My first semi-serious attempt at a DSO target - looking for feedback!
Taken at Stockport on 3 May:
Telescope: Celestron 8SE SCT on single-arm fork alt-az mount (unguided)
DSLR: Sony Alpha NEX-3
Subs: 94 x 25s @ ISO1600 (with 25 darks, 50 flats & 50 bias)
Stacked in DSS, processed in Startools, touch-up in Paint.NET
I wanted at least twice as many subs, but the clouds closed in and abruptly ended the session! (I also took some nightscape subs before I went to prime focus, but haven't processed these yet).
What I'm trying to do is gradually learn the ropes of AP basics, before buying some serious kit (long-term goal is 5" APO, solid EQ mount, cooled CCD). I realise my current equipment is hardly ideal for AP, but Wayne Anderson has shown what can be achieved with similar gear and some talent + perseverance!
My key goal over the next few months is to learn the ropes on stacking and image processing (total neophyte currently), before I think about changing my kit.
After nearly 30 years as a visual-only guy, I've been having fun stepping into the somewhat daunting world of AP over the last few months.
P.S. Original FTS, straight out of DSS, is here, in case anyone wants to muck about with it and show me what can really be done! (33 MB zip [original is 109 MB FTS file]): https://dl.dropboxusercontent.com/u/...ationalFTS.zip
Last edited by Amaranthus; 05-05-2014 at 09:58 PM.
You've definitely got better data there than your current processed image shows.
I grabbed your fits file and had a quick play:
I opened it in DSS and saved out a 16 bit TIFF. From there I used PS to muck around. Since it was quick: auto tone, auto colour, auto contrast, manually adjusted levels, brightness and contrast, ran the HLVG plugin (removes a lot of green), little bit of vibrance/saturation, and "small" sharpness filter.
I think it's possible to do way better given time, effort, etc.
Good thing is you can always reprocess these things - I should know since I often end up with multiple versions.
Not a bad effort at all.
Some things to be aware of when processing in StarTools;
- get rid of any stacking artifacts (you have a few at the borders of the image)
- don't do any pre-processing and keep the data as virgin as possible (to me it appears the data has, at the very least, been color calibrated, while there is also a suspicious lack of background bias.
- i can't be 100% certain but it looks to me like the that the data was pre-filtered or otherwise 'prettied-up' (read: destroyed) before it made it to DSS. You're not using anything other than your stacking solution to convert your RAWs, right? *DO NOT* use any of the software that came with your camera!
I was pretty sure I'd turned off all pre-filtering in my DSLR. I checked back through the settings, and the only ones that *might* have done something unexpected to the data is the white balance (left on auto) and the "High ISO NR" (the options are 'auto' and 'weak' - I left on auto). I did turn off the "Long Exposure NR", so that can't be it. The camera also has the colour options "sRGB" and "AdobeRGB" - I left it on sRGB for want of knowledge on which was better. Nothing else I could see that the camera was doing - all other corrections are definitely off.
I must admit to flying somewhat blind in DSS. I read the manual, FAQ and tech details, and watch some vid tutorials, so I'm okay now at running the program and not making gross errors. However, it remains difficult to know what might be the best register and stacking parameters to use for my RAW data. (I definitely avoided doing any post-processing of the image in DSS, so what you see in the FTS file is straight after the stacking).
Does anyone have advice on what might be decent 'default' parameters for DSS, or else what tweaks are typically highly sensitive or things to avoid? I dug around about in the IIS forums but couldn't find a thread that really covered this (might have missed it). I may well have ticked the RGB color calibration box, for instance! I figure I should work on getting DSS as optimal as possible before I start getting serious about the Startools processing (which is why I hadn't bugged you directly yet, Ivo ). I guess you are saying that I should let DSS do no corrections except those required for stacking - would this include no hot pixel removal etc.?
I did stack via a custom area (I got rid of the margins of the reference image) and junked the worst 14 images that had obvious blurring or got bad scores, but I guess this wasn't sufficient to remove some stacking artifacts.
I tick hot pixel removal, but I haven't done a comparison to see how it compares to unticked. Most parameters are default that I use. I wind the star detection slider a bit to the right to speed things up a bit.
Your image is pretty good for an early attempt. It's just a matter of tweaking things. I've seen a few full spectrum modified Nex3's pop up on ebay and was wondering what they were like.
Thanks Ivo. I'll have another go at the data in DSS tonight, with the guiding goal of trying to get DSS to do as little as possible to the data short of the alignment and stacking! (I'll send a screenshot of my selected settings).
For the stacking, is the 'median' method preferred other other options (e.g. averaging, or the various bias/variance-correction algorithms?). Is 'drizzle' worthwhile for sharpening the fine detail? (I've found it can crash DSS regularly if the area is not pre-clipped, so I guess this is the major constraint!)
