crikey this was hard work. Object has fairly low surface brightness and seeing was 3-4 arcsec for lum and ~5 for RGB. Only managed very short total time on target over 2 nights due to clouds, so the processing was a fine tuning exercise to get back at least some detail without too much noise showing and while still keeping some of the the dimmer stuff. Colour was also compromised by gradients from thin cloud .
In the end though, was pleased with the outcome - deconvolution could only do so much and there are lots of technical faults, but it is such a beautiful object that the image is still worth keeping . Looking forward to imaging it under better conditions. thanks for looking - appreciate any feedback. Regards Ray
lum 50x2min@1x1
RGB 14:15:18 x2min@2x2
image cropped and scaled by 0.7 to improve appearance
200f4 + RCC1 at 1.17arcsec/pixel
SX H694
NEQ6
edit: on the basis of helpful and constructive comments, for second image, redid colour calibration, turned down stars a little and boosted dimmer regions slightly. Think I have reached the noise limit for this data. Thanks folks for the input.
That's great for the short exposure, Ray! I needed eight hours at f/8 with a KAF-8300 to get a half decent image.
Thanks Rick. would have liked a lot more data, but this camera has high enough QE and low enough noise to produce usable results (just..) from relatively short exposures.
Quote:
Originally Posted by strongmanmike
Excellent result Ray, given the described seeing
These Starlightxpress cameras are pretty good huh? Real alternatives to the popular 8300 based cameras I recon?
Mike
Thanks Mike. Yep, I like the SX camera a lot - to the point where I do not think about it much anymore - It just seems to work very well regardless and results will be determined by other factors (eg seeing). As for it being an alternative to the 8300, I think it is different enough to have its own niche. Where you want high resolution imaging from short focal length scopes and in conditions where long exposures may not be possible or desirable (particularly using Ha), the icx694 is in a league of its own. It costs a bit more than an 8300, but I think that the extra performance could be worth it in many cases.
FYI, have been trying a range of pre-processing techniques and have got the best possible SNR by doing an initial hot pixel removal from darks and lights - before the normal calibration process. The read noise of the chip is so low low that it doesn't mask the minor noise that results from dark subtraction where the thermal noise is significant (hot pix). Getting rid of a few hundred warm/hot pixels first is worthwhile. Of course, this extra effort is only beneficial if you are digging around in the depths of the data for tiny signals - on most targets it wouldn't matter at all.
Quote:
Originally Posted by Lee
A great image Ray... I might stalk you and try it myself next new moon!
thanks Lee. looking forward to seeing your results.
Thanks Mike. Yep, I like the SX camera a lot - to the point where I do not think about it much anymore - It just seems to work very well regardless and results will be determined by other factors (eg seeing). As for it being an alternative to the 8300, I think it is different enough to have its own niche. Where you want high resolution imaging from short focal length scopes and in conditions where long exposures may not be possible (particularly using Ha), the icx694 is in a league of its own. It costs a bit more than an 8300, but I think that the extra performance could be worth it in many cases.
FYI, have been trying a range of pre-processing techniques and have got the best possible SNR by doing an initial hot pixel removal from darks and lights - before the normal calibration process. The read noise of the chip is so low low that it doesn't mask the minor noise that results from dark subtraction where the thermal noise is significant (hot pix). Getting rid of a few hundred warm/hot pixels first is worthwhile. Of course, this extra effort is only beneficial if you are digging around in the depths of the data for tiny signals - on most targets it wouldn't matter at all.
Regards ray
I had never done it before with my previous SX and FLI cameras but initial results with the H694 suggest that dithering and then median combining takes care of everything and prevents any noise being added by dark subtraction. It's pretty cool to watch on the screen too, when the guide star gets moved around and locked onto again
This (full frame full resolution) is a median combine of 6 X 5min dithered sub frames, no darks and no flats (of course it is pushed quite hard to reveal as much as possible from just 30min). NB the pause after dither move was not long enough and half the subs started before the guide star had been fully re-acquired and this showed slightly in the star shapes. Since this, Fabio from MSB Software (Astroart) has rewritten the camera control code to lengthen this pause for me
No Darks and no Flats!
Easily Amused Mike
Last edited by strongmanmike; 16-07-2013 at 11:39 PM.
I had never done it before with my previous SX and FLI cameras but initial results with the H694 suggest that dithering and then median combining takes care of everything and prevents any noise being added by dark subtraction. It's pretty cool to watch on the screen too, when the guide star gets moved around and locked onto again
This is a median combine of 6 X 5min dithered sub frames, no darks and no flats (of course it is pushed quite hard to reveal as much as possible from just 30min). NB the pause after dither move was not long enough and half the subs started before the guide star had been fully re-acquired and this showed slightly in the star shapes. Since this, Fabio from MSB Software (Astroart) has rewritten the camera control code to lengthen this pause for me
No Darks and no Flats!
Easily Amused Mike
Thanks for the info Mike - I guess that method also uses a non-linear process to get rid of hot pixels and it overcomes residual fixed pattern thermal noise by spreading it around - will give it a go. I have not tried to use dithering yet with the H694, even though phd/Nebulosity supports it - I have been wary of having the EQ6 deliberately go off target, since it can take a long time to guide back on accurately enough at ~1 arcsec/pix and 2-3 second tracking loop update rate.
The image you linked to sure is a good illustration of just how quiet these chips are.
