Juno perijove 9, October 24, 2017, near solar conjunction Options

[Gerald]

It's time to start a new topic for Juno's Perijove-09.
Due to solar conjunction, data downlink is delayed several days, but things appear to look good thus far.
Data downlink started on October 31, and if everything continues well, it should be a matter of a few days, at most, until the majority of the raw Perijove-09 images will become available. Due to the incremented available storage for JunoCam, we may get a sufficient coverage to render a pole to pole -- well, almost pole to pole -- fly-over reconstruction, despite the difficult communication near solar conjunction. I'll have a try at least, over the next few weeks.

[Sean]

More power to you Gerald, next week is going to be busy & fun!

[mcaplinger]

The PJ9 images are on missionjuno (although the front page doesn't say this yet.)

[Gerald]

PJ-09, quick & dirty version without SPICE nor shape model, therefore close-up portions not yet properly RGB aligned.
The enhanced versions will take another few hours.

[Gerald]

This is the 10-fold reduced testrun of the version I'm currently rendering:

[jccwrt]

Gerald, thanks for making the quick and dirty versions available so quickly. I have processed image #105, which is a family portrait of Jupiter with Io and Europa.


Perijove 9 Departure - JunoCam

A small amount of unsharp masking was applied to enhance some of the detail, which reveals what look like real albedo features on Io's surface. This image is simply enlarged by 3x:


Ganymede Perijove 9 detail - JunoCam

This is Io's southern hemisphere but I'm not entirely sure which features should be visible. Maybe Tarsus Regio?

[Gerald]

Great to see you finding out which moons those are, while I've been rendering and uploading the enhanced versions of the close-ups to the missionjuno site.

Here the JPG version, going to be distributed over several posts:

Perijove-09, #75, #76, #79, and #79:



As usual since Juno has been orbiting Jupiter, the close-ups start with images of the north polar region.

[Gerald]

PJ-09, #80, and #81, with close-ups of the northern "folded filamentary regions" (FFRs), mostly cyclonic turbulent storms known for their increased probability to show lightnings:

[Gerald]

#82, and #83:

[Gerald]

#84, and #85:

[Gerald]

#88, and #89, now we are already above the southern hemisphere:

[Gerald]

#90, with the white anticyclonic oval A5, if the prognosis has been correct:

[Gerald]

#91, and #92, approaching the south polar region:

[Gerald]

Approaching the south pole along images #93, #94, #95:


The transition into the departure sequence with images
#97 and #98, and with some gap the slightly overexposed #105:


Image #105 shows some artifacts near the GRS induced by my patching algorithm for camera artifacts. The algorithm uses nearby colors to fill in invalidated regions, which works pretty well in most cases, but not near the GRS.

[Sean]

Perijove 09 sequence, based on Gerald's work...

672 Megapixel


168 Megapixel


10k

[Sean]

Some processed & cropped hilites...

[jccwrt]

Striving towards an enhanced but still naturalistic processed version. This is my take on #084, which includes a photobomb by Io and Europa.


Southern Hemisphere - JunoCam

A very close look at Io shows a circular reddish spot at about the 9 o'clock position. There are some subtle albedo features around the 2 o'clock position as well. My guess (and this is only a guess), is that they are the Pele plume deposits and the region around Issum Patera, respectively. Here's a 4x enlarged image. (To view the uncompressed image, click through and add '/sizes/o' to the end of the url)


Io - JunoCam

[Brian Swift]

Some Videos:
PJ09 Approach Movie
JunoCam CCD's view of #91

[Sean]

New pass on Perijove 09 [G.Eichstadt] Upscaled, processed, patched & reframed...

PJ09_76



PJ09_78



PJ09_79


PJ09_80


PJ09_90


PJ09_91


*update*
PJ09_81

[Sean]

PJ09_82


PJ09_92


PJ09_89


PJ09_Sequence1


PJ09_Sequence3

[Gerald]

Perijove-09 inbound RGB images, reprojected, and enhanced. Illumination adjustment is based on PJ-06 TDI-2 images, and may be biased.
I'm elaborating approach and some departure PJ-09 images in more detail, since they are likely to be the only visible light Jupiter images of a major part of its surface available during solar conjunction.

Departure images are more challenging to process due to spacecraft maneuvers, and due to the position of Jupiter approaching the left margin of JunoCam's fov, where geometric calibration is particularly tricky. I'll see over the next few days, how far I can get, hopefully over the first two Jupiter days (i.e. 20 hours) after PJ-09, at least.

I might find time to infer a PJ-09 specific illumination model before PJ-10, but I'm not yet sure.