Last edited by Amaranthus; 05-05-2014 at 05:14 PM.
Following the advice above, I adopted the following procedure in DSS:
Guiding principle for Startools pre-import file preparation = minimize manipulation of images in DSS – keep them as ‘virgin’ as possible
Register Settings
Star detection = aim for ~50 stars, no noise reduction by median filter
No automatic detection of hot pixels
Stacking Parameters
Result = standard (or custom), no alignment of RGB channels, no drizzle (except small custom areas)
Light = median, no background calibration
Dark = median Kappa-Sigma clipping, no hot pixel detection, no dark optimization
Flat = median, Bias = median, Alignment = Automatic
No intermediate files, No cosmetic post calibration
Other
No post-processing in DSS, export as rational 32-bit FTS
Attached image shows what the pre-stacking settings look like.
Ivo, I imported this into Startools and have started mucking about. It is already looking better, I think, but any advice would be welcomed!
I played about in Startools (still not sure what the right workflow might be, so I just followed the general flow from Appendix 1 of the unofficial manual), and I cropped to the area around the nebula.
Does it look better/worse than the original attempt, do you think?
--- Auto Develop
Too see what we got.
--- Bin
Parameter [Scale] set to [(scale/noise reduction 50.00%)/(400.00%)/(+2.00 bits)]
--- Crop
To get rid of stacking artifacts.
--- Wipe
Parameter [Dark Anomaly Filter] set to [6 pixels]
--- Auto Develop
Final stretch. I created a Region Of Interest over the Great Spider.
Parameter [Ignore Fine Detail <] set to [2.1 pixels] to have AutoDev ignore the noise.
--- Deconvolution
Automatically made mask.
Parameter [Radius] set to [1.8 pixels]
--- Wavelet Sharpen
Default values.
--- Color
I'm having trouble with the colours. There is too much blue (not sure why this is).
Parameter [Cap Green] set to [To Yellow]
Parameter [Dark Saturation] set to [2.70]
Parameter [Blue Bias Reduce] set to [1.29]
Parameter [Green Bias Reduce] set to [1.20]
--- Life
Parameter [Strength] set to [79 %]
--- Wavelet De-Noise
Final noise reduction;
Parameter [Scale 5] set to [50 %]
Parameter [Redistribution Kernel] set to [12.2 pixels]
Parameter [Read Noise Compensation] set to [1.38 %]
Parameter [Smoothness] set to [73 %]
Thanks Ivo - looking fantastic! It seems like you did this on my old DSS extraction though? (in the new extraction, I cut down the stacked area using custom - what you see in the JPG in my post from last night is the full extent of the stacked area [and I didn't use the flux model]).
Anyway, I will try those settings you've suggested on my new DSS extraction too. I really appreciate your help - this is giving me great guidance on what to try.
I agree that it does look very blue. No filters were applied at the telescope, so I'm not sure why that might be
There's a lot of OIII in 2070, but there's also a lot of Ha. Depending on CCD sensitivity to each wavelength, the colour of the big spider can be anything from red to blue.
This shot is with a CLS on a full spectrum DSLR which lets a lot of Ha through so it ends up red. If you had your Nex3 full spectrum modified, it would let through a lot more red Ha.
There's no right or wrong colour, just a personal preference.
There's a lot of OIII in 2070, but there's also a lot of Ha. Depending on CCD sensitivity to each wavelength, the colour of the big spider can be anything from red to blue.
This shot is with a CLS on a full spectrum DSLR which lets a lot of Ha through so it ends up red. If you had your Nex3 full spectrum modified, it would let through a lot more red Ha.
There's no right or wrong colour, just a personal preference.
Good explanation, thanks. Since my DSLR is unmodded, I guess I get more blue dominance over the red Ha as well.
Thanks Dunk! If figured I'd try to squeeze everything I could out of my existing equipment and really learn the ropes of AP and image processing, BEFORE I invested in some serious AP-dedicated kit.
It's been surprising to me how much signal one can extract out of many stacked 'snaps' (subs <30s) on an unguided alt-az mount. I hope there'll be plenty more images on the way -- once the damned clouds clear!
Here is a first go at the Keyhole Nebula, NGC 3372, again using the 8SE on the single arm alt-az mount. I used 25 min of 10 sec and 25 min of 30 sec subs. Details here: http://www.astrobin.com/98885/
I drew out some interesting colours! But really, I'm still at the 'mucking about' stage in StarTools right now (but enjoying it).
). I guess you are saying that I should let DSS do no corrections except those required for stacking - would this include no hot pixel removal etc.?
Since my DSLR is unmodded, I guess I get more blue dominance over the red Ha as well.
