Almost 900 MB of "Marble Movie" data now available: https://www.missionjuno.swri.edu/junocam/processing

The newly released Marble Movie data cover 10 Jul to 28 Jul. For the most part, images are taken every 15 minutes. Until 16 July it alternated RGB and blue, after that it's just RGB. Every two days beginning July 17, there is a methane filter image. There are two gaps in the data, one between 13 Jul 17:00 and 14 Jul 02:15, and the other from 14 Jul 13:00 to 14 Jul 20:15.

Note that unlike the approach movie images, which had to be full-spin 82-frame images, now that we're in orbit we are able to use "nadir mode" which times the image acquisition based on s/c attitude knowledge of when Jupiter is in the FOV. This reduces the data volume a lot, but has the unhappy side effect of splitting the planet across two green framelets for most of the RGB images, which makes processing a bit harder depending on how one does it.

That's the output of my first out-of-the-hip calibration test run, starting with a wild guess of the camera parameters:
Click to view attachment
The second iteration run should provide better-centered images.

A very first taste of an animated version with 6 frames showing the Great Red Spot:
Click to view attachment

I've uploaded the first few hundred processed RGB still frames (a zip with about 250 MB).
After the sequence up to image 2449, covered by the zip, Juno's spin axis changed a bit. This required an adjustment of the processing parameters.
It's the kind of x-coordinate coverage with which I hope to be able to narrow down the remaining degrees of freedom in my camera model.
The processing of the remaining color images is still running. I'll upload the results later, in a few hours.

A derivative version from the last frame in that zip, with colours auto-balanced and north up:

Click to view attachment
Credit: NASA / JPL-Caltech / SwRI / MSSS / Gerald Eichstädt / Maksim Kakitsev

I've added a preliminary AVI version of the "Marble Movie" (about 24 MB).
The first half is good. In the second half, there is some misalignment, wobbling and flickering. To treat this, I'll likeley need to apply a more time-consuming reprocessing.

Edit: The remaining preliminary stills of the AVI are now online.
I've skipped image 2450, since it requires parameters specific to this image, which are TBD.

Wonderful animation Gerald, seeing the orbiting moons made it special for me

Thanks Paul, for finding time to have a look at the Juno mission!

This enhanced version of JunoCam's first "methane" image of Jupiter
Click to view attachment
shows bright regions near the poles, as well as bright bands closer to the equator.
The image is derived from a crop of the raw image
Click to view attachment
and enhanced as sketched in this animated gif:
Click to view attachment
The brightest pixel is in the northern (lower) bright band near the equator. Since it's only one pixel, it may be noise. If it's a real feature, it's not in the neighborhood of to the Great Red Spot
Click to view attachment
This gif contains enhanced frames of immediately before and after the methane image, a strongly enhanced version of the methane image in-between, and the Great Red Spot image:
Click to view attachment

The first 81 color frames of the "Marble Movie", linearized to enhance cloud band structure, and cropped around Jupiter, as a lossy gif preview:
Click to view attachment
The stills of this gif are derived from the zipped stills I've uploaded yesterday.

... I'm preparing an AVI of better quality for the "Marble Movie" RGB images until JNC image #2449.

The AVI, white-balanced version:
Click to view attachment

Interestingly, the file is of better quality than the gif, contains a lot more (602) frames, but it's of smaller file size.

This looks really good. You might consider rotating it so north is up.

Finding better ways of enhancing the moons without bringing up the noise background is something we spent a fair bit of time on. I found that masking the planet and then linearly stretching the background worked OK, though if you make the mask too large you cut off the moons near the planet and if you make it too small you end up with a ring of stray light around the planet.

BTW, the planet gets very small but marble movie becomes interesting later on as the spacecraft approaches the equatorial plane and you can see the moons' motion become more and more linear.

Thanks!
Here the north-up version:
Click to view attachment

I may implement and test some moon-enhancement versions later today. The log-version I've posted for the Jupiter Approach movie was intended to ensure, that no information gets lost on the level of the darkest colors. But specific for moon enhancement, there are several options. Later today or tomorrow, I may post the best solutions I can find in the short run.

Yeah, the countdown to PJ1 is running. Exciting, and still a lot to prepare...
Good to know, that the motion of the moons is going to become even more interesting. So it appears useful to run some more tests with the images already available.

I think the public image upload capability at missionjuno is now live -- https://www.missionjuno.swri.edu/junocam/processing and push the upload button in the upper right corner.

