The "No EQ" DSO Challenge!

[Filroden]

Just now, The Admiral said:

Yes, a point worth remembering, though I've not been that diligent in this regards I must admit! I've been lucky in that the dust hasn't been that troublesome. It'd be worth doing a sensor clean at the beginning of an imaging session I'd have thought.

Ian

My Canon would go through the routine every time I turned it on, so there was no need to do it more frequently.

[The Admiral]

23 minutes ago, Stub Mandrel said:

I can't post the pic as it's taken on an EQ tripod ;'-)

I think that the Alt-Az vigilantes would allow it, just this once mind

Ian

[Filroden]

1 minute ago, The Admiral said:

I think that the Alt-Az vigilantes would allow it, just this once mind

Ian

And the forewarning has given me time to sharpen my stake!

[The Admiral]

1 minute ago, Filroden said:

My Canon would go through the routine every time I turned it on, so there was no need to do it more frequently.

I meant, if sensor cleaning had been switched off. Otherwise, I agree there'd be no point in doing another.

Ian

[Filroden]

NGC2239 - initial integration

These 6 images (a screenshot of my Pixinsight workspace) shows the results of initial calibration, alignment and integration. They have been auto-stretched by Pixinsight. Starting clockwise from the top left corner:

H_mono: integration of 35 x 30s and 22 x 60s, aligned with the first Ha image in the set

R: integration of 33 x 15s and 10 x 30s, aligned to an L image

L: integration of 77 x 15s and 9 x 30s, aligned to an L image

H: integration of 35 x 30s and 22 x 60s, aligned to an L image

G: integration of 39 x 15s and 10 x 30s, aligned to an L image

B: integration of 30 x 15s and 10 x 30s, aligned to an L image

The last five images will need a lot of cropping as the images were collected on two nights over quite a time span and have therefore suffered a lot of rotation. I created the first image so I could create as wide a field for a mono Ha image.

Perhaps unfairly, the Ha images really stand out. I think this is for two reasons: I collected more data, the narrowness of the filter means the image has less background signal so it can be auto-stretched harder by Pixinsight (the histogram between the Ha and L is radically different, with the Ha data being confined to a tight peak). I think I will have to synthesise a super L from the L, R, G and B data and even then, I think I will be throwing the L away and using just the Ha data. I just hope the colour data doesn't add too much noise!

[Herzy]

21 hours ago, Filroden said:

My first Ha sub. A single sub of 60s at 300 gain, 50 offset and -20C. No calibration, just a quick levels and curves stretch and no noise reduction. Going to go lie down now...

Actually, I've set my first sequence to be 30 x 60s Ha, 10 x 30s RGB and 30 x 30s L. Hopefully I can repeat that sequence a few times to add to my original data.

What camera are you using? That's incredible. It looks better than several hours of normal broadband imaging from my house, and it's only a single sub.

[Filroden]

26 minutes ago, Herzy said:

What camera are you using? That's incredible. It looks better than several hours of normal broadband imaging from my house, and it's only a single sub.

My ZWO ASI1600MM-C

[Filroden]

I have something odd going on and it may be a result of calibrating my TV with the Spyder. I process a mono image and export it to a TIFF. When I open it in Photoshop it's like the exposure has been cranked up 2 or 3 stops undoing all my lovely noise reduction!

[The Admiral]

At the end of November last year I had a go at M36 and M38. I tried processing in ST but I found the results quite messy, too many stars really. Anyway, I thought I'd have a go using AstroArt to process. After all, although I don't really know how to get the best out of an image with the software, at least with star fields it's not a question of teasing out feint details.

M36. I stacked 60 x 30s at ISO400

 

M38 (with NGC 1907 I think). 70 x 30s, ISO400

Ian

 

[Nigel G]

Ken, that's Awesome   

Nige.

[Nigel G]

@Filroden Ken is the taken using your Esprit 80ED ?

I'm on the hunt for a new scope.

Nige.

[Filroden]

14 minutes ago, Nigel G said:

@Filroden Ken is the taken using your Esprit 80ED ?

I'm on the hunt for a new scope.

Nige.

It is. It's got a lovely field of view with it's 400mm focal length and it's a very solid scope with a great focuser. I have the matching field flattener too (not that you'd notice given how badly tilted/spaced my imaging system is at the moment).

