Make your own Jupiter from my data

[iceman]

Hi all

In the continuing trend of collaborative processing tips and techniques, open sharing of information and ideas, I've uploaded the raw stacked images from my excellent Jupiter session on the 26th May.

The files are "raw stacked". That means, I've ran them through registax to align and stack the raw frames, but have not done any wavelet or other processing on them. They have had mild noise reduction applied in photoshop (part of my usual processing routine).

Each file is a 1.2meg TIF file. Right-click on the link and click "save target as" to save the file.

- Red Channel (http://www.iceinspace.com.au/images/jupiter-20070526/11-r-r.tif)
- Green Channel (http://www.iceinspace.com.au/images/jupiter-20070526/11-g-r.tif)
- Blue Channel (http://www.iceinspace.com.au/images/jupiter-20070526/11-b-r.tif)

As a suggestion for the less initiated, open each channel one at a time in Registax, and it will take you straight to the wavelets page where you can start the processing.

Use AstraImage, Photoshop or similar tools to combine the image back into an RGB image.

An (un-inforced) requirement is that you post your processing steps when you post your version of the image. This will help everyone (myself included) learn from your method of processing and be able to compare the results.

Have fun, and I look forward to seeing the results!

[Astroman]

Thanks for the opportunity to have a go Mike.

[Bassnut]

Hi Mike

I try to load any of the tiffs into Registax V4 with "select" and it faults with "problem reading tiff file. processing cancelled"?.

Cheers
Fred

[ving]

load and combine them in ps-cs2.

[Bassnut]

OK, I got it now. They load into CS OK, but they are already RGBs, which causes a fault loading into registax. I converted them to grey scale in CS and now they load OK in Registax.

CS doesnt have wavelet processing, so I was keen to try that 1st.

[joe_smith]

Ok here is my attempt, I dident know really what to do and took me a while to combine then in Astaimage. I had to convert them to gray scale first? is this right? I thought RGB images were already in gray scale. tried the LC then combined them the put it through focus Magic to sharpen it up a bit. Then into photoshop played with the color and levels, then save to the web. I dont like the banding I got around Jupiter and I couldn't get the color right for me. Need heaps more practice. :thumbsup:

Edit:- did two more tried to get it darker and more detail

[Ingo]

Why TIF files? PNG Is lossless.

Anyways,

Put it in photoshop, paste into RGB channels. Messed around with contrast, sharpening, auto levels, auto color, levels, etc. Desaturated the blues that ended up around the edges.

[iceman]

So is TIF (lossless).

[Ingo]

TIF is a lot bigger. PNG is compressed lossless.

Like in audio WAV is huge, and FLAC is small. Both are lossless

[iceman]

Thanks for having a go, Joe and Ingo. I'm surprised others haven had a shot at it.

While the captures are in mono, at some stage during my pre-processing (ppmcentre etc) they end up as RGB bitmaps even though they're single channel. So it's normal (for me) to have to convert them to greyscale before combining them.

I didn't have any trouble opening them directly in Registax 4 though?

Joe - your first 2 look good in terms of data and sharpness, but they're too pink/red. Try adding more green and blue in the levels. The last 2 are good for colour, but the details are kind of blurry and not sharp. Great attempts though, well done.

Ingo - yours looks good, nice colour but could probably do with a bit more sharpening. Nice job for a non-planetary guy.

I hope to see some others, and I hope this project has helped anyway, and I'm happy to do the same again in future when I've got good data to share.

By way of comparison, here is my version of processing on the same data, done back at the time of capture.

Of course, processing is a very subjective thing and what looks good to one person might look oversharpened, or the wrong colour, or whatever, to someone else.

[joe_smith]

very nice image Mike. I think its a good idea :thumbsup: will help build up our skills, and shows how much different one image can be processed with the same data.

[sheeny]

OK I've finally got the time to have a go.:)

Here's the procedure I followed:

1. Convert each image to greyscale in PS CS3 and save.
2. Open each image in Registax V4, and apply 'Dennis wavelets" (1=20.1, 2=10.5, gaussian linear) and save.
3. Load each image into CS PS3 again and combine into RGB.
4. Levels (top slider each channel down to 192).
5. Curves (I chose standard increase contrast curve and then lifted the dark end of the curve a bit).
6. Selective colour: Cyan -60, Magenta +40, Yellow -32, Black 0.
7. Save for the web.

