Witch Head Nebula

[MarkS]

Here's a quick Witch Head Nebula taken on 25-Nov.

Just over an hour's worth of subs in 30 second exposures.  Sony A7S on Tak Epsilon.  It was this image that gave me the data for the cosmic superhighway:   http://forum.orpington-astronomy.org.uk/index.php?topic=10910.0

It was the pixel rejection map that made me realise something "interesting" was going on:



Anyway here's the final result but it desperately needs more data:



The "beautiful" coloured artifacts come from the nearby star Rigel.  I thought Altinak (the bane of the Horsehead Nebula) was bad enough but Rigel really is something else 

Here's the larger version:
http://www.markshelley.co.uk/Astronomy/2016/WitchHead20161125.jpg

Mark

[Carole]

Well done Mark, I gather this is a very faint nebula and also a very large FOV. 

Did attempt it once with the QHY8L but it was when I was having trouble with the software so never got a usable result.

Carole 

[MarkS]

Well done Mark, I gather this is a very faint nebula and also a very large FOV. 

Yes it's big - the part I've shown in this image is 2 degrees from top to bottom. It's also very faint - but nowhere near as faint as the clouds of molecular dust I've been imaging recently!

The big problem for both this image and the especially my dust cloud images is the background extraction.  I've invented an approach using Photoshop CC which allows you to "paint away" residual background gradients with different size brushes while the data is still linear - i.e. straight after stacking and before applying the tone curve.

The famous PixInsight DBE (Dynamic Background Extraction) comes completely unstuck on most of my gradients because there is no background sky showing behind the dust, on which to place sample points. 

I'll document this technique when I get a chance. 

For noise reduction I now use the Camera Raw Filter which can be applied from within Photoshop.  It has sliders for chroma noise and luminance noise and works completely intuitively.  It seems to work better on noise that looks random.  The noise in a Bayer drizzled image looks more random and less "blobby" than the noise you get when stacking the usual debayered images.  So Bayer drizzle is now an important part of my processing workflow.

So I take back some of the rude things I've said about Photoshop in the past but I do still maintain that Photoshop curves is a most excellent way of removing colour from stars.

Mark

[Carole]

I'll document this technique when I get a chance. 

I look forward to seeing this.

So I take back some of the rude things I've said about Photoshop in the past but I do still maintain that Photoshop curves is a most excellent way of removing colour from stars.

Yes I agree that is a problem and one I still struggle with.  I have recently started an experimental method of trying to counteract this and that is to take a copy of the image before stretching too much and increasing the colour for the stars BEFORE doing a bit of a stretch, and then trying to layer/ or select stars back into the final or nearly final image.  I tried is on one experimental image and it worked OK, but haven't done any LRGB since then to experiment further.  I got the idea from Jaspalchadra who gets very good star colour and although he doesn't explain his method intimated it was to do with combining layers.

Carole

[JohnP]

nice start Mark... Looks like there is a tad more nebula up top as well.. Must be difficult to tame the star top right... look forward to you getting a chance to finish this.. John

[Fay]

I really like this Mark

Fay

[MarkS]

I'll document this technique when I get a chance. 

I look forward to seeing this.

I was just starting work on my tutorial and decided to base it on Photoshop CS2 because it is likely that more people would have Photoshop CS2 than the monthly subscription Photoshop CC that I use.

I now realise the technique won't work at all in CS2, so I've no idea which versions it might (or might not) work in.

Here are the things that I regularly do in CC that won't work in CS2:

I convert the image to 32bit mode and work in 32bit mode because I then get a non-posterised interactive preview even at small scale e.g. 33%.  The trouble is, in CS2 you can't use Levels and Curves layers with a 32bit image - they are grayed out.

With CC, the histogram (on the Navigator/Info/Histogram pane) shows the 3 colour channels overlapped in a single view.  So I can interactively select areas of image background and immediately see which colour dominates and what adjustments are therefore required to neutralise it.

With CC, the curves tool displays the histogram and overlays the individual curves applied to each R,G,B channel on that histogram.

In CS2, the Camera Raw filter is not available from the Filter menu.  The Camera Raw filter is a powerful tool.

So I take back some of the rude things I've said about Photoshop in the past but I do still maintain that Photoshop curves is a most excellent way of removing colour from stars.

I've just added more rude things about Photoshop - the earlier versions anyway!  Photoshop CS2 is more or less unusable.

Mark

[Fay]

I was wondering whether to change to the monthly PSLightroom. I have CS5. Thing is, I have a lot of plugins on my present version and am wondering if I would be able to move them over to a newr version

Fay

[Carole]

Thanks for the info Mark, I use CS3 so might be better on there.

Carole

[MarkS]

I use a variation of Roger Clark's technique here:
http://www.clarkvision.com/articles/astrophotography.image.processing2/
See the section on "Light Pollution and Airglow Removal"

However, instead of selecting an area of the image and applying curves to it, I add a curves adjustment layer.  The RGB levels can be set to cancel out a particular coloured part of the gradient and then you apply the correction by using the brush on the layer mask.  The size and opacity of the brush can be altered and you can apply multiple brush strokes with various sizes and opacities - start with a large brush and low opacity and work incrementally with successive smaller brushes with higher opacity.

So if your image has an background area of say green and another area of purple you can add one curves layer to subtract green and another curves layer to subtract purple.  It's worth also having a temporary layer right at the top to apply strong colour saturation to make it easier to see the effect of what you are doing.

With the selection tool you can go round selecting small areas of background and their histogram automatically appears in the Histogram window - this allows you to check how close you are getting to black background without clipping.

To prevent the annoying Photoshop behaviour of posterising the image when you add an adjustment layer then first convert the base image layer to 32bit mode. Once you finish you can flatten the layers and then convert back to 16bit mode - when the HDR Toning dialog appears, simply select "Exposure and Gamma" as the Method.

The above approach will work on an image that's already stretched but it's not a mathematically correct approach for subtracting a colour cast or colour gradient.  However, most of the time you won't notice any colour shifts caused by doing it this way.

Myself, I prefer to work on the original linear data.  The technique to do that is quite a bit more complicated and I'll describe it another time.  However the essence is to add a number of levels layers to boost the intensity and to remove most of the background glow - the bright parts of the image will be saturated to death but that doesn't matter.  Then add adjustment layers (preferably colour fill layers set to subtract) to deal with the colour gradients visible in the background - again by using brushes on the layer mask.  Then add further levels layers to reduce the image intensity back to where it was to start with.

Mark