An attempt to upload the AVI failed. So, I've provided a link to the Youtube version Emily has been so kind to load up.
One needs to pay a little bit of attention not to happen to confuse the upload for amateur observations with the upload for JunoCam products.

Re moon-enhancement: I've completed brainstorming several approaches, and am now going to implement. The first step will be related to a breadth-first algorithm to cover a single Voronoi cell with onion shells in order to determine the distance from the bright part of Jupiter for each pixel. This follows a fill algorithm for Jupiter, which assigns a zero-distance from the pixels to the bright Jupiter area.
The onion shells will also be stored explicitely to allow for quantile calculations within one or a sequence of shells.
The analysis of the shells in terms of percentiles, and the number of the shell, will go into dark bias (for noise and stray light), gamma, and stretch functions. These functions will be parameterized to allow to play with. The functions should be continuous in order to avoid visible ring artifacts.

The technique worked as expected, at least for the first 50 test images:
Click to view attachment

Hot pixels are also enhanced, but not yet filtered out.

602 frames of juggling Jupiter as AVI.

Wow! This looks extremely promising and should look awesome once you have figured out a way to get rid of the hot pixels/noise.

Yes, JunoCam takes great images! All you need to do is processing them appropriately.
I'm going to implement a crude "a-posteriori" filter for removal of the hot pixels and eventual CRs on dark background. It will simply look for single-filter bright pixels without neighboring pixels with a non-low value on a different color channel within some radius, and set them to black, or maybe to a point noise filtered value.
This exploits, that bright noise usually occurs in only one color band, and it considers some misalignment or low signal at the same time.

----

I've seen a question on Reddit about the black feature in the Marble Movie.
There are several stills in the Marble Movie showing shadows of moons. Here an example sequence:
Click to view attachment
The image is a crop of a synopsis of 600 images I've submitted to the missionjuno website for approval.

Very nice I think you should add "SwRI" to your list of image credits (between JPL and MSSS).

For the approach movie we converted the moons to grayscale because otherwise the color fringing from slight misregistration was too distracting.

Thanks! Re crediting: I'm a little confused, since the missionjuno website credits only to NASA/JPL/MSSS. But I may include SwRI in future runs again, as I've done before.

Here is the cleaned version of Marble Movie part 1 AVI (602 RGB still frames), together with the synopsis of 600 frames, as submitted to missionjuno.

I'm not yet quite happy with the dimming of the Moons near Jupiter in the AVIs, since in the meanwhile I've been able to demonstrate, that this isn't a necessity:
Click to view attachment Click to view attachment

However, the revised method requires accurate alignment between subsequent images, in order to allow for good differencing results.
This is not identical, but related to better color alignment. To be able to improve this, for the Marble Movie stills beginning with #2450, and for accurate PJ images, I should focus again more on fine-tuning the camera calibration. I'm still far from having inferred the accuracy feasible with these images. It might even be possible to do this fully automated.

Re grey scale: This would be possible. However, it's to be applied after cleaning (postprocess) from hot pixels and CRs. But I'm unsure, whether it wouldn't be better to work with RGB as a test for RGB alignment accuracy, and as a motivation to improve it. If the resuts are urgently required for public release, it's something else. I may add this option, just in case.

Thanks again for your interesting observations and advices!

Well, that's an excellent point. I'll ask for clarification.

As a fallback option, here a grey-scale version (for the moons only, Jupiter remains in color) of the AVI, and according stills as zip, for the first 602 color frames of "Marble Movie".

I'll continue with trying, whether the same processing is able to produce at least a draft version of the movie starting with JNC image #2450.
As far as possible in parallel, I'll look for remaining minor software glitches, implement the differencing method for the moons (which is related to motion enhancement), and work on inferring constraints for camera calibration based on distant Jupiter images. I'll adjust the credit line in the processing software, if/when it turns out to be appropriate.

Hi Gerald - Thank you very very very much for posting your marble movie on the missionjuno website. It's great! Regarding the credits, NASA/SWRI/MSSS is preferred, but it's not a big deal. Some contributors are just using their name. (If we put out a press release then we will add the institutions on the front.) Thanks again for contributing! Candy

Hi Candy - Thank you very much! I'm happy, that you're enjoying my processings. It's a big pleasure to see - and help - your long-planned vision coming true.
-- I'm intending to submit only selected products (maybe a fast draft and a finalized version) to the missionjuno website in order to avoid to be too dominating, but I'll try to cover all non-black raw images at least once. If you happen to see intermediate products I'm posting on UMSF - or on the junocam.pictures webspace - you're interested in, don't hesitate to use them. To some degree, and possibly with some delay, I may also be able to provide modified or derived products according to specific wishes; just let me know.

