What Next?

[fatwoul]

This probably belongs in the beginner section, but I'm feeling pleased with myself for the image, so forgive my arrogance asking here.

I've shoved 130 lights, and a handful of darks and flats through DSS, to get this:

This is the unaltered image given to me, together with these settings:

I'm getting really confused how to get the best out of the image, though. I've made some progress with the RGBK settings, but I'm really struggling with the Luminance, which is embarrassing considering my occupation.

[Demonperformer]

Well, I'm no expert, but my routine tends to be:

Saturation up to 20%, use the centre triangles on RGB to align the three histograms, use the lower midtone slider to move the line so it slices through the 'climb' on the resulting column.

This has been arrived at a bit through trial and error and a bit through including the occasional comment I have read on here. Once I have achieved that, I then save the result under a different name (usually use suffixes -1, -2, -3, etc) and start 'tweaking', saving each change that makes what may pass for an improvement. Then once I finish with DSS, I compare the results and select the best one to go into (in my case) GIMP.

You say you have used flats, but there still seems to be quite a bit of vignetting at the corners. Don't know enough about flats to offer an explanation/solution to that one.

HTH

Nice result so far by the way.

[fatwoul]

Thanks so much, Dp, I will give this a go today.

Yeah, I seem to get massive vignettes. Usually it doesn't matter too much, as my subject is small enough that I crop. The only explanation I can offer is that I am using a full-frame DSLR, so might be pushing the limits of the coverage area of the scope. The flats do seem to remove it a little, but nowhere near as much as I would like. (EDIT: another explanation is that I am screwing up the flats in some way, which is also very likely).

It is one of the list of problems I have to solve. I should write down the list - I think it would be really helpful, and encouraging to be able to tick them off...before writing others! All part of the fun, right?

[fatwoul]

OK, after fiddling around as per Dp's suggestions, and then tweaking further in PS, I have this:

full size image here

[ollypenrice]

Better!

Curious that the vignetting has persisted despite flats, though. Did you get the flats to peak about half way up the histogram? Also in an OSC camera you did debayer the flats? Maybe DSS does that automatically, I don't know it.

Not much star colour so I wonder if the white point is clipped.

Olly

[rfdesigner]

RE flats

Did you take a bias frame off the flats before using them?

i.e. flat = bright image of 1s long.

bias = dark image of 1s long (same length as the bright flats)

dark = dark image equal in length to the image subs.

Final flat = average(flats) - average(bias)

Final dark = average(darks)

Derek

[fatwoul]

Olly - if I did get them half way, then its by accident, as I wasn't really paying that much attention! I didn't debayer, and I don't know if DSS does either. I'm finally sitting down with all the RAWs from the autumn/winter and trying to make something out of them. I have a lot still to learn.

Derek - no. They were literally just flat exposures taken through a diffuser box I made last year, and then just shoved into DSS. I haven't done anything to them at all.

Good advice from both of you, of course. I will add it all to my list of things I have to figure out...

[rfdesigner]

Derek - no. They were literally just flat exposures taken through a diffuser box I made last year, and then just shoved into DSS. I haven't done anything to them at all.

Good, it's a simple fix then. If you remove the bias frames you may find the aparent vignetting disappears.

Derek

[ollypenrice]

Bias frames are the shortest exposures your camera can take. There is no need for other darks for flats. Take 50 bias, (or more), median combine them and treat this as a master dark-for-flats. That is what many of the imagers do on the forum. Don't subtract the bias twice.

Olly

[fatwoul]

Thanks again, Derek and Olly. I have some "historic" bias and flat frames from last year, but they were taken in the Autumn when it was much colder, so I'm thinking they shouldn't be used now? I'll make sure to rattle off 50 bias next session. At 11 frames a sec, and 1/8000th sec, 5 seconds work is hardly something I should be lazy about. Or should I not do them in continuous? Should they be seperated out? Is there some characteristic of DSLR chips (heating up or something) that might mean I should do them more slowly?

[fatwoul]

OK, I had another bash:

I may have overdone some of the post-production, but I am much happier with it, especially the colour that has survived.

[rfdesigner]

Oh yes, much flatter well done, looks like the colour balance is OK.. what did you do that made the difference?

Actually looking at the histogram, I think you can just drop the background a touch. (about 9 counts in the online 8 bit version)

Derek

[fatwoul]

Cheers Derek, I'm quite pleased. All I did was add a bunch of bias frames, and ran the whole show again through DSS. The result that time was a lot easier to work with, and following the advice here I was able to get more out of that result aswell. I darkened a little in PS (selectively darkening the "black" areas) upped the saturation a touch, and then unsharp masked a bit to make the 72dpi version look a bit more snappy.

You're right about dropping it a little further. I did all that last night on my monitor, which is in desperate need of calibration, but looking at it today on a different monitor it could do with being a little blacker.

Thanks again, everyone.

(PS - My POD arrived today!)

[rfdesigner]

If you darken it do look at the histogram, you shouldn't eat into the 'background curve' (I'm not actually sure what the name is for it).. but below this curve (normal distribution shape) there shouldn't be any information so it's fine to just subtract the offset, but don't eat into the normal distribution curve itself.

Derek

[fatwoul]

Too much?

[ollypenrice]

Too much?

Yes, just. That looks a bit clipped. I bet the foot of the peak on the left hand side looks sightly chopped off on the histogram? But what a transformation! In Ps if I click on the background sky I like to see a value of at least 15.

Olly

[Beyond_Vision]

Coming along really nicely, but as Olly says post no 11 looks better since the last is clipped.

Regards

Kevin

[RikM]

#11 is better. Looking very nice.

[rfdesigner]

yup, as per Olly

over clipped... When I suggested using the histogram I should have made up a pic (now included) to show what I meant.

Crop up to but not beyond the point marked with the point of the upturned V on the histogram X axis, what you've done in the latest image is to crop to the peak of the histogram.

Derek