AI meets Jupiter

 Warning: boring details regarding planetary image processing ahead.  

Planetary imaging is best done these days with a web cam capturing short video segments with as many frames as possible.  software then combines the sharpest frames, discarding frames distorted by atmospheric turbulence.  the "stacked" image is that further sharpened.  too much sharpening produces unsightly artifacts enhancing noise inherent in the imaging process, rather than fine detail.  too little sharpening leaves a blurry image.  recently "AI" techniques have been used to sharpen images or "denoise" sharpened images.  the results have been very impressive, but controversial.  

Here's an image of jupiter with very conservative sharpening.  the image is nice and smooth, but blurry:

Jupiter 12/16/2023 05:20 UTC

next up is an image with aggressive sharpening applied.  fine details are more evident, but there are clearly sharpening artifacts, sometimes referred to as "noise":  

next a conventional "denoise" has been applied, removing the artifact, blurring the image slightly, but preserving some of the sharpening for a smoother sharpened image:

here's what topaz denoise AI did to the sharpened image:

the image is far more aesthetically pleasing than the others, both silky smooth and razor sharp.  this is what has generated the controversy.  there are 2 main objections:

1. the image is too smooth.  
-personally i think this is silly. hubble images of jupter are smooth and sharp at this scale, why not mine?  

2. the AI generates structures that are not real.  
-now this is a problem.  a number of advanced imagers have demonstrated AI processed images with structures that look real, but aren't actually there.  pretty damning criticism, thus all imagers submitting images for scientific work are avoiding AI processing.  i think you can make the case that the false AI structures are the result of pushing the AI processing too far, but now that the objection has been raised, AI processing is widely condemned in the planetary imaging circles.  

Image Details:

i'm submitting images to ALPO so now include image detail in the frame for the submission.  

cloudy nights thread on the topic--mostly civil, only one post was banned (but you can still see it's residue)

more jupiter images 2023

Renowned amateur astronomer Christopher Go appealed to the OCA for images of Jupiter so I've been cranking them out and working on my technique.  

Jupiter 11/26/23 05:53.3 UTC

Jupiter 11/26/23 and Io

Jupiter 11/29/23 05:51.9 UTC 

Jupiter 12/6/23 07:10.6 UTC RGB
"chimney open"

imaging details:

details mostly provided in image annotations

new laptop is allowing much higher capture rates without dropped frames, especially at larger frame sizes ;) 20,000 800x800 frames in 90 seconds!

the downside is the video files are much large so the terabyte hard drive fills up after 3 or 4 imaging sessions :(

processing in autostakkert also takes much longer

tried stacking unsharpened images in winjupos, then sharpening the combined image

the result was horrible

so will probably be doing more sharpening prior to winjupos combine

tried oversharpening final images, then used topaz denoise to remove sharpening artifacts

the result was very impressive

working on planetary autoguiding with firecapture.  having trouble with RA oscillations occasionally throwing the planet way out of the frame (editing these frames out with virtualdub).  

Jovian navigation

Jupiter is the fastest rotating planet in our solar system, with a day lasting only 10 hours.  thus Jupiter's Great Red Spot (aka small salmon spot) is only visible for a few hours each day.  so if you see it one night, it will be visible 4 hours earlier or 6 hours later the next (multiples of 10 mod 24).  this makes it tough to catch night over night.  it's visible for about 2 hours and can be followed rotating across the face of the planet during that time.  

Jupiter
11/22/23 05:22 UTC
2023-11-22-0522_5
CM1 60.8, CM2 59.9, CM3 102.6

I recently noticed my planetarium software wasn't getting the position of the great red spot quite right. so off to the rabbit hole of jovian navigation...the short version is that clouds in jupiter's atmosphere do not rotate a constant rate.  

a planetary frame of reference was created, roughly defined by the position of the great red spot.  however, the great red spot's rate of movement across the surface varies with time, so it's position has to be intermittently updated relative to the planetary reference system.  

the following is largely verbatim from references listed below:

Because Jupiter is not a solid body, its upper atmosphere undergoes differential rotation. The rotation of Jupiter's polar atmosphere is about 5 minutes longer than that of the equatorial atmosphere (wikipedia).

Three systems are used as frames of reference for tracking planetary rotation, particularly when graphing the motion of atmospheric features. System I applies to latitudes from 7° N to 7° S; its period is the planet's shortest, at 9h 50 m 30.0s. System II applies at latitudes north and south of these; its period is 9h 55 m 40.6s. System III was defined by radio astronomers and corresponds to the rotation of the planet's magnetosphere; its period 9 h 55 m 29.710s is Jupiter's official rotation (GJI).

"most (but not all) higher-latitude atmospheric disturbances including the Great Red Spot and the three white ovals move only slowly in system II." (GJI)

the Red Spot was at Jovian System II longitude 46° in November 2023 and continues to drift 1.75° per month, based on historical trends noted by JUPOS (sky and telescope)

note the planetary image above is annotated with 3 different central meridian (CM) positions, one for each system.  

historical trends and recent observations are listed at JUPOS the organization associated with winjupos, a sophisticated planetary mapping program designed for advanced amateur astronomers to track planetary features.  ironically the GRS position in the winjupos planetary surface texture is way off.  

GRS longitude (Sy. 2)
JUPOS database

in the image above and my 11/12/2023 06:45.8 UTC image, i measured the GRS longitude at 46 in winjupos.  sky safari lists it at 50, accounting for the difference in the image and planetarium software.  

references:

https://en.wikipedia.org/wiki/Jupiter

On the rotation of Jupiter
R. Hide
Geophysical Journal International, Volume 64, Issue 1, January 1981, Pages 283–289, 
https://doi.org/10.1111/j.1365-246X.1981.tb02668.x

https://skyandtelescope.org/observing/interactive-sky-watching-tools/transit-times-of-jupiters-great-red-spot/

historical trends and recent observations JUPOS.org:
http://jupos.privat.t-online.de/

imaging details:
eastbluff, CA
11/22/23 05:22 UTC

CM1 60.8, CM2 59.9, CM3 102.6

celestron 11" Edge HD

baader IR/UV block filter
ZWO ADC

ZWO ASI 290MC (one shot color camera)
11x90 second captures 05:15-05:30 UTC
FPS (avg.)=27-110
Shutter=5.545ms
Gain=351 (58%)
Histogramm=85%
downsampled to 66 % :(
software:
firecapture
autostakkert
winjupos
registax
photoshop

 

Io transiting jupiter's great red spot

Jupiter put on a nice show for west coast observers last weekend: Jupiter's moon Io and it's shadow transited the planet's surface just as the great red spot (GRS) was rotating into view.  

Io and shadow transiting Jupiter's great red spot
11/12/2023 06:45.8 UTC
CM1 331.7 CM2 46.6 CM3 86.7 CLat +3.8
Winjupos combination of 4x90 sec captures with independent processing of Io

The dark circle to the left of the great red spot is Io's shadow.  Io hovers over the top portion of the GRS. Close inspection of Io reveals detail on the jovian moon: dark patches top and bottom with a bright equatorial region.

Io and shadow transiting Jupiter's great red spot
11/12/2023 06:35.7 UTC
CM1 325.5 CM2 40.5 CM3 80.6 CLat +3.8

Animation of Io transit (larger version below):

Io and shadow transiting Jupiter's great red spot
11/12/2023 05:19-8:05 UTC

imaging details:
Eastbluff, CA
11/12/23 
still shot 06:35.7 UTC
CM1 325.5 CM2 40.5 CM3 80.6 CLat +3.8

winjupos shot 06:45.8 UTC

CM1 331.7 CM2 46.6 CM3 86.7 CLat +3.8
celestron 11" Edge HD
ZWO ASI 290MC (one shot color camera)
2x barlow
90 second captures at 5 minute intervals

05:19-08:05 UTC
FPS (avg.)=102
Shutter=2.976ms
Gain=351 (58%)
Histogramm=78%

animation scaled to 50%

i'm guessing capturing at higher histogram level facilitates detail on the jovian moons

software:
firecapture
autostakkert
winjupos (for CM measurement only)
registax
photoshop

started with poor seeing, which fortunately improved during the GRS transit

very dry so dew was not an issue

unfortunately, collimation was way off. wondering if i focused from the wrong direction, introducing mirror flop.  might be time for a crayford focuser.

on the plus side, it's rare for me to capture detail on a jovian moon, so a win despite poor collimation.  

PS full scale animation processed for frames with better seeing:

Io and shadow transiting Jupiter's great red spot
11/12/2023 05:19-8:05 UTC

jupiter near opposition 2023

jupiter is bright in the evening sky now.  well positioned for northern hemisphere viewers, it transits just before midnight at an altitude of almost 70 degrees from my location.  

here's jupiter on halloween night, shortly before opposition, the seeing wasn't as good as i'd hoped, but managed to salvage something:

  

Jupiter 11/1/2023 06:53 UTC
little red spot
just above center?

imaging details

eastbluff, CA

11/1/23 06:43-0658 UTC

CM1 37.8, CM2 196.6, CM3 233.7

celestron 11" Edge HD

ZWO ASI 290MC (one shot color camera)

2x barlow

ZWO ADC

baader IR/UV block filter

best 5  of 90 second captures from 06:43-06:58 UTC

FPS (avg.)=27-110

Shutter=3.174ms

Gain=351 (58%)

Histogramm=32%

inadvertantly put the ADC after the barlow

resulting in massive over magnification on a night of mediocre seeing :(

interestingly, learned that displaying the very large image during capture on my laptop monitor slowed the capture rate

viewing at 25% improved the capture rate

downsampled to 37.5 % :(

software:

firecapture

autostakkert

winjupos

registax

photoshop

Annular eclipse at the Albuquerque Balloon Fiesta 10/14/23 part III: super-annular eclipse montage and animations

 You may recall that there was a supermoon in late september.  the corollary to this is a "micro-moon" crossing the face of the sun during the solar eclipse at the subsequent new moon, resulting in an annular eclipse rather than a total eclipse 😢

or, to look on the "bright side" a super-annular eclipse with the widest band possible.

in creating a montage of the partial phases, i viewed the annular phase and decided on a clock-like arrangement with small images of partial phases forming a ring around the annular eclipse.  after meticulously arranging the partial phases, i looked at what i had done...not exactly what i was going for, but i'll take it.

flower power!
annular solar eclipse and preceding partial phases 10/14/23
balloon fiesta park, albuquerque, NM
(click for full size)

more conventional montage:

annular solar eclipse and preceding partial phases 10/14/23
balloon fiesta park, albuquerque, NM

partial phases:

annularity animation:

the disturbance mid capture above is likely due turbulence from the burners firing off in the field at annularity :(

Annularity 10/14/23
Balloon Fiesta Field
Albuquerque, NM

if  you look at the final images of annularity, you'll note that the moon does not contact the sun evenly.  mountain peaks contact the sun first:

3rd contact
10/14/23

by offsetting each image (captured at 4 frames per second) by 1 pixel, one can generate a magnified image of the lunar surface:

magnified moon mountains
lower left
Cf. APOD

bigger partial phases
click for full size

field view/filtered view composite:

click for full size

sky view of above:

click for full size

Imaging details:
10/14/23
Albuquerque, NM
nikon D850
AF-S NIKKOR 300mm f/4E PF ED VR
Baader solar film, Fotga IR/UV cut filter
iso 100 1/1250-1/2000 sec exposure
Vello Shutterboss Version II Timer Remote Switch
skywatcher star adventurer tracking mount

Samsung galaxy S22 ultra

for field shot and filtered cellphone image

filter

the final 2 composites are a combination of the cell phone view of the field and nikon filtered images of the sun, scaled down to the sky view size and positioned roughly in the appropriate location in the sky

Annular eclipse at the Albuquerque Balloon Fiesta 10/14/23 part II: Ring of Fire

a few clouds and a lazy contrail on the horizon made for an interesting sunrise, but didn't impact the rest of the eclipse (click on image for full size):

sunrise balloon fiesta park
abluquerque, NM
8/14/23 06:37:27

first contact:

first contact (upper right)
09:13:22

partial phase:

09:30:12

bye sunspots:

10:14:12

The announcer called out annularity and all the grounded burners in the field fired up.  note the annular reflections below the sun:

annularity, balloon fiesta park
abluquerque, NM
8/14/23 10:36:11

RING OF FIRE: there was a lot of talk about a "ring of fire" eclipse, but no images really show it.  here's my attempt, I digitally overexposed this image to simulate an unfiltered image of the sun (which would melt my camera sensor) to portray an actual "ring of fire"

ring of fire

here's a more conventional portrayal.  bullseye with the camera--my location was dead center eclipse path:

annular solar eclipse
balloon fiesta park
abluquerque, NM
10/14/23 ~10:36:32

cellphone:

filtered

broken ring:

image rotated horns up
10:39:09

Next up: animation and montage

Imaging details:

10/14/23

Albuquerque, NM

nikon D850

AF-S NIKKOR 300mm f/4E PF ED VR

Baader solar film, Fotga IR/UV cut filter

iso 100 1/1250-1/2000 sec exposure

Vello Shutterboss Version II Timer Remote Switch

skywatcher star adventurer tracking mount

Samsung galaxy S22 ultra

for field shot and filtered cellphone image

filter

eclipse image shot with solar filter at 30x

processed in photoshop

    Arriving early for the pre-dawn activities allowed me to dial in my polar alignment for my tracking mount, only adjusted my alignment once.

    While i tested my exposures repeatedly at home, to be sure my images were not clipped, as the sun rose higher in the thin albuequerque air the face of the sun became slightly clipped, losing the subtle surface granulation. fortunately, i caught it relatively early and decreased the exposures from 1/1250 s to 1/2000.  if you look closely at the partial phase above, you'll note a lack of subtle granulation across the bright face of the disk (09:30:12), which can be seen in the subsequent image (10:14:12)

    The nikon D850 has a built in intervalometer which is great for meteor showers and overnight sequences, but awkward to adjust on the fly.  there is software to connect a smart phone via bluetooth or wifi, but it disconnects every time you switch apps and takes >30sec to reconnect.  less than ideal for the eclipse experience unless you use a dedicated tablet.  

    I chose to used the vello shutterboss version II timer remote switch, a wired solution that can trigger the shutter manually without shaking the camera and be used as an intervalometer that can be rapidly adjusted.  i chose 4 minute intervals for the partial phases and 1 second intervals for annularity.  for the 3rd contact i held the manual trigger down with the camera set for continuous low speed shots, having determined 4 fps is the fastest my camera will write full frame raw files (8288x5520 ~53 meg) to the chip.  it can do bursts much faster but only for a few seconds.  

    I found these an excellent focus aid -o-o-

    In terms of processing, the auto color balance with the IR filter and baader film was all over the place.  By trial and error, I came up with something close to white by setting the temperature to 7100 and tint to -40 for raw conversion.  this still gave a slightly non-uniform color white in the center and slightly red red at the edges which was acceptable, though I desaturated the white images above.  not sure why the color was not uniform.  either the camera sensor is non-linear or the solar color is non-uniform.  here's the "natural color"

10:14:12

    for images with a substantial amount of the central solar disk showing, i applied a high pass filter at 1.5 radius soft light, then again at 2.5 radius hard light 50% blend (excluding the solar and lunar edges). this brought out the sunspots and surface granulation.  color images colorized at 25 hue, 100% saturation, -20 lightness