-- I'll re-insert "SwRI" to the credits in future renditions.

Draft AVI of parts 1 and 2 of the "Marble Movie" with grey-scaled enhanced moons, sometimes filterd out by noise filter.
I'm working on several of the issues.

Spectacular!

Parts 3 and 4 of the raw Marble Movie images are available on the missionjuno site.
If my computer doesn't go to produce small white clouds of smoke, I hope I'll be a little faster with a draft version this time.

Edit: File naming convention changed. Filename contains image number in hexadecimal and iso time. Applied filters are provided in attribute "FILTER_NAME" of json file.
See also json attribute "SOURCE_PRODUCT_ID" as reference to filename.

On a Windows/DOS operating system, this batch executed in the same directory as the unzipped raw image files renamed the files for me to the usual JNCE raw filename convention:
Click to view attachment

A first draft of "Marble Movie" until frame #4676 (parts 1 to 4).

The sequence ends near apojove, when Juno crosses the plane of the Galilean satellites.

Sorry about that. Those are the names we receive the data files as, while the JNCE names are more or less the same as the equivalent PDS products. The former names were used by mistake. We'll correct that as soon as possible.

I've identified the cause of two flickerings in the movie:
- C4192 doesn't show Jupiter, and
- C3242 violates the divisibility of the height by 128.

---

In the meanwhile I've been able to find a good ("level 1") parameter set for part 3 of the Marble Movie.
I'll prepare a post with more detail in a few hours.

"Level 1"-ish images of Marble Movie part 3, and an according preliminary moon-enhanced "level-2" rendition.

I've used 16 images near the beginning of part 3 to infer four almost-best-fit parameters for the Jupiter color channel centroid alignment for each of the images, within a 7-parameter camera model family, after choosing three parameters as constant within three chosen degrees of freedom.
Click to view attachment
For rendition, I've used the three chosen constants, and for the other four parameters roughly the respective mean.

Chosen and inferred parameters of the calibration run as CSV:
Click to view attachment
The four inferred parameters are a function of the three chosen constant parameters. The method allows to choose any three of the seven parameters freely, and to infer the other four from the images.

---

In the meanwhile, a similar calibration run for part 4 is performed. The parameters look rather similar to those for part 3. I'll try to provide the according level 1-ish stills of part 4 later today.

---

Part 2 is more tricky, and might require recalibration for individual images over some parts of the sequence.

I've added the level 1-ish stills and preliminary level 2 AVI animation of a refined calibrated version of part 4 of "Marble Movie".

Remarkable is an occultation of Jupiter by Ganymede, combined with Io's shadow around image #4046.
Annotated preliminary level 2:
Click to view attachment
(edit: submitted to missionjuno for approval)

Level 2 without annotation:
Click to view attachment

Animated gif of 8 level 1 images:
Click to view attachment

For the model parameter oscillations starting near images #2449 and #2529, I've determined a parameter set for rgb aligning the Jupiter centroids for each image:
Click to view attachment
Determining the parameters took about 5.5 minutes per image.
Here the CSV version:
Click to view attachment

With these time series I'm pretty sure to be able to render well-rgb-aligned images for the tricky end of Marble Movie part 1, and start of part 2.

Here the preliminary level 2 version of this difficult wobble sequence as AVI:
Click to view attachment

For those of you who may be wondering what makes some parts of the movie different from others -- we're not doing anything different as the movie progresses, other than changing the number of frames and the mix between RGB and CH4 a little as we experimented. Nor do I think the camera intrinsic parameters are changing very much. The main thing that's changing is the spacecraft pointing and spin rate. The spacecraft nominally has the HGA pointing at the Earth, so its pointing slowly evolves as the Earth moves (the tracking is done stepwise rather than continuously.) Every once in a while there's an adjustment to maintain the axis and spin rate. And there have been some fairly large pointing excursions from time to time for calibration activities for the other instruments. The latter is the largest effect -- if you look at where Jupiter is in the raw frames you can see it jump across the field by a few hundred pixels and then jump back. All of these pointing changes are visible in the spacecraft C kernels.

There may also be some timing slop in the reported image timestamps relative to when the first frame also starts exposure, and perhaps some drift in the interframe time spacing, which is set by a crystal oscillator in the camera. I'm still working to characterize that, but it only amounts to a few pixels of offset.