[Filroden]

NGC2239 in Ha

Skywatcher Esprit 80, ZWO ASI1600MM-C and Astrodon Ha filter

-20C, 300 gain and 50 offset

Integration of 35 x 30s and 22 x 60s and processed in Pixinsight and Photoshop

Version 1: uncalibrated, no background removal, no noise reduction and just a simple histogram stretch using levels and curves

Version 2: calibrated with bias, darks and flats, background removal, noise reduction, histogram stretch using levels and curves, further noise reduction, localised histogram balancing to enhance dark structure

In all honesty, the first version is so much better for not having me process it! I have no idea what I've done differently this time compared to my earlier images, but I just can't control the background in this.

[Stub Mandrel]

50 minutes ago, Filroden said:

In all honesty, the first version is so much better for not having me process it! I have no idea what I've done differently this time compared to my earlier images, but I just can't control the background in this.

I like it, but I'm still not entirely sure what 'faults' I should be looking for sometimes :-)

[Filroden]

3 minutes ago, Stub Mandrel said:

I'm still not entirely sure what 'faults' I should be looking for

The main thing that is bugging me is the harsh graduation from background to nebula in the second version. I think I might have pushed the processing too hard as the Ha image in my colour version is nowhere near as stark.

[Stub Mandrel]

I see, the first one does feel more 'natural'.

What I do when I have two images that both have good points, is put one as a layer on top of the other in Photoshop and use modes (e.g. luminosity if one is sharp and clear and the other is colourful) and/or use different percentage transparency to see if I can get the effect I like.

[Filroden]

I think I'm narrowing down my problem moving from PixInsight to Photoshop. The exported TIFF is being brightened to extremes and using curves to bring it back is causing my noise to shoot up and losing the natural transition. I need to check my setting as I think it's related to the image being mono.

[rotatux]

1 hour ago, Filroden said:

In all honesty, the first version is so much better for not having me process it! I have no idea what I've done differently this time compared to my earlier images, but I just can't control the background in this.

I prefer the 2nd version, being more contrasty especially in faint zones and details (such as dust clouds "digits" around the center).

Maybe you have applied too much noise reduction: the background is almost flat, when I would expect some grain... or is the jpeg.

BTW about your TV calibration problem: is the Spyder actually calibrating the driver / or graphics card in your Windows PC, or the TV itself ? My TV is calibrated (though I don't use it for PC) but it's a builtin feature, not done with an external device. Conversely, calibrating my PCs involved selecting an existing ICC color profile by eye and then doing minor adjustments, but it acts on the color-lookup-tables in the graphic card's rendering stage (CRTC for VGA/D-SUB or whatever it is for HDMI/DVI).

[Filroden]

46 minutes ago, rotatux said:

BTW about your TV calibration problem: is the Spyder actually calibrating the driver / or graphics card in your Windows PC, or the TV itself ?

The Spyder software has me adjust the settings on the TV. So I had to adjust the LED Backlight setting (much brighter). It then went on to calibrate colours and adjusted the driver which turned the image a very warm red so I reset it. I've now deleted the adjusted profile and reset my TV settings back to default. I need to start again!

Anyway, here's a new version of the Mono Ha image. I found my problem - it was sat between the keyboard and the chair! I had changed a setting in Photoshop which meant I was applying a different correction to the one I though. Much happier with this one.

This has had background removed, initial noise reduction on the background, levels stretch, more noise reduction in the background, sharpening and some localised contrast enhancement. I then moved into Lightroom and added a little more clarity and adjusted the black and white points.

And here's my attempt at a colour version. This uses the above Ha image for luminance (I discarded my other luminance data as it made the image noticeably worse) combined with colour data using a blend of R + Ha, G & B. I now understand the comment that Ha images can take on a salmon pink colour! I did try a HaRGB without adding the Ha into the R colour channel but it also didn't look as nice.

And for comparison, here's the Rosette Nebula I processed using the earlier data (i.e. not including last night's data).

Edited January 24, 2017 by Filroden

[Filroden]

And here's 15 minutes of LRGB data on M81/82:

[Nigel G]

2 hours ago, Filroden said:

The Spyder software has me adjust the settings on the TV. So I had to adjust the LED Backlight setting (much brighter). It then went on to calibrate colours and adjusted the driver which turned the image a very warm red so I reset it. I've now deleted the adjusted profile and reset my TV settings back to default. I need to start again!

Anyway, here's a new version of the Mono Ha image. I found my problem - it was sat between the keyboard and the chair! I had changed a setting in Photoshop which meant I was applying a different correction to the one I though. Much happier with this one.

This has had background removed, initial noise reduction on the background, levels stretch, more noise reduction in the background, sharpening and some localised contrast enhancement. I then moved into Lightroom and added a little more clarity and adjusted the black and white points.

