Jupiter processes

[matt]

Interested to get opinions. Here are two images of Jupiter featuring markedly different processing routines using the same avi.

Which one do you prefer?

Will reveal what processing took place later ;)

[iceman]

I like the second one.

[hevelsky]

Like Mike i prefer the second
Great Image :thumbsup:

[Dennis]

Gee, Matt, you've come a long way, despite the atrocious seeing reputation of Canberra. Those are big Jupiter's with good detail. The second one just looks more pleasing to the eye, especially when viewed side by side.

Cheers

Dennis

[Harpspitfire]

i wouldnt be able to say which is more color accurate- but the second definitely shows more detail and contrast- even though they are the same 'images'- the second process has a better 'focus' (for lack of better words)

[gbeal]

I'm with the above crowd, but both are good.
It would be better if they were open side by side.
Spill the beans, what's the scoop.

[matt]

OK.

First of all, thanks for the replies. I agree. My preference is for the one on the right. Which is a bit of a surprise.

Here's the thing; the one on the right is a straight run through Registax with some very mild wavelets. That's all.

The one on the left was given the full treatment. virtualdubbed into BMPs... RGBSplit and ppmcentre (using the great new GUIs) ... each channel thru registax, stacked, mild wavelets... AstraImage ... LR decon, recombine ... in to photoshop.

The works!!!!!!!!!!! and yet ...

Go figure???:shrug:

I'm not sure if it made any difference but I did perfrom the RGBSplit before the ppmcentre.

Perhaps this was a mistake, Mike????

[iceman]

I wouldn't call it a mistake - it's not the way I'd do it, but the result should not be any different in terms of the quality of frames.

Interesting results you have, and yes surprising. I just performed my routine on one of Adam's (hitchiker) runs from the LPI, and I'm waiting to see his original version again but i'm quite confident still that the processing routine I use will produce a better image most times.

In this case, it didn't.. it could be the seeing, it could be the settings used, the colour is slightly different - the first one is too yellow where as the second one has a nice dose of blue.

Interesting results for sure!

[matt]

Yeah. The colour's the strange thing.

I didn't make any adjustments there at all. The yellow just seemed to happen all by itself as a result of the full-on processing routine.

Mike. What do you mean by "it could be the settings used"???

Cheers

[iceman]

What I mean is, the amount of LR deconvolution applied, amount of wavelets applied etc. Every image has to be treated on its merits and sometimes it needs more or less "treatment".

I hope you don't mind, I took your two images into AstraImage, side by side.

I performed almost exactly the same processing on them.

The top left is your original left image. The top right is your original top right image.
The images below are the same images after my processing.

- Split into RGB
- left image: LR 2 @ 1.1
- right image: LR 3 @ 1.1
- recombine
- gamma adjust 0.8
- take into photoshop
- curves adjust on centre of image
- slight/light unsharp mask
- colour balance blue channel +7 (got rid of the yellow hue).
- Save as web

That's it.

What do you think?

[iceman]

btw it's a very good image! Congrats, you must be pleased.

The image scale is large too - is that a resample or did you use a bigger barlow than normal?

[matt]

Well .... I'll be jiggered!!!!

Mike, of course I don't mind. Well done, mate. I'm thrilled, and humbled at the same time:thumbsup:

My idea was to just compare processes as far as detail and image quality is concerned.

Re: image scale. That was achieved in PS. Just kept resizing image by 5% increments. That way avoids artefacts.

Cheers

[matt]

Interesting that the image on the right was able to take 3 (LR) iterations compared to the one on the left (2) - if I've got that right?

Would that not suggest the original image at right is slightly better???

[iceman]

It doesn't necessarily suggest the image on the right is slightly better, but I felt from looking at it, it could stand up to more sharpening (through deconvolution).

When I looked at them side by side, I feel the image on the left is actually a bit sharper (it's had more processing of it, already), but it lacked colour depth, contrast and was too yellow. Maybe that's why people (myself included) favoured the right one initially.

Because the left image had already had deconvolution applied to it, I went light on it so as not to oversharpen it and introduce any artifacts. It may have stood up to 3 iterations, but I didn't try it. I did 2, it looked good, so I stopped there :) Sometimes there's no method to my madness :)

[matt]

Very nice work... and the feedback is much appreciated.:)

[Harpspitfire]

wow mike- you really brought that out- you just had the JPG i assume, can you just run the LR decon on a finished image like this?

