I asked John if I could have a play with his amazing Jupiter image, posted in this thread.
He kindly sent me the R, G and B channels after stacking and wavelets were applied in Registax.
I did some mild LR deconvolution in AstraImage, followed by making a synthetic Luminance channel from the layered R,G,B channels in Photoshop.
The L channel had some mild noise reduction, curves and levels applied, and was used as luminance for the colour data from the recombined R,G,B image.
Further saturation, levels and curves adjustment gave the attached image. There's a bit of cross-hatching on the left hand side due to underexposure + resampling.
Hey Mike, I know zip about how you achieved this, I am at least a year away from astrophotography, but I do have a good pair of eyes and I know what I like. I compared John's original with your reworked version, side-by-side and the amount of extra detail you have been able achieve is remarkable, there is a significant and marked degree of detail that is crisp and clearer, you have done a really great job.
Sometimes getting an untrained and inexperienced eye, such as mine, is more of an insight than collaborating with other professionals, so I offer you my observation in the spirit of friendship and admiration.
and it would be great if you can share some more about the concept of creating a synthetic Luminance channel from the layered R,G,B channels in Photoshop. Sounds interesting.
A good job, Mike, and it's really interesting to compare the two. I'd have to disagree with Ron. I think your image and your colour balance especially is a little more natural, closer to a visual image (it's considerably less blue on my monitor) and maybe a little smoother despite the artifacts you mentioned. A few bits that were saturated in the orginal have fared better too.
However overall I can't see specific detail that wasn't there in the original. In fact, to my eyes, the heightened colour contrasts in John's makes some fine features stand out a little more. Maybe it just comes down to visual judgement and personal taste - a question of aesthetics vs scientific data.
I think the differences are typical of LR deconvolution (Mike's reprocess) versus ME Deconvolution (mine). I prefer generally to use ME Deconvolution on my Jupiter images as I end up with a less grainy looking image. As well, my image has had some mild noise reduction applied to it which has enhanced this variation.
As Graeme has said sometimes it is all about asthetics vs scientific data. I can add to this also by saying that I think that eventually everyone develops some type of their own "style".
However can I just say that if it wasn't largely to Mike running IIS and the fantastic information sharing that goes on the IIS forumns I doubt I would be posting any images that were worthwhile discussing!
I agree with Graeme in part - my processing hasn't revealed any more detail that wasn't in John's original. It's mostly what comes down to personal preference and how you like an image to look.
If you look at every feature, it's in all 3 versions. Heavy contrast adjustments might make it stand out a bit more, but at the sacrifice of a more natural, pleasing look (at least, to my preference). From my understanding, even the smallest features are able to accurately measured by the ALPO guys even without heavy contrast adjustments.
So I'd say that it's not really "aesthetic vs scientific" either - the same features are evident and can be measured.
It all comes down to smoothness, colour balance and personal preference. Like John said, we all have a "style".
and it would be great if you can share some more about the concept of creating a synthetic Luminance channel from the layered R,G,B channels in Photoshop. Sounds interesting.
John,
I take the processed R, G and B channels and layer them in a new image in photoshop. I stack them by changing the opacity of layer 2 to 50% and layer 3 to 33%. I flatten the image and then do some mild noise reduction and a high pass filter (sharpening).
I then adjust the curves (contrast) and flatten the image.
I then copy that image, which is now the synthethic luminance image, and paste it as a new layer on top of the original RGB image. I change the blending mode of that layer to "Luminance" - which you can see the effect of by "blinking" the visibility of that layer on and off. It's using the "detail" from the luminance layer and the colour from the RGB layer. It (usually) has the effect of producing a smoother, more detailed image.
I then process the combined image in the normal way, to personal taste. Curves, levels, colour balance, saturation etc.
The same method (but with real luminance, not synthetic) can be used by those using a monochrome DMK/LU075 (Matt/DP) to capture the luminance and a ToUcam to capture the RGB (colour) data.
John K's amazing Jupiter - reprocessed
