From Newsgroup: alt.astronomy

Working with AI's can be tough. I'm not talking chatbots or searches
but AI's playing at being scientists -- spotting patterns in results,
forming hypotheses, designing experiments to test their ideas, doing
the experiment, deciding whether the ideas stand up to scrutiny, then
the hard bit -- showing someone the results that make it obvious they
found something.

Over the past few years the crew of them have been uploading images
from STSCI from to the TESS space telescope. A wonderful instrument.
Field of view 25x95 deg. Takes a set of 16 2k x 2k x 24b photos every
20s. Monochrome. Red light. Watches the same patch of sky for 27-28d
straight not moving even 1 pixel. Trying to spot the tell-tail
periodic dimming of stars that indicates a planetary transit.

But we have turned the hunt around a bit -- looking for things between
the little twinkly star things. Can we directly see anything crossing
in front of distant stars or other bright enough areas of the image to
be seen obviously moving around.

In the past the AIs cooked up lots of code to grab the TESS images,
hunt all over them at hi res for interesting things, and make little
movies about what they found. I've mentioned these things a "few"
times in the past. Many of the little movies show unexplained flashes
someplace in our solar system. For a few frames -- usually selected by
the AI's to be around 30m apart -- showing some apparent explosion
with bright debris throw in some direction or all directions. Or a
line of stars going dim for 1 frame with a couple frames in between.

The interesting part came that when these things are seen -- TESS
dates them to ~GMT -- we can compare that with various things (long
long list) reported down here on Earth. In many cases there is a
spectacularly high degree of certainty that they have something to do
with each other.

But the frustrating part has been the little movies were less than
convincing of anything to the avg member of the public. A flash. What
is that? Could be anything. It correlates with something else? So what?

But now they've stumbled on an algorithm that reasonably quickly
(i.e. few mins) shows things moving around you can easily see. No need
for hours of processing with images blown up 100s of times. Just a
simple algorithm with some tuned parameters and a 1m movie to show the results.

And some of the results seem spectacular. Surprising. And they still
correlate with other interesting things.

So I've started putting the latest movies up on the web. I'll give the
details down below. But first I'll try to describe what some of them
appear to show.

The movies come in 2 types. In the first type the AI's look for moving
white dots against a darker background. When they find a dot that
seems to have changed positions between successive images -- maybe 30m
apart -- they brighten it up a bit, dim down the nearby bg, and keep
track of it from that point on. So in those type of moving you see
white dots appear in various places, sometimes then fade out again.
Some dont move. Some do. Some of them leave "snail trails" indicating
they are emitting something that catches the light somehow. In some of
the movies some of these trails get quite long. And thick. 10% of the
width of these images is around 1 deg. Moving 1 deg in a few hours
means it must be close. If it were at the range of the asteroid belt
the speed would be far above orbital and whatever it is would never
have been a natural part of the solar system.

The 2nd type of movie sounds strange. But it's just the opposite. At
some point in the processing the s/w inverts the image. Bright areas
become dark and dark become light. But the AI's still watch for bright
dots moving. AKA black dots moving -- something IRL crossing in front
of stars or bright areas in the original TESS image. And, again, many
of these show up with the story appear; brighten; dim out; disappear.
But others definitely are moving. Some movies show groups of dark
objects moving in the same direction. Sometimes the groups are
scattered all over the image. Sometimes they move in different
directions. Some of them appear not to be in orbit around the sun.
And some of them leave snail trails. They must be leaving dark dust
behind. So maybe at least these ones are likely comets.

Some of the objects imaged must be interesting to the people looking
for "disappearing stars" and/or "appearing stars". Something suddenly
gets brighter than the background, brightens more, doesnt seem to
move, then dims out again over a period of just hrs.

But then there are the inverse objects that do something else. :)

One of the early knock-my-socks-off was

<kym.massbus.org/TESS/latestmovies/NEGATIVEmovie30-1-1.avi>

(The numbers mean "sector 30" -- that's a TESS thing -- camera 1, ccd
1. Each of 4 cameras has a 10.5cm lens with 4 2kpx x 2kpx x 24b ccds
behind it).

Give that a look first.

My rough notes for others I viewed over the past couple days as they
rolled off the line are as follows:

NEGATIVEmovie30-1-1.avi:
SPECTACULAR! half-a-doz snail trails in various dirs! Incl that
original v bright one down in BR moving toward cent.