The wavelets might be a bit harsh for this image - it looks a little grainy to me when combined to RGB.

Thoughts?

Al.

[sheeny]

Here's a second go... This time I:

1.Converted original images to greyscale, doubled them in size and saved them in PS CS3.
2. In Registax 4 I Stretched intensity levels, applied heavy wavelets (22.1, 11.2, 10.5, 10.5, 21.2, 25.3 gaussian linear), gamma 0.7, histo stretch top slider to 175.
3. Back in PS CS3 I combined images into RGB.
4. Adjusted levels (R 219, G 209, B 212)
5. Curves (dragged down a bit like reducing gamma to increase contrast in highlights)
6 Saved for web reducing size back to original.

[sheeny]

:lol:Just goes to show... I'm not getting the sort of detail out of that data that you are Mike! Care to share details?

Al.

[iceman]

Nice work Al, I prefer the first one for sharpness but the colours of the 2nd are nice.

The first one will clean up quite nice with a "dust and scratches" or "despeckle" filter in photoshop (under filters->noise).

I'll explain mine in depth when I have a bit more time. Great attempts!

[Ingo]

Here's another version, just did unsharp mask. Lacks in quality due to me not having the original high quality version of my first edit anymore.

[joe_smith]

Just tried it out and it on Al's first image and it did make a big difference, good tip :thumbsup: just used the default settings

[sheeny]

Good work Joe! I'll play with my originals!:thumbsup:

Al.

[sheeny]

And here they are. I also backed off the selective colour a bit on the first one (it was a bit red!)

Al.

[joe_smith]

That first one is very good :thumbsup:

you get my vote :)

[Ingo]

Here's another one, very sharp.

[joe_smith]

Ingo thats pretty good what program are you using?

[ronnierigel]

Thanks Mike for the opportunity to use your data--I wish my data was half as good-----it's quite easily recognisable by the way your images hold up to intense processing that they contain lots of information.

ANyway I've processd your data in photshop7 --so had to work at times in 8bits (not ideal). I did not use registax.

I did a series of unsharp masks at different levels on the different color channels, combined them, adjusted levels, curves, and hue/saturation.

Thanks again.

ron

[sheeny]

OK here's another go - different method (well sort of)... All done in PS CS3.

1. Each channel processed as follows:
Blue: unsharp mask 2px, 500% then 1 px, 198%; despeckle; Levels 202, curves - increase contrast.
Green: unsharp mask 2px, 500% then 1px, 205%; despeckle; levels 205; curves - increase contrast.
Red: unsharp mask 2px, 500% then 1px, 225%; despeckle; levels 211; curves - increase contrast.
2. Combine into RGB
3. Saturation +41
4. Brightness -35, Contrast +32
5. Flatten the image.
6. Unsharp mask 0.7 px, 131%
7. Despeckle.
8. Save for the web.

Seems to me to be a lot of double processing - sharpening/despeckling - so there's got to be data loss. It comes about from responding to the appearance of the image at each stage, and probably not getting the USM right in the first place (maybe wrong radii?:shrug:).

Al.

[Kal]

Another rainy weekend so I decided to have a crack and learn a bit more about processing.

1) Load up each image in registax and set wavelet slider #2 to ~20
2) Load up in photoshop, despeckle, set the levels "input levels" to '1,1.00,190', set contrast +20, brightness -20, then a slight despecle again through dust and scratches with a radius of 1 pixel.
3) Load up in astraimage, convert to greyscale, run a LC deconvolution with 5 iterations at 1.1 setting, then combine RGB, set gamma to 0.7.
4) Screenshot it so I can save it in MS paint :P Open in photoshop, save for web.

[Ingo]

Photoshop CS2 :thumbsup:

[Robster]

I tried opening the original image in Registax 4 and got the same error message :shrug: so I renamed each of the files with a .tif extension rather than with .tiff and they opened fine. (my computer course training coming in handy here) ;) Registax just didn't recognise the four character filename extension.