[iceman]

No probs, my pleasure! :D

I also think it would look even better at its original size.

[iceman]

Yep, I just used the two jpeg's in the first post above. To run LR deconvolution on a finished image, you have to split it into the individual colour channels first. LR deconvolution (in Astraimage) only works on greyscale images.

Potentially the image would be even better if I had the original avi's to process (hint hint Matt) ;)

[matt]

I hear ya, Mike.

PM details where I can send the avi.

It's only a hand-picked 109 frames (VirtDub) so shouldn't have too much trouble getting that to you.

[davidpretorius]

top work all, great data matt!

[asimov]

Good capture Matt, & nice reprocess Mike!

I had a crack at it too, hope you don't mind. I only worked with the one on the right.

Purely in the name of experimentation I wanted to see the effects of different methods in Astraimage. Heres one without deconvolution ;

I saved Matts image as a bmp & did the RGB split in Astraimage, unsharp mask 2.0 was applied on all 3, recombined. Unsharp mask of 2.5 was applied. A gamma reduction of 0.8 was applied. Screen shot it & taken into my editor (Picture Publisher) saved as a jpg, reduced in size. Thats it.

I found it interesting that it got rid of the blue hue just by doing that without deconvolution!:confuse3:

[matt]

Nice work John.

Don't mind you having a go at all. The more the merrier.

This colour situation is very weird.

Love the avatar, by the way

[matt]

OK

For those who want to download the full 109 frame avi from which these images have been produced, here's the link:

http://www.artwarehouse.com.au/Docs/dubdub.zip

Sorry for folks on dial-up (like me). It is what it is;)

[asimov]

6.59 megs...That must have taken yonks to upload on dialup!

Dialup here too. I may do it tonight as I'm going to bed ;)

[matt]

Luckily, John, I have very understanding and helpful girlfiend (member name raezee) who has broadband and uploaded it to her website;)

[davidpretorius]

astra image will darken some images with deconvolution. I do not worry as a auto colour in photoshop corrects it straight away.

You notice generally a brightening as the histostretch takes place when you split or convert to greyscale, and then darkening with some deconvolutions

As I am going through 30+ avis over the last two months, to reprocess with the new ppmcentre and split batch programs, this one cropped up.

[davidpretorius]

easily fixed, marry her and words like "understanding" and "helpful" words are quickly dropped from your vocabulary!!!! :D

[matt]

:lol:

But back to your prior post Davo

That's amazing how your top left image (combine) has the same overall blue colouration.

Even your avatar has a blue tinge.

Any idea why?

[asimov]

:D

[davidpretorius]

working with the split r,g,b and processing them all differently ie differenet wavelets and deconvolution will darken / brighten each channel. upon recombining, the colour will be off. The auto colour will fix it though in PS

[matt]

Cheers

I've never used auto colour in PS

Will check it out:thumbsup:

[davidpretorius]

no worries mate!!!

[iceman]

OK here's my process of your avi, Matt.

Great data. Just needs more image scale!

[matt]

Wow:scared:

That's really something, Mike. Thanks so much for going to the trouble of downloading and processing.

I'm really encouraged. Must have been one of those rare nights in Canberra when the seeing climbed above 3/10:)

Must get this scope out of Canberra sometime and along to a star party and do some imaging alongside you ... and other IIS-ers. See what she can really do:thumbsup:

I guess you followed your usual processing path: VD split into BMPs, ppmcentre.and.. RGBSplit (with new software)... run each channel thru regi, stack and mild wavelets, AI (LR decon) and AI recombine back, and then a little work in PS?????

Cool stuff. Again, many thanks

[iceman]

That's exactly what I did Matt. Glad you like it, it was nice data to work with.

In AI, I did 5 iterations @ 1.15.

And I did use the new PCFE and RGBSplit front-ends.

[asimov]

Truly a great job reprocessing Mike! :) Well done!

[matt]

Mike

Been meaning to ask for days.

In your earlier post in this thread you refer to the above in PS

How is this done? How do you work curves on the centre of the image?

Cheers

[iceman]

Hi Matt
It's explained in the tutorial, under the Photoshop section. Let me know if you have any q's about it.