This is data from last year, actually from my first light with my Sky Rover 130mm and QHY163M. I've revisited this data as I've wanted to see how much detail could be extracted from 4.44 hours. Some of the stars aren't quite right in the corners, this was mostly caused by field curvature in having incorrect back focus. It is an area that I've been thinking about getting printed, wanted to centre the Pillars more but I don't have enough data on the right hand side of the image for it to be usable - just a giant black patch.
I went way over board and did 250 iterations of deconvolution to eek out every tiny bit of detail possible. Some of it was lost when combining the Ha (luminance layer) with the OIII and SII but that is what happens
Last edited by Atmos; 06-07-2018 at 05:32 PM.
Reason: changing link to high res version
250 iterations of decon in one hit is asking for trouble. The problem is it sharpens up the noise as well, tends to put rings around stars etc. So your image is a tad noisy in the full rez image.
Its better to do several decon levels and blend them in at different opacities and only the bright areas not the dimmer areas to avoid emphasising the noise.
Nice image Colin just need to tone down the colour on the stars or add some RGB to ease on the pinkish stars. narrow band is not easy very good effort.
Martin
Looks really nice Colin. The stars are a bit strange in narrow band but I'm sure you will get the hang of that in no time.
Getting everything looking right in narrow band is a whole new learning curve isn't it? I think when I do get an astro cam it will be OSC.
It’s been a year since I last played with this data but last time I did try to go starless before processing and there just wasn’t enough data. My spacing was a few mm away from correct so I have some bad star shapes and there was a meridian flip involved so I’ve got a few conflicting bad star shapes in different corners all conspiring against nice round stars. After removing all the purple halos it left some weird funkiness.
I’ll have to have another go at full star removal!
Quote:
Originally Posted by gregbradley
250 iterations of decon in one hit is asking for trouble. The problem is it sharpens up the noise as well, tends to put rings around stars etc. So your image is a tad noisy in the full rez image.
Its better to do several decon levels and blend them in at different opacities and only the bright areas not the dimmer areas to avoid emphasising the noise.
There is a nice amount of detail in the pillars.
Greg.
I’ve found that there isn’t much of a difference between 100 iterations and 250. I find myself generally making 3-5 star masks for the different star sizes and then creating a master star mask out of those adding different intensities to get a decent protection on everything but the smallest of stars. It’s only the smallest stars that have less contrast than nebulosity that haven’t been protected.
You are correct on the noise however. I found that I could easily remove that with some kinds of noise reduction but decided against it as i also lost some finer details. I’m planning on a test print soonish and i don’t imagine that it’ll be visible in print.
My main purpose of this was to thrash the approx 1.5 hours of Ha to give up as much detail as possible
Quote:
Originally Posted by willik
Nice image Colin just need to tone down the colour on the stars or add some RGB to ease on the pinkish stars. narrow band is not easy very good effort.
Martin
Pink stars? Apart from a few tiny orange ones they’re all white.
If yes, unless my eyes mislead me, I think there is a tad more detail on the pillars in the original version - in particular on the top of the largest pillar.
If yes, unless my eyes mislead me, I think there is a tad more detail on the pillars in the original version - in particular on the top of the largest pillar.
It is indeed the same data! I've hit this data a few times now.
I've done and reprocessed just the Pillars of Creation as it is where so much of the interest lies
A worthwhile experiment. There seems to be a huge amount of very fine detail but the noise amplification makes it hard to know what is real. I think on balance I prefer your previous version.
Perhaps that amount of deconvolution might work wonders if there was a huge amount of signal.
Well done Colin and definitely worth giving it a shot even with the relatively limited amount of data. There is some nice detail in the image and the colour works well.
A worthwhile experiment. There seems to be a huge amount of very fine detail but the noise amplification makes it hard to know what is real. I think on balance I prefer your previous version.
Perhaps that amount of deconvolution might work wonders if there was a huge amount of signal.
Best,
Mike
I know what you mean Mike, the previous version probably had 10 iterations and more masks. It has less fine detail but doesn’t show the deconvolved noise in the background. Continually trying to get the best of both worlds now
Slowly slowing improving but more data would definitely make it easier. Or maybe a 20” Light bucket
Quote:
Originally Posted by SimmoW
Looks fantastic i reckon Colin! Great central colour too.
Max decon would be better applied via a layer mask, maybe. But when i opened the image it didnt hit me as false, just detailed
It has mostly caused that look in the background but I’ve since figured out how to fix it while still leaving the detail.
Still trying to get better looking stars though.
Quote:
Originally Posted by Ryderscope
Well done Colin and definitely worth giving it a shot even with the relatively limited amount of data. There is some nice detail in the image and the colour works well.
Thanks Rodney, lack of data is a bit of a pain when dealing with various processing steps.
After listening to all of the feedback I have since gone back and spent the last two days slowly chipping away at this image. I've reframed and been far more delicate in the processing; trying to bring out as much detail as possible while keeping from having too many processing artefacts.
There are still some dark rings around the smallest of stars, there hasn't been enough data in the Ha to be able to separate them from the noise in the background. Going to put some thought over the next day or two in how to fix this but at the moment, this is where I am at.
I think I have extracted every possible bit of resolution out of my 130mm refractor in this. Even after drizzle integration my average FWHM was still just a tad below 4 pixels (2.25"). Not earth shattering but I need more focal length to get anything more
Thanks Lewis and Greg.
I've just come to realise that the crash Astrobin had a few days ago wiped the image that I'd uploaded and it has since been replaced with a wrong one! I've updated the Astrobin links to the correct one