I've a bit of familiarisation ahead of me with Registax though, before I can share some results. :whistle: I took some digital photos of the Juper/Moon conjunction a few weeks ago using my Olympus on a tripod - no telescope needed - which I want to tidy up with Registax, so any practical experience I can get will be very helpful. The Moon was over-exposed on some photos while Jupiter came up sharp, and on others the Moon came up sharp and there was no Jupiter! How do I line them up in Registax or can I just stack them on top of each other in say Paint Shop Pro (I don't have Photoshop)?

Thanks Mike :thumbsup: for providing some good quality photos to start with.

Robster
Woy Woy

[joe_smith]

thanks Ingo thats what i have to :)

Al thats good mate i like the color in your ones :thumbsup:

The colors in my ones are pale looking I got one with good colors but cant seem to do i again :doh:Also when I convert them to gray scale in astra image they go heaps lighter. oh well practice, practice and more practice

Robster check this out might help or not :)

http://www.luminous-landscape.com/tutorials/digital-blending.shtml

http://www.adobe.com/digitalimag/pdfs/highlight_recovery.pdf

http://www.luminous-landscape.com/tutorials/hdr.shtml

[Robster]

Thanks Joe. I went to the links and grabbed a heap of info. Now to put it to some use.

By the way, I tried using the techniques suggested by one of the earlier mails on this thread by Al (sheeny) and found that if you convert Mike's original 3 images to greyscale BEFORE you take them into Registax 4 it doesn't like it at all. I created copies of the original images and took them into Registax OK, carried out the wavelet conversions, then took them into Paint Shop Pro and converted them into greyscale then recombined them using the RGB tool, then carried out some tweeking with some of PSP's tools. Still fiddling at this stage.

Robster
Woy Woy

[iceman]

Excellent work guys.

Ron - very nice, great sharpness and detail. Probably too flat looking though, and not 3D. Maybe try increasing the levels a tad, and then an overall curves reduction to give it more 3D appearance. Also the hard limb can be removed by using a guassian blur on a feathered selection around limb.

Al - your last version is nice. Perhaps a tad too oversaturated for my liking, still a bit too red/orange. I know what you mean with successive sharpening/despeckling. I usually limit the despeckle to one iteration right at the end with a minor USM after.

Kal - nice attempt. Probably a tad too dark around the limbs and could probably stand a bit more sharpening but a good first attempt.

Robster - please feel free to post your versions, no matter how you think they look. It's a good learning process for everyone!

[rat156]

Here's my attempt.

Opened in Registax. Forced it to stack the three images in B&W. This simulates a luminance channel. Wavelets (4, 5 and 6) until it looked nice. Save as Tiff.

Exported four images to PS. Combined the three original files to a new document as R, G and B channels. Copy this image and paste it over the lumiance image, change blending to colour. Flatten image. Final tweak with Unsharp mask (1.5p 32%) and despeckle. Final boost to saturation by about 14%

Thanks for letting us play with your pics Mike!

Cheers
Stuart

[iceman]

Nice job Stuart. You should try using curves to increase the contrast (drag curves down) - it would give a lot more depth to your version.

Also, be careful of using registax to create the synthetic luminance. If the colour channels are slightly misaligned, registax may not pick up the differences and the resulting L channel may have the features out of alignment, slightly blurring them.

The best way to create a synthetic luminance is to copy the G and B image on top of the R image in Photoshop as new layers, and change the opacity of layer 2 to 50% and layer 3 to 33%. You can then use the move tool, blinking each layer on and off, to align the layers with respect to the features.

Once the features are aligned, flatten the image and now you've got your L channel.

Thanks for having a play.

[Phoenix]

G'day all. I am a newbie to this forum but thought I would give the Jupiter image processing ago. In the spirit of the original post - ie. to share information - I am keen to see how some basic image processing techniques that I use in my own remote sensing profession perform. Apologies in advance if these are processes that you are all already familiar with.