[JRehling]

These pictures are convincing me that there will eventually be enormous scientific payoff from the Juno imagery, despite the official disclaimers. The still images alone speak to the dynamics taking place, and in concert with the deep-looking radiometry and lower-resolution images from Earth that provide greater temporal coverage, seemingly must speak volumes to the atmospheric dynamics at large and small scales.

And also: They are stunning! You amateur image wizards are working wonders.

[Tom Tamlyn]

The Atlantic posted an article today by Marina Karen titled The Photoshoppers Behind Dreamy Jupiter Photos, which quotes members Gerald Eichstädt, Seán Doran, Björn Jónsson, Roman Tkachenko, Emily Lakdawalla, and Jason Major (listed in the order mentioned).

https://www.theatlantic.com/science/archive...ictures/546146/

The article includes a passage that echoes JRehling's observation about the scientific value of these images:

JunoCam wasn’t designed for scientific purposes—its sole mission is, quite literally, to take pretty pictures—but the Juno team has used the images to better understand the meteorology of gas giants, in our solar system and beyond. “Jupiter can be considered as representing a population of gas giants, likely a widespread population of celestial bodies in the observable universe,” Eichstädt said. “Understanding Jupiter means understanding non-negligible portions of our universe.”

[Bjorn Jonsson]

Here is a montage of several versions of PJ-9 image 93:



A: An approximately true color/contrast image.
B: An approximately true color/contrast image where the effects of global illumination have been removed. This reveals dimly lit features near the terminator.
C: Same as B but in addition, the colors and contrast have been exaggerated and small scale details sharpened to better reveal various features.
D: Same as the previous version with the addition of a latitude/longitude grid. This reveals the location of Jupiter's south pole.

A subset of the metadata:

IMAGE_TIME = 2017-10-24T18:39:10.279
MISSION_PHASE_NAME = PERIJOVE 9
PRODUCT_ID = JNCE_2017297_09C00093_V01
SPACECRAFT_ALTITUDE = 79834.2 km
SPACECRAFT_NAME = JUNO
SUB_SPACECRAFT_LATITUDE = -78.1351
SUB_SPACECRAFT_LONGITUDE = 279.7929
TITLE = Southern timelapse
Resolution at nadir: ~54 km/pixel

[Roman Tkachenko]

At the edge of the abyss (Jupiter at 4467 km)

[Sean]

Some PJ09 hilites after Gerald's work...

79


80


89


90


91


91


91

[Kevin Gill]

My JunoCam workflow still isn't 100% satisfactory, but was able to work these up over the weekend after realizing I had nothing good from Perijove 9. I'm using a script which simply grabs each strip and stacks them by color band. It doesn't yet do any spice-based alignment or reprojection, but I'm hopeful I can figure that part out. The script also does a histogram equalization and grayscale contrast adjust. I take the output into Photoshop, do some manual alignment, brightness adjustment, and sometimes a HDR toning for contrast and structure. Finally, I make final adjustments for tone, presence, sharpening, & noise in Lightroom.

Attached thumbnail(s)

 

[Kevin Gill]

A couple more shots I was able to go back and reprocess. The color isn't quite there and I am focusing on enhanced contrast.

Attached thumbnail(s)

 

[Roman Tkachenko]

Io as seen by Juno

[Kevin Gill]

A couple more from various points in the mission

Attached thumbnail(s)

 

[Gerald]

In the meanwhile, John Rogers released two reports about Perijove-09:

Juno at Perijove-9 (Part I): Global maps, and
Juno at Perijove-9 (Part II): What the close-up images show.

Despite the difficult observational conditions, we were able to create a global cylindrical map.
Of particular interest is a newly formed south tropical disturbance near the GRS.

[nprev]

The Atlantic posted an article today by Marina Karen titled The Photoshoppers Behind Dreamy Jupiter Photos, which quotes members Gerald Eichstädt, Seán Doran, Björn Jónsson, Roman Tkachenko, Emily Lakdawalla, and Jason Major (listed in the order mentioned).

https://www.theatlantic.com/science/archive...ictures/546146/

A bit belated, but congratulations to you all for this article and recognition! The Atlantic is a widely-distributed US magazine. The article also name-checks our 'no-frills' Forum.

[Gerald]

Well, UMSF has been playing an important, if not essential role in bringing people together.

---

Here a link to a Perijove-09 flyby animation.
The link is submitted to missionjuo, too.
I've implemented a "nadir mode" for cylindrical views. This helps to make the movie's horizontal center continuously pointing to Jupiter's center. Vertical displacement is allowed in order to fit with JunoCam's fov. For the youtube version of the movie, I've chosen half second blendings, since the alignment of the scenes isn't quite perfect, and would hence result in doubled features over too long periods of time.

