First Centaurus A ...

[CarlJoseph]

Hi folks,

Keen for some advice on this one. It's my first image after a reasonable polar and drink alignment.

Stellarvue SV105 f/6.2 APO
Canon 450D unmodded
400 ISO
2 minute subs
20 minutes total exposure
No darks

Out to the right of the image the stars look elongated. Is this what you call coma or something else?

:thanx:

Cheers,
Carl

[CarlJoseph]

I've also uploaded the stacked RAW image (http://carljoseph.com.au/misc/2011_04_07 Centaurus A - Stacked.TIF) (70MB). Would be keen to see what others can get out of it.

[A23649]

i love the photo but have no idea why some of the stars near the edge are elongated

[Hagar]

Hi Carl, I have had a quick process on your image. The unequal stars would probably be caused by your camera not sitting quite square in the focuser or the focuser sagging a little. Overall the image could use a bit more exposure but as it is you have done quite well. I would have been very happy to get an image like this when I started imaging.

Well done.

[desler]

I probably would have said the same as Doug, Possibly the camera not sitting square or focuser sag.

Having said that, you've picked up some great data (as Doug's repro shows) and should be very encouraged about your future prospects! That set up is going to provide some nice images. Well done!

Darren

[miker03]

Carl while I have no experience of using SLR cameras for astrophotography, the lack of symmetry of the elongation points to something like how the camera is mounted. Short of taking another image to see if it appears again I have no other suggestions. A great image nevertheless.

Mike

[mr bruess]

interEsting photo

[CarlJoseph]

Thanks for all the comments folks. I really appreciate it. I'll take a closer look at the camera mount and focuser next time to see if I can pinpoint what's happening there.

Doug, thanks so much for reprocessing my image. It's useful to see what others can get out of it. Curious that the colour is different and you've manage to keep away much of that reddish noise. Hmm ... time for more practice me thinks! :)

Cheers,
Carl

[Hagar]

Hi Carl, The biggest tip I can give you at this stage is to keep an eye on the histogram when processing an image. Try to keep the histogram close to the left hand wall of the histogram and adjust each colour using levels to align all 3 colours with each other and close to that left hand wall (0 end of the histogram) (black point).

[CarlJoseph]

Thanks for the tip Doug. Keeping that histogram to the left and the colours aligned seems to help a lot. :thanx:

I've tried another round of processing and it's looking less noisy this time and much more "in control".

[batema]

I have used curves to start with in photoshop and increased the bottom number to 13 and the left hand side number to 65 and then do the flatten of with about three more points.
Then levels but i open up the histogram and individual colours to the left hand side without cutting any off.
Repeated a cople of more times but less aggressive with the curves and then flatten, then smart sharpen advanced with numbers in first in Sharpen 80, 3,6 and Lens blur in drop down. In both Shadow and Highlight had numbers on 50, 50 and 20. Then filter and remove noise. I hope it helps.

Mark

[CarlJoseph]

Thanks Mark! :thumbsup: The details you've provided are an excellent help. :thanx:



So does more exposures = more subs, or longer subs?

[A23649]

more carl

[CarlJoseph]

I spent some time last night taking some more subs for this one. This time I have 20 subs of 2minutes each at 400 ISO. I only got 10 darks and no flats.

I can now see a little more of the internal structure and some more "wispiness" at the ends. Slowly getting better I think.

Any insights, thoughts and tips from others?

Cheers,
Carl

[batema]

Hi Carl,

I remember my first image of this object and we sort of could maybe possibly see the galaxy shape with star trails prevelent and was thrilled so your image is about 10000 times better than mine.
I think initially when I processed an image I would try to process more than the data would allow and I think this is what you have done here in your excitement. The very distinctive white round sphere should not be there and shows you have pushed your data too much. A rule of thumb I use in phtoshop is increase the curves initially so the numbers on the x axis are 13 and the y axis is 65 and then I soften the curve to make the dog leg shape. Then go for levels of esch channel. Then repeat with x axis still 13 but you may have to not quite go to 65. You decide but as soon as you start to blow out the core with that white circle I would stop and reduce slightly. The white circle will reduce when you dogleg the curves line so just watch. I think after 4 yrs of learning I'm for the subtle look. Hope this helps.