With the objective of keeping the process as simple as possible, the original Jupiter image was processed using a method called subtractive smoothing to create the first image below. Despite its name it is an image sharpening technique. Here a smoothing operation is performed on all 3 bands of the original image (Mean or median kernel, etc - 3x3 or 5x5, etc). The new smoothed image is then subtracted from the original image (hence the name of the operation) - resulting in a 'difference' image that contains high values where there were higher spatial frequencies and values close to zero where there was low spatial frequency in the original image. This difference image is then added back to the original image making the higher spatial frequency areas higher while keeping the low spatial frequency values the same - thus sharpening the image (I have attached a figure that attempts to show the process). Notably, with a little experimentation I multiplied the difference image by a factor of 6 (as it is a subtle sharpener) before adding it back to the original image. I then used a gamma correction for contrast stretching. The result provides no improvement over the excellent results from other contributors, but I thought interesting nonetheless.

The second image attached below was produced using a process called principal components analysis (PCA). Put simply, the process transforms the original RGB image into 3 new bands called components ordered in terms of the amount of information each component carries. That is, component 1 (PC1) contains the majority of information from the original image. The second component carries the next greatest amount of information that the first PC did not explain, while PC3 contains even less information. The significance of this is that the later components (in this case PC3) typically contain more noise - ie. isolating the noise from the 'real' information. Once this is done, PC1 is sharpened using the subtractive smoothing method above, while PC3 (the noise band) is smoothed to reduce image noise. On completion the PCA image is then transformed back to an RGB image where the information has now been 'enhanced' and the noise reduced. Again, the results show no real improvement over others posted but I found it quite refreshing to apply these techniques to planetary remote sensing instead of terrestrial observations.

There are so many terrestrial digital image analysis techniques that one can use, particularly atmospheric correction processes that I am keen to apply. So I hope to make some further contributions. Thanks Iceman for the opportunity.

Cheers, nix.
Happy sensing!

[iceman]

Hi Steve, welcome to IceInSpace! :welcome:

Thanks for having a go at processing the data - especially given your non-standard techniques! The explanations make some sense, and it's great to see terrestrial image processing techniques can be of use in astrophotography.

Your results are very good, too - some of the better processed versions I've seen done.

Thanks again and I hope you enjoy it here at IIS!

[Phoenix]

Many thanks Mike, nice to be a part of IIS and thanks for the encouragement.

Cheers, nix

[beren]

Thanks for letting us have a crack at some top data :thumbsup:

Heres my attempt.....used Astra Image first and worked each channel through the ME filter and experimented with the settings, the green channel allowed more filtering. Then into PS CS2 for some tweaks in levels/curves and a little work in Images Plus. My preference is for a more subtle overall image with saturation and sharpening so I don't know if this translates for others :)

[ballaratdragons]

Hows this for a different approach!

I grabbed all 15 trial images in this thread, saved them as .bmp's in PS, then stacked them in Registax.

For Wavelettes, I used the 3 bottom sliders only: 2.4, 3.1, 3.8

Then back into PS -
Colour Balance: +2 Red in Midtones
2 clicks on 'Sharpen'
1 click on 'Despeckle'. :thumbsup:

This is what everyones images look like stacked together :)

[Phoenix]

Wow.... very nice results beren. Nice uniform contrast over the entire face. What is Astra Image?

Nice work also ballaratdragons. I am a newbie to the forum and I assume Registax combined with PS is the typical software choice for image processing?

Cheers. nix

[joe_smith]

That looks good Ken, you are always thinking outside the square.

[joe_smith]

Steve do you have any links for any atmospheric correction processes info?

[beren]

G'day Steve
If you check out this article about Plantery Imaging in the "Aticles and Projects" section of the site it details how Astra Image another image processing program fits into working your images http://www.iceinspace.com.au/index.php?id=63,306,0,0,1,0 . Well worth it to go over, fantastic tutorial :thumbsup:

[Phoenix]

Thanks beren for the information, I will check the link out and start my learning curve.