I'll try to upload stills, and mp4 files onto the junocam.pictures webspace later tonight (CET).

[Sean]

Another excellent effort Gerald! I think this sequence really sells how close Juno gets to Jupiter. I hope the tweening software can handle the closer frames without too much trouble.

[Guest_avisolo_*]

This is so cool Gerald! Here is my humble attempt at a more poetic rendering of your animation:
https://vimeo.com/244473772

[Gerald]

Thanks! I like that sound track of Vangelis, Avisolo.

The stills are online since a few hours; the online version is compressed to 1.7 GB of JPG files, I've used 98% quality in order to save upload time and storage volume, and to maintain good quality.
I've rendered two movie versions, one with short blending for Youtube upload, and another one with long blending resulting in an overall more continuous impression, but with considerable feature doubling.
The individual scenes before blending are provided as MP4, as well.
The MP4 scenes have been rendered from BMP files, the same files the JPG stills are derived from. Each of the BMP files has been rendered directly from the respective raw files in one monolithic step using double precision floating point arithmetics.

[Guest_avisolo_*]

Thanks Gerald!

I've replaced the original video with the files you uploaded, makes for better viewing:
https://vimeo.com/244473772

[Gerald]

Beautiful! I've recommended it via email to several members of the JunoCam team.

[Sean]

It appears that these animation renders are reversed/mirrored. Can you confirm this Gerald? At least when compared with the initial processed stills.

[Gerald]

You are right, the cyclones are winded in the wrong sense. I should have noticed that. I'll try to mirror the movie. The stills will remain as they are.

[Sean]

It's easy enough to correct, I just wanted to make sure you knew.

[Gerald]

These things can happen, especially when I'm working around the clock. Thanks for your checking!

The flip-corrected version is on Youtube, now.
The previous version on the missionjuno site hasn't been approved yet. So, I've removed it and added the revised version, instead.

[Sean]

Here is an early pass on Gerald's video [flipped, rotated, variably retimed & processed] set to 'Orphic Hymn' by Jóhann Jóhannsson



60fps version on Youtube

[Sean]

Tinkering with re-scaled stills from Gerald's flyby...




*update*

...and another...


( uploaded corrected, flipped version )

[Sean]

This is a processed, masked blend of projected images [ 90 & 91 ] from Geralds' flyby...

[scalbers]

Amazing planet and excellent work from Sean and Gerald. I'm gradually getting better at figuring which belts/zones we are seeing when we're at somewhat lower latitudes. Post #31 helps with this.

[Sean]

Someone on Twitter sent me this...



I think it really ties the room together!

[djellison]

Someone on Twitter sent me this...

Damn - that's a nice dining room table.

[algorithm]

Walnut? or birds eye Maple perhaps, nah, too dark, which brings me onto that!!! couch....

The slightly fuzzy, wrapped canvas isn't too bad though.

[Sean]

Haha... thanks Doug, I love coffee coming out of my nose!

[Jay]

Hey everybody, first time posting to the site after learning of it's existence the other day. I've been getting back into space-oriented side projects lately and wanted to add Juno image processing to the list. I haven't dug too far into the "proper" methods of processing these images yet but did a quick test last night regardless. I was really surprised to find that compiling the color separations into one image seem to leave the result pretty far feeling from true color.

I'll be looking at more methods to bring that to a better accuracy but for now I wanted to share what I came up with, on the left is what I got just from compiling the red green and blue channels and on the right was my first run at altering it to try and bring out some of the detail. I didn't do any drastic colors, I was mostly trying to sort out how to bring out the clarity of the features from the North Pole image I used. I would love any feedback you could provide and plan to post more as I continue to refine my process, and will probably also experiment with some other styles just for fun.

[kgreene]

Hi Jay (and everyone). I was in your position about a month ago. I wrote an automated processing pipeline to process the raw images for a class, though it still needs some improvement. Most of the info I needed I found either in technical papers or on this forum.

Did you decompand the image data? I found that helped a huge amount, though the color channels still need calibrating afterwards. If so, you probably just need to weight the color channels. I found several in these forums, the one I ended up using was one Gerald suggested in one of his papers: http://www.unmannedspaceflight.com/index.p...st&id=42174

Linear radiometric weights applied to decompanded raw data are 0.82 for red, 1.0 for green, and 2.17 for blue. This may induce a slightly greenish cast.

