Eta C Repro with selective sharpening

[TrevorW]

Did a rework with some selective sharpening as instructed by Paul and suggested by Doug

Thoughts please

[desler]

Looks very good to me Trevor! Nicely done!



Darren

[Paul Haese]

Looks a lot better but some refinement is most likely needed. At least you are on your way to sharpening images now.

[TrevorW]

Thanks Paul for your help and for something different I did this playing with min filters etc

[gregbradley]

Nice image, earlier version. The minimum filter is only good in extremely small doses as it is quite a destructive filter. That image has lots of blotches caused by the filter. if you want to tighten up an image deconvolution is better. See Ken Crawfords tutorials on it.

Greg.

[TrevorW]

Ok Greg I ran min filter with a small dose then smart sharpen and noise reduction

PS doe not have a Decon filter percei

Ken runs a multi stength decon through CCD creating two image layers I think before processing in PS

[bmitchell82]

Trev you have well and truly cooked that one, I think something is amiss further back in your capture as the data coming up originally isn't clear and the RC does produce really clear images. and at 9 min worth of data each sub i would have expected alot more nebulosity. here is 3 x 5 min with the 40d in my back yard just outside of mandurah so LP is alot less than the city, but still there

http://i700.photobucket.com/albums/ww3/brendanmitchell1982/Eta-panel-2-post.png

same image just a crop to show at native pixels or there about depending on photobucket.
http://i700.photobucket.com/albums/ww3/brendanmitchell1982/fickle-finger.jpg

Have you looked at collimation? ect ect, because its not a guiding problem the stars are round. I believe when you figure out the root cause then everything else will start playing the game. That is just my opinion for what its worth.

[Octane]

I think it's clipped shadows, particularly in the red.

I'm sure there is an enormity of data there.

Check your levels, Trevor. Once you get that balance right, your images will shine.

H

[multiweb]

Will probably be the party pooper here but your first rendition in the other thread (http://www.iceinspace.com.au/forum/showpost.php?p=599244&postcount=10)was right on the money IMO. Very nice details and natural looking. These couple of repros are a bit harsh. Sometime less is better. Selective sharpening must be applied with "hideall" masks and blended in with very small levels of opacity to reveal the details otherwise you'll "cook" the whole field.

[Hagar]

The sharpen looks alright but the min filter is a doosy. When using a min filter, even at 1 pixel it is nearly always nessesary to use the filter then go to edit menu, fade minimum and adjust the filter back to 50 or less percent. By doing this you can reduce some of the artifacts caused when using this filter. It is nearly always a hard filter to use and nearly always leaves behind little tell tale artifacts which require a lot of work to correct.

Good try.

[TrevorW]

Doug the min filter was an experiment only to see what it does as I haven't used it before, it really strips the image data, cheers

I did notice when stacking the lights in DSS the red channel wasn't as strong as I'd expect from 9 minute subs when flats where applied as compared to when using no flats

so maybe their was an issue with the flats

also remember this was taken under a full moon in LP suburban skies,

Brendan I've got better results like you when using the ED80 with a Canon 350d but then again that is a much shorter focal length with subs around 5 minutes and I haven't overstretched this image to bring out the nebula more.

Also nothing wrong with the scope as far as collimation is concerned probably more conditions some poor focus and the need for longer subs see the crop

I tried the selective sharpening as an exercise but still working on refining this technique

Thanks H will go back to the beginning and check level again

Thanks Darren

Marc have to agree sometimes less is better.

Thanks all for the feedback much appreciated

[bmitchell82]

looking at the crop trev, it looks as if you have applied some heavy noise control, its so hard to explain to you what i see on this end, though i don't believe that its your focal length that is causing your issue, looking at Grahames gso rc10 which is practically the same scope the images come out clear as anything, furthering to this the stars seem to have a large soft halo around them which i havn't seen in other RC8's, sure they have that nice glow but not a soft fuzzy look.

http://paulhaese.net/ThorsHelmetHaRGB.html

just for example, yes i know there's a lot more exposure time but that doesn't change the soft/sharp look.

Im sorry if you think im harsh, but i believe that nearly anybody who posts images up on this site are looking for ways to make their images better, if it was to do with colours, well that is totally up to the user to decide as its personal pref, but things that everybody can comment on is the mechanics, focus, flattening, guiding, ect ect.

[TrevorW]

I don't consider you are being harsh Brendan.

I will recheck the collimation once the clouds disappear

all comments are welcome :thanx:

[gregbradley]

Not sure what camera you used - was it a DSLR?

Collimation issues and focus issues aside the best technique in my opnion to sharpen up an image is the Ken Crawford approach or something very similar. He has a tutorial on multiplle deconvolution.

But, as with any image processing, processing is always easiest with excellent data and no amount of processing should be considered a substitute for excellent data. So the basics done well sets up the excellent image.

Dark skies, exact collimation, accurate polar alignment, good autoguiding with low PE, accurate focus, cold and clean camera with a good chip and of course good optics.

Greg.

[TrevorW]

Thanks Greg can't agree more, as too Kens tutorial I think you need CCD to do the deconvolution. It has been brought to my attention that the GSO RC may appear collimated into focus but may need collimation at back focus

It was a one shot colour QHY8

[bmitchell82]

I can de-convolute my DSLR images very well so its not a specific kind of camera that it works on.

If im correct in saying, de convolution is something to do about the pixel wells themselves and that as light hits them it bends slightly, de-convolution adjusts this and re aligns the image so to speak, and hence making the fine details come out, though somebody with more experience should chime in :) . I don't believe that it should be used to fix a image like Greg said having excellent data to start with makes the world of difference.

One other thing to be extremely wary with de-convolution is over cooking it as you can see when somebody gets a bit to frisky because it looks like sand has been chucked across the image. i use CCD stacks algorithim as its one of the best i have come across, 30 - 50 iterations for mild 50-90 iterations for heavy. Once again use hide all layer masking technique with a soft fine hand to blend it in. eg, 20% opacity 20%flow other and gradually blend it in.

If you can find somebody with a autocollimator grab it and have a look at the reflections, that will quickly tell you if your close to the money.