Joe, I will have a look at work and see if I can find some appropriate references and post them back here. Atmospheric effects are the bane of most terrestrial remote sensing researchers and while I’ve got a pretty good handle on it, most of the modern techniques use atmospheric models that measure atmospheric attenuation in specific parts of the electromagnetic spectrum that lie outside of human vision (and thus not RGB cameras), which are used to reduce its effect across all wavelengths. Unfortunately, we don’t have these wavelengths to use, so I will have a think about some earlier but very simple and useful techniques that may be of some use for RGB imagery and find some appropriate references – albeit earth observation based. One approach for instance is that blue wavelengths scatter in the atmosphere to the 4th power more than red (or something like that) so the overall ‘brightness’ of each image band could be adjusted relative to this value. Also, the angle between the declination of the celestial object and the zenith of your position is highly correlated with the amount of atmospheric influence; this could also be adjusted for. In the end this may all be academic as the brightness adjustments that everyone seems to be applying to their imagery is kind of doing the same thing, but applying objective corrections might (might!) be more realistic. I am starting to dribble on so I will go and try and find some references for you.

Cheers, nix

[iceman]

Hi Steve

The effects you describe are certainly noticeable while doing high-resolution planetary imaging, such as the Jupiter image in this thread.

Monochrome RGB imaging certainly reveals that the blue channel suffers worst from the atmospheric dispersion and turbulence ("seeing"). We also notice the focal point can change slightly between the red, green and blue wavelenghts - moreso in bad seeing, and moreso when the object is lower on the horizon.

With RGB imaging, the settings can be changed during capture, so things like brightness, gain, exposure can be adjusted to suit the conditions, rather than having to worry about it too much in post-processing.

The post-processing routine usually involves aligning all the individual frames; ranking them in order of "sharpest" to blurriest (based on edge-detection type algorithms), and then "stacking" the sharpest frames to increase the signal to noise ratio.

The better the seeing, the better (more accurate) the alignment and ranking, and the better the seeing, the more frames can be stacked together to give you more signal, smoothing out the image.

More post-processing is then done using sharpening algorithms such as wavelets and typical blur/sharpen filters in photoshop.

Of course the individual colour channels can then still have adjustments applied including levels, colour balance etc.

[Phoenix]

Hi Mike and Joe

Firstly, sorry for such a long posting – I won’t do it again, promise.

Mike, thanks for the overview of the processing procedure in your last post – this has been a big help. I should make it clear that I am under no allusion that I can waltz in and provide any real improvements to your already excellent results. You guys really know your stuff and I have a lot to learn. I doubt now that the more conventional techniques used in remote sensing for atmospheric correction are transferable to planetary RS as they are not designed to cope with multiple frames of the same wavelength band.

These techniques I refer to are either absolute models (that use non-visible wavelengths to measure atmospheric absorption) or relative models (that use ground-based target spectral signatures to correct the effect of the atmosphere ‘relative’ to image derived signatures).

There is one method though that I alluded to in my last post that may be used on DSOs and not dissimilar to the correction technique you use at the pre-processing stage – ie. altering the gain of the sensor for each wavelength - called Dark Object haze reduction (some times called the histogram method of haze reduction) and is described by Pat Chavez (1988) – I can email via PM the paper if anyone is interested.

Basically, It requires the analyst to look in dark areas of an image where there should be no reflectance (earth: deep clear water or shadowed areas; Jupiter image: the dark space to the side of the disk). While not strictly true, these pixels should be zero in value, but of course they often have values significantly >0 due to atmospheric scattering and residual noise of the sensor. These minimum values (different for each band due to different amounts of scattering) can be used as a Starting Haze Value (SHV) – the amount that should be subtracted from each image band respectively to reduce haze. However, Chavez describes how only one SHV is typically used from one wavelength band (e.g. the blue band, as it scatters light the most) and is then used along with a simple scattering model to predict the SHV in each of the other bands. For example:

Simple scattering model:
Very Clear = λ-4
Clear = λ-2
Moderate = λ-1
Hazy = λ-0.7
Very Hazy = λ-0.5

So given the approx. centre wavelength for RGB with a Very Clear sky, then:
B = 0.485um x -4 = 18.07
G = 0.560um x -4 = 10.17
R = 0.660um x -4 = 5.27