And here's my attempt at a colour version. This uses the above Ha image for luminance (I discarded my other luminance data as it made the image noticeably worse) combined with colour data using a blend of R + Ha, G & B. I now understand the comment that Ha images can take on a salmon pink colour! I did try a HaRGB without adding the Ha into the R colour channel but it also didn't look as nice.

 

 

Wow stunning . Blooming stunning.

Well done Ken.

Cheers

Nige.

[Filroden]

2 minutes ago, Nigel G said:

Wow stunning . Blooming stunning.

Well done Ken.

Cheers

Nige.

Thank you! Under an hour of imaging and about six hours of processing  But it's now my favourite image that I've taken though I'm torn between the mono and colour versions.

[The Admiral]

10 hours ago, Filroden said:

Anyway, here's a new version of the Mono Ha image. This has had background removed, initial noise reduction on the background, levels stretch, more noise reduction in the background, sharpening and some localised contrast enhancement. I then moved into Lightroom and added a little more clarity and adjusted the black and white points.

And here's my attempt at a colour version. This uses the above Ha image for luminance (I discarded my other luminance data as it made the image noticeably worse) combined with colour data using a blend of R + Ha, G & B. I now understand the comment that Ha images can take on a salmon pink colour! I did try a HaRGB without adding the Ha into the R colour channel but it also didn't look as nice.

 

That's very nice Ken, and I'm amazed at what you've done with such a short exposure. I'm not sure whether I prefer the monochrome version; to me the colour one looks a bit flat colour-wise. Do you think a greater exposure in RGB would give more colour variation, or would one need to use other narrow-bands as well and colour map? Or is that just a natural consequence of using Ha?

Ian

[Filroden]

2 hours ago, The Admiral said:

That's very nice Ken, and I'm amazed at what you've done with such a short exposure.

Thanks Ian. I think it helps that there is so little background in the narrowband. It cuts through light pollution, etc like butter, so what little I have (about 40 mins) is much easier to stretch than the equivalent L data. There is a definite point on the histogram where it transitions from black into signal. When I process L data I have to cut into the data a little (possibly clipping up to 2,000 pixels or less than 0.001% of the total data) but I can get a darker background clipping no or only a handful of pixels with the Ha data.

2 hours ago, The Admiral said:

I'm not sure whether I prefer the monochrome version; to me the colour one looks a bit flat colour-wise. Do you think a greater exposure in RGB would give more colour variation, or would one need to use other narrow-bands as well and colour map? Or is that just a natural consequence of using Ha?

I have so little RGB (less than the Ha for each channel) and it's weaker (see comment above about having to remove backgrounds/stretch/clip). There was enough to colour the stars which I'm pleased with. The star size comes from the Ha data, which gives beautifully small stars (ignoring my eggy stars in the corners) but it's actually taken colour from the RGB data. I think I need a lot more RGB and I need to push it harder. If I get time I may reprocess the colour again and take the RGB data into Photoshop where I can boost it harder without affecting star colours. I might also try reducing the contribution of the Ha into the red channel. I think because I've used the Ha for the luminance and it dominates the colour, they have matched 1:1, giving that flat feel.

My plan now is to relocate my scope to the other end of the garden so I can avoid the neighbour's steam vent and gain an extra hour or so before the target goes behind the house. I would like to add to the Ha (maybe 80 more minutes) so I can get more from the background, but I need to match that with at least an hour each for RGB - more than I can achieve in a single night so I may have to collect the data in chunks of 20 minutes per channel. It also means getting a power extension cable and moving the laptop outside (which then means I have to follow it)

Anyway, for now I'm going to resist reprocessing today as there forecast suggests a chance of some clear skies over the next couple of days. Instead, I have 15 mins of Ha data on M81/82 that I didn't use and which shows some extensions in M82. I foolishly didn't read the tutorial enough and in my eagerness to process something yesterday, I took my LRGB non-linear too quickly before blending the Ha and then I got impatient to move onto the Rosette data!

Edited January 25, 2017 by Filroden

[Stub Mandrel]

Saw your posts giving me special permission to sneak in an EQ rosette. it's widefield so much smaller and not particularly brilliant (bear in mid done with an ancient 10D that cost me £30 off ebay). I couldn't find a raw stack but I was able to salvage some not-too processed layers from unflattened tiffs and tried my 'new superpowers' on it. The Christmas Tree is here too, but the cone is invisible:

It does suggest that the background of the Rosette isn't very 'black'