Mark

[CarlJoseph]

Hi Mark.

Thanks so much for the advice. Looking back now I have pushed the image quite a bit and blown out that centre part. I followed your instructions and got the following result which I think is quite pleasing. :)

Cheers,
Carl

[batema]

Much better Carl. I look at the second image and it looks much more natural rather than the previous. Have a look at the difference between the stars as the latest image has much more realistic stars. I would definately have a read of Rod Wodaski's "The New Astro Zone System for Astro Imaging" and there is a guy here who has done some movies about processing which I have found very helpfull as well as Google Utube Astronomy Imaging Processing and have a search. Keep them coming.

Mark

[joecool]

OK, I'm new to this processing as well, but given another image to play with, this is my attempt.

Process.......
Photoshop - Cropped top blank bar. This was mucking up fits liberators auto settings and clipping histogram.

Maxim DL - Auto Flatten Background, white balanced, Remove Background Colour, saved to tiff.

Tried fits liberator ArcSinh(x). Way too much background noise. hmmmm. Works well on bright nebula...

Photoshop - Auto Tone, Auto Contrast, Auto Colour.

Noel Carboni's Astronomy Tools v1_6 - Color Blotch reduction.

Noise Ninja.

Um, resized to HD 1080, but the jpg to keep the file under 200k adds grain all over, oh well.

A crop of the center makes the nebulosity brighter and kills off the distorted stars, but I left it like this for now...

Mark.

[joecool]

OK, here is the image i did before just cropped to 1080 high in original resolution from the camera. Lets me do better quality jpg so less noise added and removes the bent stars to the right of the image.

I use the 450D, and I recommend doing 4 or 5 by 5min exposures. On fainter targets if there are no bright stars in shot then go for 10minutes (possibly longer? I have not yet tried with 450D???). It does make a big difference. On objects with great contrast then do both 5 min and 10 min and do HDR or create a gaussian blurred mask of the brighter image and apply the dim image over the bright image using this to bring out over exposed cores (M42 comes to mind).

Doing 2min exposures on a faint target is not real good as the edges of the object are about as bright as the noise of the camera. Once stretched I'm seeing a lot of red/blue blotchy pixelation between the stars, we're stretching out the noise of the camera. Doing lots of exposures is great if you have high initial brightness of the target being captured. Exposures are not added together as they are stacked. What the software is doing is comparing them to remove the noise hence increasing the contrast of the brighter target and letting you stretch the image more without showing up more noise and grain.

Don't waste too much time at first doing 20 shots. Try for 5 or so longer shots. More is better, but not if you have limited time. Spend more time polar aligning and setting up. If your stars are not round at 2min, then they will be worse at 5min.

Allow for dark, bias and flat capture time. I noted you mentioned nothing about bias shots. They are important to remove readout noise from the darks and flats and images. Again, just run 5 of these at first. You can use darks and bias from previous nights work if you forget to take them. I run them with the cap on the camera body whilst I pack up the scope so they are taken at the same temperature as the main images.

Your image without flats caused the center to go bright. Without a lightbox this can be removed pretty well with software and is important to do, especially with this round target near the center. When you push to 10min exposures on faint objects you will bring out the dust donuts on the camera sensor, so a light box is better.

Hope this helps,
Mark

[CarlJoseph]

Hi Mark,

Thank you so very much for all that info. :thanx: I really appreciate the detail. It's all incredibly useful and helps me greatly.

I'm keen to try guided exposures but before then I'll spend more time polar aligning to see if I can push the exposure length a little more.

Let me take in your notes and I'll give it all another "shot" next time I'm out. Now just to wait for a clear sky down here.

Cheers,
Carl

[midnight]

Well done Carl.

You should be very happy with that image. Centaurus A is a target I would like to have a go at one day - if only I can get the time...

Darrin...

[Nugeorge]

That's an awesome looking galaxy.. Well done Carl, looking forward to my first shot at it.

[CarlJoseph]

Thanks for the feedback folks. I'm quite chuffed at the response. :) I know I can get a better shot so will keep practicing at this one for a while once the skies clear again.

[bmitchell82]

Ill pitch in my 2 bobs worth on the processing side of it. Although I agree with Mark, with the levels and curves you should try not to use the curves feature till you have fully stretched your data.