Preliminary level 2 "Marble Movie" Parts 1 to 4. RGB alignment issue resolved, moon dimming near Jupiter pending.

Next, I'll work on a zip of the according level 1 stills. Might be completed later today, or early tomorrow.

Level 1 rgb stills, parts 1 and 2, most of which revised for RGB alignment since the first preliminary version.
(Parts 3 and 4 uploaded before.)

Edit: I've added a zip with processing parameters. Sorry for the filenames only matching in respective substrings; making everything fully formally consistent would have consumed more time than I can currently spend for this detail; filenames should be sufficiently unique to avoid confusion.

I must admit I don't understand all the math behind how you do these, but this is great - it's very stable and the moons are really bright.

I've got a similar issue with the moons of Uranus, which are pretty dim compared to the planet, so I might need to enhance them also at some point - maybe some kind of logarithmic brightening (?).

Have you thought of posting these to Reddit? There are ~4000 subscribers on https://www.reddit.com/r/junomission that I'm sure would appreciate these.

Can't wait to see the rest of the approach sequence!

As soon as I'll be at a quality level I'm happy with, I'll submit the results to the official missionjuno site, as a central location for the Juno project. But if anyone likes to use the intermediate versions as they are, they may work with them. I'd think level 1, parts 1-4, are ok by now; level 2 can be improved; that's what I'm going to elaborate.
There are people at Reddit who observe the missionjuno site, so I'm rather sure, that some of the processings will eventually show up there; actually part of the Marble Movie has already been discussed at Reddit. Maintaining a Reddit thread is currently beyond my limited ressources.

Re moon enhancement: I've used a (2d) distance estimate from Jupiter to gradually change the enhancement functions within a ring-like zone around Jupiter. It's implemented as an intentional side effect of filling a Voronoi-like cell around Jupiter defined as distance zero.

To obtain an image like this
Click to view attachment
from a level-1 image, I'm first calculating a Voronoi cell filled with onion-layer-like equidistance shells, roughly visualised by this image:
Click to view attachment
The distance function (including its explicite representation as an indexed array of equidistance lines) is the basis for gamma-, stretch-, greyscaling, and percentile functions.
I'm using percentiles within each equidistance line to estimate background level and brightness maxima neglecting outliers.

But I think, that adding kind of motion enhancement to the method bears the potential to provide better results. Exploring this refined approach is my immediate next goal.

Thanks for the thumbnails, Gerald.

Here is a browse page to all the Marble Movie data released so far.

First, I'm grateful, that you've found a usable way to manage this large number of images, Emily!
So, I barely dare to ask about the source of the "range" data, since I don't see a connection to the "SPACECRAFT_ALTITUDE" attribute of the json files.

Woops, an error crept into my spreadsheet. Sorry...stand by and I'll fix.

...fixed! Thanks for pointing that out.

A "magic" threshold in the brightness of small/pointlike objects in an image is whether or not the apparent size of the object equals about one pixel. If so, you shouldn't run into any problems with its brightness – the outer planets moons are usually (not always) of albedo comparable to the planet, so if the planet is visible in an image, a moon (or other object) that covers about a pixel should be, too.

If the moon/object is much smaller than one pixel, then you have to increase the exposure to make the pixel(s) that it falls into bright enough to show it… and that may overexpose the planet.

When I take pictures of Jupiter and Saturn, I can resolve the Galileans and Titan easily, and they display the correct colors. Rhea is just about one pixel. But the moons smaller than that, I have to expose the image more and turn Saturn into a blob of glare. In these cases, you need to do something tricky to get the moons and planet to look nice in the same image. But for moons larger than a pixel, there's no extra effort required… it just works.

Since Emily's uploaded version of the preliminary Marble Movie to youtube appears to get some considerable attention, I've notified it to the missionjuno site.

A short sequence of motion-enhanced Marble Movie images showing moon transits:
Click to view attachment
It avoids the dimming of the moons near the bright Jupiter.
At the center of the very right you may notice a faint moon. I'm working on a more sophisticated noise filter, which will hopefully be able to let the faint moons pass, and at the same time filter out the considerably brighter hot pixels, and statistical stray light and background noise.

Parts 5 and 6 of raw Marble Movie images are available.

And here the superfast service, rgb images of part 6 as preliminary level 2 avi:
Click to view attachment

Movie of part 5 and level 1 stills will take a little more time, and will probably become available later today.