Thanks for the info - that method similarly uses a non-linear process to get rid of hot pixels and it overcomes residual fixed pattern thermal noise by spreading it around. I have not tried to use dithering yet with the H694, even though phd/Nebulosity supports it - the last thing I want my EQ6 to do is deliberately go off target, since it can take a long time to guide back on accurately at ~1 arcsec/pix. Regards ray
The original pause after dither move in Astroart was only about 3sec and this was not long enough at all, so Fabio re-wrote that piece of the camera control software for me and increased the pause to 6 sec. This was better but still not long enough for 50% of dither moves, particularly if the move was in Dec, so he has since lengthened it to 15sec for me - this should do it and I can set the max pixel movement to just 2 pix, which is enough for dithering to work.
The original pause after dither move in Astroart was only about 3sec and this was not long enough at all, so Fabio re-wrote that piece of the camera control software for me and increased the pause to 6 sec. This was better but still not long enough for 50% of dither moves, particularly if the move was in Dec, so he has since lengthened it to 15sec for me - this should do it and I can set the max pixel movement to just 2 pix, which is enough for dithering to work.
Mike
phd/Nebulosity uses the measured tracking error to decide when to start imaging again. If it is set too tight, it can take a long time and since I use the 694 with short subs, the imaging efficiency takes a hit.
An additional advantage of dithering is that it would allow super-res to be applied to any data you get in exceptional seeing conditions - that could be very useful.
Forgot to mention that 2x2 binning worked as advertised. Like most CCDs, the horizontal shift register in the 694 cannot accommodate the full signal from 4 pixels, but there were no signs of any unwanted overload artefacts seen that apparently can turn up (the 8300 for example is reported to produce slightly elongated bright stars). Seeing was pretty bad though when I used binning, so that may have masked any overload effects.
Quote:
Originally Posted by johnnyt123
that is a great image from the short exposure time....
I have been looking at buying a new CCD camera.
Do you have any experience with
Starlight Xpress SXVR-H814 Monochrome CCD?
I hope to have a fast system eventually.
John
Hi John. I don't have experience with the 814, but it has specs that are ~identical to the 694 in all but pixel size and number. On that basis it would be a great foundation for a small high res system with good sensitivity. It should be suitable for full res imaging with scopes up to about 1m fl in very good seeing conditions.
Awesome work Ray. As others have mentioned, you've extracted the faint galaxy extensions very well. Love the spiral structure and dark lanes along with the HII regions. The stars with a magenta cast are a little distracting but nothing major. I suspect this may have been caused by the blue channel highlight clipping noted in the histogram. However its not that severe so there could be something else at play. Normally magenta is attributed to a lack of green which may be the clue as its the weaker of the three channels. i.e could probably be slightly stretch more to better match the strengths of Red and Blue. Really its a minor point, but I thought I'd offer this feedback in the scenario you are looking to improve on the data set. Thanks for sharing. Hope to see more.
edit: something I find useful is the RGB colour space chart such as this one - http://www.colorspire.com/rgb-color-wheel/ Work with the opposites .i.e. to get a balance when there is too much magenta, add green, if there is a cyan cast add red. Of course there is only so much addition you can make to an additive colours space that is RGB i.e. all white 255,255,255 can't be coloured which is a primary reason why managing star profiles is important. I hope this provides greater insight into what you can do with RGB channels.
Last edited by jase; 17-07-2013 at 04:18 PM.
Reason: colour chart
Awesome shot Ray. Great colors and details. You've captured the outer arms nicely. It's bloody faint hey?
thanks Mark. sure is dim - I hadn't quite appreciated how far down in the sky noise it would be.
Quote:
Originally Posted by David Fitz-Henr
Nice shot of a difficult object Ray; good detail and you've captured a lot of the fainter extensions as well
Thanks David. I had a look at your excellent 6744 - something to aim for.
Quote:
Originally Posted by jase
Awesome work Ray. As others have mentioned, you've extracted the faint galaxy extensions very well. Love the spiral structure and dark lanes along with the HII regions. The stars with a magenta cast are a little distracting but nothing major. I suspect this may have been caused by the blue channel highlight clipping noted in the histogram. However its not that severe so there could be something else at play. Normally magenta is attributed to a lack of green which may be the clue as its the weaker of the three channels. i.e could probably be slightly stretch more to better match the strengths of Red and Blue. Really its a minor point, but I thought I'd offer this feedback in the scenario you are looking to improve on the data set. Thanks for sharing. Hope to see more.
edit: something I find useful is the RGB colour space chart such as this one - http://www.colorspire.com/rgb-color-wheel/ Work with the opposites .i.e. to get a balance when there is too much magenta, add green, if there is a cyan cast add red. Of course there is only so much addition you can make to an additive colours space that is RGB i.e. all white 255,255,255 can't be coloured which is a primary reason why managing star profiles is important. I hope this provides greater insight into what you can do with RGB channels.
Thanks Jase. Really appreciate you taking the time to have a close look at the image. I knew the colours were not quite right, but was not sure why and what to do about it - your comments have been very helpful and I will go back and do some more detailed digging around to find out where things have got out of balance and what I might do to improve this image and my technique. As you point out, handling stars that are saturated before colour combination is a major issue just for starters. I know enough about the electro-optics side of things to get images, but am on a steep learning curve when it comes to processing them in colour, so really appreciate your help - thanks.
Well Ray, your hard work has paid off. I was expecting an image no where near that good after reading your intro....
Chris
Thanks very much Chris - it was a case of grab about 3 hours of data between clouds and then spend days trying to drag something out of the results, so I was very pleased to end up with something that shows the galaxy. Stars are pretty blobby though - nothing much I could do about the seeing ..
Have attached a repro to the original post based on input from those who responded to the thread - grateful for helpful advice.
. 
. Looking forward to imaging it under better conditions. thanks for looking - appreciate any feedback. Regards Ray
