LRGB Image processing help and ATIK USB filter wheel.

[JohnH]

Hi,

I have finally had some time to get my scope out again after a lengthy layoff due to family issues. I was determined to get my USB filter wheel working properly as I had been limited by the manual version to monochrome imaging. I had some fun with Ha but LRGB and Ha, Oiii, Sii by hand is no fun at all - even with a fast ccd.

So fwiw I share my story, some while ago I obtained an Opticstar 145m camera, to compliment this I bought a set of Astronomic LRGB filters and an ATIK motorised filter wheel, this last worked fine with the manual hbx but was very flaky under usb control. After many frustrating and failed attempts to collect images I found last night that it would only behave well when connected into an external, powered usb hub. Odd as this also connects to both cameras and the Gemini controller.

Anyway it is working now and I shared that only in the hope that it might help someone else having usb wierdness, it seemed more sensible t ome to have the wheel directly connected to the laptop on a dedicated port.

Which brings me to my question - LRGB processing - I am using AA4 which lets me do the stacking/calibration and LRGB processing but I wanted to check my workflow with those who know.

I am collecting L at 1x1 binnning at 2 or 4 mins typically and the RGB at 2x2 and 30s or 60s.

I calibrate and stack each set of lights to give RGB and L, on examination I find the B frames stars are more bloated than the R and G frames. I focused using the G filter in line using a Bahtinov mask and locked off the focuser. The filters are (supposedly) parfocal and the scope is (semi?) APO so I was a little surprised at this.

First question then : Do you refocus on each filter change? I do not have a motorised focuser so this would be a pain, also I run LRGB, LRGB, LRGB... not LLL...RRR...GGG...BBB... to ensure I get some colour data in the event the imaging run is shortened (eg by clouds) meaning many many focus adjustments.

Now I need to convert my 4 individual stacks to a single colour frame - for this I need calibration ratios.

Second question - Do you do any image processing on the individual stacks before you combine them, if so what and must you apply the same to all stacked frames to keep the data true?

Now I have an RGB image and a L image - these I align and combine - but I seem to loose a lot of the colour infomation when I do so - the bright stars in particular go white.

So I have much to learn about this but I am happy to have got started, I have attached my test image of the Jewel Box - it was LRGB - 8x2mins L binned 1x1 and 8*30s each for RGB - total of 20 mins.

[Geoff45]

You could try an individual focus with each colour just to check how parfocal they are. Then at least you'll know whether or not the big blue stars are due to a focus issue.

[Tandum]

Lots of questions in there John but I'll have a shot.

USB. I'm guessing the atik wheel has no power source plugged in apart from the USB. USB only supplies .5amp and if you run that over a 5meter lead you will have a voltage drop. So you where probably getting low volts to the wheel. Now it's on a powered hub the voltage drop is gone.

I normally shoot colour at 2x with 1/2 the exposure time. So 10 minute L @ 1x1 and 5 minute RGB @ 2x2.

Use the mask and focus on a star with L then rotate the other filters in and check they are all in focus (parfocal). For bloated BG, check the filters are the right way round in the wheel. Imagine no wheel and they should be sitting the way round you would normally screw then into the nose. If they are right way round, try a light pollution filter in front of everything, that solved all my bloated star problems.

Colour mixing ratios are established by doing a G2V star calibration. Search for that one.

Sometimes I just combine and stack them as is, other times I'll de-convolve the L and other times de-convolve all sets. I find mixing in Ha kills the colour but you can up saturation in photoshop to compensate. Trial and error is the order of the day till you get a flow that works for you. I've found DDP stretching helps me a lot at times and other times it destroys the image :P

[JohnH]

Hi Robin, thanks for the response, I had not thought to put an LPR filter in line I might just try that or I could shoot the B then LRGHa with a re-focus between.

I think it is normal for blue channel to be the worst - most subject to scattering iirc so perhaps I should aim to shoot B with the object at its highest point?

The ATIK whell has a 12V supply as well as the USB connection - but it can only be the drive for the motor not the electronics.

[Tandum]

John, If you scroll through this 1st light post (http://www.iceinspace.com.au/forum/showthread.php?t=55443)you'll find the individual panes from the horse image. You'll see blue looks very similar to all the others apart from reflections. This was from a dark site.

If you look at this early post (http://www.iceinspace.com.au/forum/showthread.php?t=48511)you'll see the bloated star problems I had.

I'd be checking that the filters are parfocal.

[Bassnut]

Dont bother with LPR filter. Its normal for blue stars to be worse. Short of setting different focus offsets for each filter (eg with DL), just apply decon to blue to match other filter star sizes (or the min filter in PS).

[Tandum]

Fred, you got any idea why I always see worse reflections in Green and Blue than Lum or Red? I've had this on 3 different brands of filters now.

[Bassnut]

Green and Blue are generally dimmer than red or Lum, so they need more stretching to get similar histograms. As a result, star bloom, refections and noise are much more obvious. You can push them hard and then blur filter in PS for an LRGB blend, because all the detail is in Lum. Lots of blur and level clip on RGB wont affect the final image quality much, this is a common problem.

[JohnH]

Thanks Fred - that is a relief - thought I was going mad. I think I read about shooting blue when the target is high - that plus a refocus for blue should improve the raw data - and I will not be shy about hitting my blue frames with decon....thanks for the tip.

I have re-processed - it is now better but I have realised that I have actually over-exposed the L at 120s and this has burnt out the star cores, my colour balance is also not great, I used the auto feature in AA4 because I have not got calibration values for my filter CCD combo yet and there is light pollution in the signal (shooting S from North of Sydney for this object)...still a lot to learn..

[AlexN]

First off mate, what are your optics? Some refractors will simply not focus blue as tightly as it will focus other colours.. I'd agree with Fred, Deconvolving the blue channel to to get the stars the same size as the rest is the way to go... My setup required a slight focus change L and R were equally focused, G was slightly out, and blue slightly out again. (with my refractor) however refocusing between filters fixed that up no worries.... I use an LPS inline as robin mentioned, however I don't know how that would make much of a difference? unless your LRGB filters are not UV/IR blocking and the LPS is. My LRGB are UV/IR blocked - I saw no difference in star sizes when using the LPS filter...

[JohnH]

I am using a 127mm APO from North Group, the filters are the Astronomics series II and are both IR blocking and parfocal and the CCD is an Opticstar 145M (Cooled Sony EX ICX285AL at 1.4" per pixel). I know focus on these is critical from using it with my 20D but I have made careful adjustments using a Bahtinov mask and using the focus routines in AA4 either way for the same exposure my stars in B are 50% bigger than the R and G frames.

I have done a few more trials since my last post - but even if I run LRG then refocus for B I get still significantly larger blue stars, I have tried several tricks to tackle this but with only limited success, RL deconvolution does reduce the star sizes a little but it causes rings to appear if pushed hard, I guess I may be better off blurring the G and R and then trusting to the L frame to get the details out?

I think the optics are ok - the attached Ha image of NGC3372 was done as a test and I am happy with that - perhaps NB is my best bet anyway due to the light pollution...