What are you trying to do?

Dynamic range. Bright to dark areas, M42 Orion nebula has a massive dynamic range from faint whispy's to bright enough for the naked eye to see.... Stars are a large dynamic range

Linear Vs Non Linear stretching. Levels = Linear, Curves = Non Linear
This is how a lot of more advanced images keep star colour (that and having a big pixel well depth more photons before the well floods (turns white).

If your using Photoshop I know you can work with the RGB or individual R G B.

Step 1. Levels

Iteratevely clip the black point to the start of the histogram (left hand side like doug said. My own process i don't take it right to the edge as you may loose some really faint data a small tail is desirable in my eyes), Leave the white point alone for most case's, and the middle point slide about 1/3rd closer to the start of the histogram from the right hand side. As a general rule i will do this iterations 4-5 times but you know when your done, the histogram will not change any more. You have now streched your data to fill the full gamut (range of colours).

Step 2. Curves

Use sparingly this is non linear, eg you can hold one section while changing the other. Highlights, Mid tones, shadows, RGB or R G B, this is a good way to boost certain colours while keeping the others the same. play with it get a feel for it and then use it like a feather not a black smiths hammer.

Black point and White point, If you change your black point (clipping) then you will change the value of the pixel under that point to zero. eg, the pixel has a value of say 100 ADU (it escapes me right now but its basically the count of pixels in the pixel well) if you change the black point to clip everything under 900 ADU, the pixel with 100 will be zero essentially turning it black.

Be aware of the colour of the back ground, HLVG is great for removing Green from a image, it is a photoshop plug in, though it only works with 32bit.

Once you master the levels and curves you will find your white balancing and colour reproduction will be very good and your star colours will stay really nice!

Hope that helps
Brendan:thumbsup:

[irwjager]

That's a wicked shot! Honestly, a data set and object like this has no place in the beginners section anymore :)

Many, many great tips here already. Once you're comfortable with all these great pointers, you may want to try & get your head around masks/layers and working on your data selectively. It's what the greats like Ken Crawford do.

When it comes to color correction; trust your data. Try to understand exactly why things look the way they look. If your colors are off (which is almost always the case), chances are there is something polluting the signal (light pollution, secondary reflections, IR or UV fuzz, etc.) Methodically find the causes and simply remove the isolated pollution from the signal. Pixel math will help you here. Once you have subtracted all polluting influences, chances are you will be left with something close to the real colors of the object. A lot of people seem to struggle with color correction; it's not magic, just science! :) One last tip on color correction is that some objects in the sky are perfect samples to calibrate your white balance against (notably a lot of galaxies).

If you use an OSC or DSLR and you have oversampled your data (e.g. the CCD resolution was higher than the scope could resolve under the seeing conditions at the time - or in laymans terms, your image looks blurry due to seeing), then use software binning to reclaim signal fidelity by trading in resolution. Many DSLRs and compact cameras these days have the exact same option. They call it something like 'low-light' mode and it is the reason why they can attain these high ISO ratings; they trade in resolution.

I took you dataset and put it through StarTools. Here are some of the steps;


Binned it to 71% (2x noise reduction)
Had the CPU have its way with the levels (manual Curve/histogram manipulation is sooo last decade :P).
Modeled the light pollution (at least I assume it was light pollution - a signature orange glow) and subtracted it. After that no color correction was necessary (a good sign), although I did bump up the saturation a bit.
Then applied a lens correction model to the image. This sorted out the worst of the distortion.
Cropped the image to center Centaurus A.
Created a mask with just Centaurus A, minus halo and applied a tiny bit of deconvolution to it.
Created another mask to preserve Centaurus A + halo.
Applied a median filter, combined with a luminance mask to modify only the darkest parts (deep space noise reduction).
Selected all the stars and put them through the Repair module to warp them into perfectly round shapes.
Applied a touch of Synth (8" refractor - I wish!) to the stars to soften them up a bit.
I'm sure I forgot some steps here.

I didn't bother with the star colors too much - I like how it keeps the focus on Centaurus A as the only distinctly colored object.

Great image and I hope to see more in the Deep Space section! :thumbsup: