Juno perijove 8, September 1, 2017 Options

[Phil Stooke]

Here's an image to compare with the Juno view, almost the correct orientation. It's rendered, not an original image.

Phil

[Sean]

Here is a detail from Gerald's PJ08_118...



Upscaled, stretched, processed, edges filled.

[Sean]

Here is a portrait from Gerald's PJ08_108...



Upscaled & processed.

[Gerald]

125-fold Perijove-08 animation on Youtube.
MP4 video, single overlapping scenes, and stills zipped for download.
Actually, except the first and the last scene, I've rendered each scene twice, based on two consecutive raws.
The stills are JPG, but should be in good quality. PNG would have been too large, regarding upload time and volume.
Note, that the stills of consecutive scenes overlap by one still. In order to obtain a seamless movie, skip the first frame of each scene, except the first scene. However, for frame interpolation you might need the first respective still of each scene.

[Sean]

Thanks for sharing these Gerald, there is so much to see in this perijove pass! You've done a grand job blending the different images together also. For me, the only 'issue' is the snapping of the camera path, although this is much less problematic now that the images have smooth blending.

[Sean]

I shared Gerald's PJ08 sequence over on Twitter and the 150dpi 11m x 2m image was actually printed out to 8.5m x 1.3m fabric!

Tweet thread

[Sean]

Here are some upscaled, processed & retimed clips from Gerald's PJ08 animation. They are still works in progress but they do feature a new re-timing method which reduces the appearance of the stepping issue when the animation switches between projected textures. The nested clips are 3 levels deep and the motion vector calculation never has to synthesize more than one frame each time. The previous solution was not nested and the software had to guess up to 10 frames which gave rise to motion artifacts.



This clip features the transition between PJ08 121 & 122


PJ08 108 - 109

[stevesliva]

I shared Gerald's PJ08 sequence over on Twitter and the 150dpi 11m x 2m image was actually printed out to 8.5m x 1.3m fabric!

Tweet thread

Awesome... so where is it hanging?

[Sean]

It's hanging in Lund, Sweden.

[Bjorn Jonsson]

After making a few improvements to my Juno-related software I now at last have some processed PJ8 stuff. This is from the PJ8_118 data, first approximately true color/contrast versions:




And the enhanced versions:





A subset of the relevant metadata:

IMAGE_TIME = 2017-09-01T22:03:23.474
MISSION_PHASE_NAME = PERIJOVE 8
PRODUCT_ID = JNCE_2017244_08C00118_V01
SPACECRAFT_ALTITUDE = 12143.0 km
SPACECRAFT_NAME = JUNO
SUB_SPACECRAFT_LATITUDE = -28.5406
SUB_SPACECRAFT_LONGITUDE = 329.8497
TITLE = POI's: Dalmatian Zone, Eye of Odin
Resolution at nadir: ~8.2 km/pixel

[Sean]

Beautiful work Bjorn.

Here is a montage of Gerald's perijove 08 clips set to Brahms

[Bjorn Jonsson]

BTW, note that image 114 ("Collision of colours"/"Sharp edge") was taken in lossless compression mode; several people had asked if images near PJ were being degraded by lossy compression. I'd be curious to hear if people thought it made a significant difference.

Yes, this makes a significant difference in low contrast areas. Here is a heavily processed image showing the area near the nadir in PJ8_114:



I was able to apply a more severe contrast stretch and unsharp mask than I usually do - and without having to worry about compression artifacts. This is a good thing because this is a lower contrast image than most of the Juno images. In a 'normally' compressed image the low contrast regions would have a 'contoured' appearance and/or squares of uniform (or near uniform) brightness and color would be obvious (Gerald posted an example earlier). The lower part of the image is more heavily processed than the upper part since the contrast there is lower in the original data. The difference from ICT compressed images is visually biggest in the blue channel - it is darker than R/G and frequently looks rather uniform or even noisy in ICT compressed images (it has even occurred to me to try to generate synthetic blue images by combining low frequency brightness variations from the blue channel with high frequency information from the green channel).

Would it be possible to generate a version of the raw image showing what it would have looked like with typical lossy compression parameters? If this is possible I would be interested in applying the same processing to that version of the image as I did in the image above.

By the way I have noticed one minor issue with all of the very high resolution images I have processed (I haven't processed many of these though). Here is an example:



There are very narrow gaps between the framelets. This only occurs near the nadir and only in the red channel. I don't think this is an error in my processing. It's easy to fill these gaps and they are very narrow so I consider this a minor issue.

[Bjorn Jonsson]

And here are two more images processed from PJ8_114. They show the northern and southern limb:



I processed these images more heavily than I usually do, both because of the low contrast in most of the areas covered by PJ8_114 and also because of the lossless compression. The image at left is especially interesting and appears to show long, narrow and elevated clouds plus associated cloud shadows and vertical relief near the horizon.

And here are approximately true color/contrast versions of all three images. The last one isn't exactly impressive and reminds me of Saturn because of the low contrast. I suspect the generally low contrast and fuzziness of these images is mainly because the contrast really is lower here than it is farther from the equator; if this is correct the very high resolution (~2.3 km/pixel at the nadir) is not the main reason for the low contrast.





And the associated metadata:

IMAGE_TIME = 2017-09-01T21:49:21.459
MISSION_PHASE_NAME = PERIJOVE 8
PRODUCT_ID = JNCE_2017244_08C00114_V01
SPACECRAFT_ALTITUDE = 3472.2 km
SPACECRAFT_NAME = JUNO
SUB_SPACECRAFT_LATITUDE = 10.1385
SUB_SPACECRAFT_LONGITUDE = 319.5073
TITLE = POI's: Collision of colours, Sharp edge
Resolution at nadir: ~2.3 km/pixel

[MahFL]

I shared Gerald's PJ08 sequence over on Twitter and the 150dpi 11m x 2m image was actually printed out to 8.5m x 1.3m fabric!

That's pretty cool.

[Bjorn Jonsson]

Here is an image from the departure sequence:



This is the best view of the Great Red Spot from the departure sequence. I processed it mainly as a 'sanity check' to confirm that my color processing (still) results in global color that looks about right.

This is image PJ8_161 from the departure sequence. North is up. The image was obtained on 2017-09-02 07:30:25.957 when Juno was about 800,000 km from Jupiter's center.