Actually, I was thinking maybe we need a thread specifically for how to process junocam images? I spent a long time reading through all the other threads and little bits of info were scattered all over the place. It would be nice to have them all grouped together and we can all help improve it.

Kevin

[Gerald]

In order to remove most of the greenish cast, I've been using 0.88 for red since PJ-06 for enhanced images. But this might undergo another refinement later.

In general, I'm not sure, whether there is a fixed "how to". It depends on your purpose. I've written almost 10 MB of code for various purposes of JunoCam image processing and related tasks, almost everything from scratch. Might be the code could be shortened to 5 MB with sufficient effort in "refactoring", but I've still a long list of TBDs after four years of development. I might double the code next year in order to cover some of the items on the (ambitious) list.
I'm interested in developing and understanding all technical detail about camera calibration, processing, data reduction, evaluation, and beyond, independent of possibly existing partial solutions, but that's probably not the recommended way for people, who just like to create beautiful images, nor for professionals who need to reduce specific project costs and risks.
The semantics of the code overlaps with that of NAIF/SPICE and ISIS3. So, if you are happy with "the standard", you might consider to base your work on these libraries and tool sets. The JunoCam extension for ISIS3 is work in progress since quite a while, and might be released next year, for those who are used to work with ISIS3 in a UNIX environment.

Regarding threads, I guess, that Candy would be happy to learn about possible extensions of the missionjuno website. One of her primary objectives is "Science in a Fishbowl", as far as that's possible without running into science publication or privacy issues. "Evaluating JunoCam images with ISIS3" might become one of the considered topics.

[mcaplinger]

The JunoCam extension for ISIS3 is work in progress since quite a while...

There is a beta version of ISIS3 that supports Junocam. I was told not to distribute it "outside the team" but since everyone here is on the Junocam team I'm not sure what that means. If you're interested PM me with details about which platform you want it for -- it's available for the 64-bit version of Mac OSX 10.11 and various 64-bit flavors of Linux. It's worth noting that ISIS3 has a very very steep learning curve so this is not something you should ask for casually -- you might start by downloading the production version of ISIS3 to see if you can figure out how to use it on a simpler case.

As for a unified thread about Junocam processing -- I'm not sure there's a unified thread about anything on UMSF, but it's a good idea in theory. The thread over in http://www.unmannedspaceflight.com/index.php?showtopic=8353 is an example, perhaps overly specific to the standard SPICE formalism. Some amateurs have documented their workflows in fair detail, some have not (which is fine) -- I am always curious to see how people are doing what they are doing.

[kgreene]

The code I developed was for a computational photography course, so for a variety of reasons I used only related methods. So I only did 2d alignment methods to register the framelets, which turned out OK but obviously was not perfect. It seemed like I would just be able to find the homographies for perfect framelet alignment but I wasn't able to get acceptable/consistent results doing it. I was hoping to be able to open source the code for people to play around with but I need to check guidelines related to the class first.

The Juno Software Interface Specification document says they perform a white balancing such that a white surface has a value of 10000. (For planetary targets). I did not attempt to implement this.

I was thinking that a thread that could summarize information about the spacecraft and specifically the instruments (including junocam) would be useful. It could encompass "official" pipelines as well as unofficial ones by giving enough data to help people implement it themselves or modify existing code. For example, originally I wanted to try to visualize data from JIRAM or MWR and merge them with junocam imagery.

I have a list of links to a variety of documents and posts that could be helpful to start.

Also, were the flat fields ever released? I found mention of them in the calibration report but could never actually find them.

[Jay]

Thank you all for sharing the information, I'll be sure to read through the provided links when I get home tonight. I did not decompand the image, to be honest I'm a bit like Alice tumbling into the rabbit hole, not having realized how deep the work with these images goes, but I love it. I focus mainly on 3D work but want to be able to process and include real imagery for future projects (and its also a blast) so I'm starting at the ground floor with processing actual spacecraft imagery.

That said I haven't had as much fun with Photoshop in a long time as I have with my first runs with Perijove 9. I can see though Gerald why many people base their work off yours as a starting point. Really loving the community and work this place provides, so thank you all for that.

[Gerald]

There is a beta version of ISIS3 that supports Junocam. I was told not to distribute it "outside the team" but since everyone here is on the Junocam team I'm not sure what that means.
...

I've access to the beta, in principle, but no system it is running on. A member of USGS considered providing a VM. This would simplify installation "infinitely".
But I've loads of other tasks. So, I may not be the first one to test the ISIS3 extension for JunoCam.

@Jay, you can really learn lots of interesting things from JunoCam image processing, once you are ready to dive into this adventure.

