vastly better bodies I want to try one day. I bought this with the suspicion I'll mod it one day. Did you use a lot of noise reduction, or was it not really necessary at ISO 800? It's pretty noisy at ISO 1600.

Agreed re the red; I kinda forced it in that direction because that's what most other images I could find of it looked like.

Thanks for doing that - really appreciate seeing what someone else does with the same data.

[rustigsmed (Russell)]

Originally Posted by thegableguy

Wow!

Love them both but probably prefer the second one. I never even considered not using darks or bias but the difference is enormous. I often wondered how others got their amazing blanket of stars; now I know. Darks sure do take out a lot of the data.

Yep camera is stock. It's the cheapest Nikon DSLR they make, but we've got some vastly better bodies I want to try one day. I bought this with the suspicion I'll mod it one day. Did you use a lot of noise reduction, or was it not really necessary at ISO 800? It's pretty noisy at ISO 1600.

Agreed re the red; I kinda forced it in that direction because that's what most other images I could find of it looked like.

Thanks for doing that - really appreciate seeing what someone else does with the same data.

no worries. its been months since i've taken any shots myself

yes definitely the first one is way too stretched ignore that one.
i read somewhere ages ago that you don't require bias frames for dslrs so i have never bothered (when using a dslr). Flats in my opinion are the most important one to get right (along with the lights )

if you've got darks go for it, i was in a rush so didn't bother downloading them (no nbn here) the extra stars are mostly from stretching the data further (rather than from not applying darks). saying that i stopped using darks even when using a dslr (most would disagree with me) but instead aimed for dithering between subs and a minimum of 16 light subs to further reduced noise (getting a library of darks at the right temp and lengths for a dslr was too time consuming for me). I didn't use any noise reduction mostly simple levels and curves (took a lot of iterations). i made a duplicate converted it to LAB mode, then extracted the luminance layer and put it over the original and blended via luminance. the bottom layer (colour) you can crank the saturation before blending the luminance layer in it seems to help with colour noise.

modding will help a lot with sensitivity and picking up the Ha.

cheers

[Atmos (Colin)]

I personally would recommend making a Master Bias of maybe 200 exposures. Takes maybe a minute to get them and having a good bias is great to have for both the flats and lights. Russell is correct that flats are the most important of the calibration frames, this is why I would suggest hitting these with a master bias along with the lights, can remove a fair bit of noise.

You can largely get away without darks by dithering between each frame as Russell suggests. Darks are mostly for removing hot pixels and slow dark current build up, with one 120s subs the build up is not too bad anyway.

It looks like you have some good data there. Realistically, it is going to take you a good year or so to start to get the hang of image processing. I've been slowly muddling my way through it for the last 9ish months and what I have most learned is just how much further I have to go haha

[Luke.]

Alright I had a go at it. I am only just starting to learn processing now.
I have no idea how to control the colours and I lost a lot of detail towards the center but its a start!

[thegableguy (Chris)]

This is all great. There's stuff I really like about that last one Luke - the fainter stuff really punches, but yeah somewhat at the expense of the finer details and brighter areas.

It would be a lengthy process to overlay two or more versions in Photoshop and layer them that way, but I'm guessing that's the best way to get the full dynamic range of everything you want to see. RAW images have great dynamic range for daytime photography, but I'm realising this is a whole other level.

Having said that, this is a pretty cheap nasty DSLR. Next time I take a shot at Carina I'll try with a better DSLR and see if there's any significant benefit.

Mostly the different versions are showing me that there's way more than one way to skin a cat and that there's no one right answer.

I'll keep slogging away! Thanks all who had a crack at it. I'm trying for the Leo Triplet tonight but I think the alignment is off; just went "meh - good enough" after an hour of fiddling around. Fingers crossed. Lovely clear skies we're enjoying at the moment.

[rcheshire (Rowland)]

It's a nice image and you may be expecting too much of the data given the total exposure time.

As it's a DSLR image the camera has applied a human eye response curve to the data which can make calibration a bit tricky at times.

The bias and flat frames are likely more nearly linear than the longer lights and darks.

The darks should improve the image. So rather than following the conventional method of calibration in which the bias is subtracted from the lights and darks you could try this as an alternative.

1. Subtract the master bias from the flats as normal. Don't subtract the master bias from the lights or the darks.

2. Subtract the master dark (which still has the bias in it) from the lights and divide by the master flat (with the bias removed)

That way you 'largely' avoid a problem that sometimes creeps in with DSLR data where the whole set of images may not be taken at the same temperature, further complicated by the corrections applied to the data by the camera software.

Colour calibration is much easier with a properly calibrated set.

I wrote a little utility for performing the task above but unless you are a Linux user it's not much use.

Hope this is helpful.

EDIT: had a go at processing on a laptop. My monitor has been seconded. Colours might be a bit off. Ignoring that because it's fixable, the image set calibrated well using the method above and stacked easily in Pixinsight, which is a good sign with DSLR data and usually means calibration didn't leave too many zeros. Finished off in StarTools.

The image has retained its nebulosity and a lot of small fine detail and the stars have colour.

You've done a good job of acquisition and more bias and dark frames would be good but overall there is a lot of detail.

If the camera is not modified attempting to increase red saturation will throw the image off. Best thing is to go for detail and then worry about colour.

Here's a dropbox link to the preprocessed 16bit tiff file