Now calculate a multiplication factor to predict haze:
18.07/18.07 = 1
10.17 / 18.07 = 0.56
5.27 / 18.07 = 0.29

So finally, if the SHV in the blue band is say 40 (for 8 bit data) then the predicted SHV for all other bands would be:

Band 1 (blue): 1 x 40 = 40
Band 2 (green): 0.56 x 40 = 22.4
Band 3 (red): 0.29 x 40 = 11.6

Notably, these SHVs should be determined from the raw radiance data (measured in physical units - watts) and not the quantized 8, 10… or 32bit image data. Most RS data like that from Landsat’s ETM+ sensor, is supplied as 8 bit data. All users are supplied with the transfer gain and offset values for each wavelength band – each band will record a slightly different gain – ie. The range of the incoming signal (radiance) relative to the range of the signal output (the image quantization level), which effects contrast. The offset is the recording of energy when there is no energy present, which effects brightness and is equivalent to the SHV of each band.

So after this very long winded explanation, you guys are already altering the contrast and brightness during both the pre and post processing stage – ie. you are basically doing the same thing. The only difference is, in RS they might be applied a little more objectively rather than subjectively (primarily for consistency between image dates for change detection work). Clearly, the objective approach does not make for better visual reconstruction.

I am going to stop over theorizing now and learn more about what you guys do and just start having a go – Bird’s July 25 Jupiter image is telling me just that. Thanks for listening.


Nix.

[Tamtarn]

Great to see all the different versions it's really very interesting

[matt]

Well...this is my first ever attempt at RGB imaging/processing, certainly with good data.

Didn't really do much with it other than a little Registax wavelets processing before taking all three tiffs into PS7 and combining the frames....a little levels and curves, some colour adjustment, some unsharp mask.

That's about it...and it shows.:lol:

Possibly a little too green/blue...but I'm happy with it for a first crack.

I'm stoked that at last I'm finally starting to get the hang of how to use PS...and how to create colour images from composite RGB frames.

I have no idea how Mike gets his final images looking sooooo good.:thumbsup:

[iceman]

Great start Matt, well done.

It's a little on the "pale" side - really needs a boost in saturation and contrast (to my eyes), using the layers->add new saturation layer, and layers->curves.

Grab the middle of the curve and drag it down/right, see how it changes things.

[matt]

Thanks, Mike.

Was thinking exactly the same thing re: the image looking a little 'pale'. Will follow those two bits of advice.

Out of curiosity, why wouldn't you just 'boost' the saturation of the final image in PS, rather than create a completely new saturation layer?

Many thanks.

[Northcanadian1]

Thanks for those files Mike. I have a 9.25 SCT and a ZWO ASI120mm with filters that I have been experimenting with (weather permitting)

Sure are a lot of learning curves for this old guy.

Here is my attempt at processing your excellent rgb's, using wavelets and Photoshop.

Cheers:)

Roger

[Nikolas]

My attempt at the data given. I'm not adept at working with greyscale but hopefully did this some justice.
I just wish we had the seeing conditions in Melbourne that you had back then at your locale.

[Northcanadian1]

Hi :

Thought I would send in one more.

Cheers

Roger

[iceman]

Blast from the past! That brought back memories :)
I'm surprised the original files are still there :)

[ralph1]

Here's my attempt.
Steps:
wavelet sharpening in Registax 6
combine to get colour in Registax 6
Adjust colour balance and saturation in GIMP
Adjust brightness/contrast in Registax 6
Convert to compressed JPEG in GIMP

[Troy]

For a second I thought it was a new image from you Mike, maybe next time ;)
Thanks for letting us play around with your data :D

I did the normal work flow
Wavelets
adjust colour
sharpening
denoise

Troy

[Somnium]

i have had very limited experience with processing planetary data, mainly because my data is no where near as good as this, but here is my attempt. i didn't do any wavelets in registax, i started to but saw that it was downgrading the quality of the image too much so i did all my processing in PS. not as good as yours Mike, but i am happy at how it turned out.