[mcaplinger]

I have a list of links to a variety of documents and posts that could be helpful to start.

If you want to post that or start a new thread it seems completely reasonable. It could go in http://www.unmannedspaceflight.com/index.php?showtopic=8143 or http://www.unmannedspaceflight.com/index.php?showtopic=8353 -- the latter is misplaced IMHO because "Image Processing Techniques" is intended to be non-mission-specific.

Also, were the flat fields ever released? I found mention of them in the calibration report but could never actually find them.

As I've said previously, the flat field is more complex than just putting up an image. Our ground flat fields consist of readouts of the entire active part of the sensor, only 128-line pieces of which per filter are sent down in flight. And then with TDI active the image moves across the sensor and blurs out the blemishes. All of this could be documented in a self-contained fashion, but I just haven't had the time. I've attached a normalized 8-bit version of the flat (IIRC this appears as a figure in the Junocam paper) and a binary blemish map of the RGB part of the sensor that gets read out (the latter is what we are currently using in our processing flow to repair blemishes, but it's fairly quick-and-dirty.)

[Bjorn Jonsson]

A bit late (since the PJ10 images are now available) but here are my versions of PJ9_081 ("northern coverage"). Approximately true color/contrast versions and then enhanced versions:







And some metadata:

IMAGE_TIME = 2017-10-24T17:32:07.116
MISSION_PHASE_NAME = PERIJOVE 9
PRODUCT_ID = JNCE_2017297_09C00081_V01
SPACECRAFT_ALTITUDE = 10107.7 km
SPACECRAFT_NAME = JUNO
SUB_SPACECRAFT_LATITUDE = 41.8417
SUB_SPACECRAFT_LONGITUDE = 220.0255
TITLE = Northern coverage
Resolution at nadir: ~6.8 km/pixel

[Bjorn Jonsson]

There is an area in one of the enhanced PJ9_081 images in my previous post that I find especially interesting:



Here I get the impression that we are seeing thin haze or tenuous and largely transparent high altitude clouds (at several locations indicated by arrows) overlying a deeper and darker layer of clouds. Of course this could also be something different.

[Sean]

Here is an update on PJ09_80 [G.Eichstadt]


Detail


Someone sent me a picture of a print they made of my previous rendition...


I'm not sure how healthy it is to stare into Jupiter's soul too long.

[Sean]

A second look at PJ09_90 [G.Eichstadt] detail...

[Sean]

Artist Lucy West recently posted her take on PJ09_80 so I decided to create a sequence showcasing the inspiration for the final piece...

In order; Juno-Gerald-Seán-Lucy



Lucy's artwork is 36" x 36"

Lucy West Studios

[Gerald]

When you try a cross-eye or a parallel-eye, you'll notice, that Lucy's geometry is slightly different from the original. So, that's an actual painting, not just a reproduction with some printing or morphing technique.
It's a long time since I've seen a naturalistic complex painting so similar to the original, but still with a personal artistic and stylistic note, an art already rare half a century ago.

[Sean]

PJ09_90_new pass

[PhilipTerryGraha...]

I'm not sure if Sean or Gerald are aware (they probably are), but to those who are not aware, this image is a finalist in Wikimedia Commons' Picture of the Year contest!

[Gerald]

Thanks a lot! I feel honoured, of course.
I've actually been busy with identifying certain subtleties in images like these:

(derived from PJ09, #89, and #84)

[volcanopele]

JIRAM data from PJ9 is now in the PDS. Slowly working on the Io data from this encounter and PJ10...

[volcanopele]

Still a work in progress but I can't help but share this:



This is a stack of 6 JIRAM images from PJ9. You can make out some surface features: large scale stuff like Media Regio and Tarsus Regio, and some smaller scale features like Karei Patera and Fjorjynn Fluctus, and maybe a ring around Nusku Patera.

Reason for edit: Wrong patera with possible ring

[JRehling]

Wow, that's about 35 distinct locations in one hemisphere. I didn't realize that Io showed so many distinct hotspots at one time.

[volcanopele]

48 distinct hotspots from my count (50 from PJ10 in the other thread, though two maybe the same hotspot, but the source is extended...).



(color roughly equates to brightness [brightest yellow -> red -> teal -> blue dimmest], but no corrections have been made for emission angle or sunlight contribution)

[nprev]

Astonishing, and astonishing work, VP.

[Sean]

PJ09_81_GE/SD Detail

[Sean]

PJ09_90 GE/SD Detail...

[Sean]

PJ09_81_Detail [ GE/SD ]