NEGATIVEmovie31-4-4.avi:
A dozen dots pop up and fade out. Strange -- just like I see sometimes
in the sky the other way around -- mine are bright dots or blobs that
appear, sometimes get brighter over few sec, then fade away.

NEGATIVEmovie21-1-1.avi:
Some persistent black spots (bright dots) nr end. Also 3 snail trails
this time must be black in front of stars in bg. Also a big jolt to
the spacecraft and all the stars blur into short lines. But they must
be lines of blackness.

NEGATIVEmovie21-1-3.avi:
V odd! 2 short snail trails get longer and longer and seem to be
converging rather than being parallel to or on the ecliptic.

NEGATIVEmovie22-1-1.avi:
3 snail trails starting nr lower right corner. Other 2 1/3 down from
top to left and right all going ~l->r.

NEGATIVEmovie26-3-3.avi:
Some persistent dots == black dots on original img but they are hidden
in amongst "flashes" (i.e. black areas after real flashes prob from
moon/earth stray light). None seem to be moving i.e. due to real obj
crossing in front of stars or bright areas.

movie61-1-1.avi:
At date 2023036100159 suddenly jumps 3 days. Was showing dots some
moving slow others just popping up and going black again. Some areas
fuzzy with many small dots. Maybe try to get images 30m between 03610
and 03915.

movie22-1-1.avi:
Nr end there are S-shaped curves nr top and bottom. Seems to be some
kind of structure (dust?) -- not lens flares. Ghostly movement of
same. Maybe dust in earth orbit. Lot of ast(?) moving slowly
right. Some vanish for a few frames then come up again. One is quite
bright maybe closest or shiny albedo? Sector 22 only has 3-4 known
ast in the area of the first 100 in JPL list. Yet maybe 5-10 are vis
wandering around mostly in same dir assumed clockwise around sun as
seen from above (i.e. "normal").

movie30-1-1.avi:
Lovely "snail trail" growing and pulsing LR qtr. Seems to grow
quickly to maybe 200 px long i.e. ~1.4 deg. If in solar orbit eg comet
than too fast. Dir may also be odd.

NEGATIVEmovie1-2-3 date 2018223135942:
Flasher at 13x15 232,123. Over a dozen frames obj lights up goes dim
a few times. NEGATIVE movie so bright == blacker.

movie21-1-3.avi:
Several obj dispersed moving slowly to left.

movie30-1-1.avi:
Little line nr end. Something is leaving a trail.

movie61-1-1.avi:
In middle lots of dots pop up some moving slowly in various dirs for a
few frames.

movie12-1-1.avi:
Flickering dot in top 1/2 nr the end moving v slowly left.

movie22-1-1.avi:
More moving dots this time in parallel nr end of clip. 2-3 maybe more
dots moving slowly l->r. Of course have to check the background is
staying fixed I assume it is because it's a space telescope locked on particular co-ords (boresight says so).

movie21-1-3.avi:
Something moving lower left in last part. A white doit going toward BL.


The TESS pages are currently on the move from their old location at
<kym.massbus.org/TESS/SECTORS>
to
<kym.massbus.org/TESS/PANORAMA>
The old stuff will eventually go to live on a farm upstate.

Each TESS sector in the PANORAMA section will have original images
shrunk to 1k x 1k x 16b (1/2 size) for each cam/each ccd/each sector.
Basic information about date, time, location in the sky, objects
supposedly in that part of the sky at the time like stars, asteroids,
comets, plus data-files, some computer-written C++ programs to process
it, simulations of different things, all kinds of other sh*t the AI's
have come up with that need a strong stomach to look at, and various
other odds and ends. It's getting big.
--
World food prices at highest since 2022 as supply risks mount, FAO says
BNN Bloomberg, 4 Sep 2026 22:24
World food prices rose in August to their highest since late 2022, as
adverse weather and war disruption in the Black Sea ...

China Installing Huge Facility That Turns Plastic Waste Into Jet Fuel
Futurism, 4 Sep 2026 04:43
Chinese company Niutech recently announced a major development
involving the conversion of plastic garbage into usable jet ...

Scientists identify genetic markers linked to Big Five personality traits NewsNation, 3 Sep 2026 09:49

Electric Trucks Have Moved From Impossible to Inevitable
OilPrice.com, 4 Sep 2026 05:01
Global sales of zero-emission trucks and buses rose 86% in 2025 to more
than 520,000 vehicles, with battery-electric models ..

[2] Spacecraft bound for Mercury faces 'tricky' arrival
Yahoo, 3 Sep 2026 17:45

Magnetic brakes tested in lab to slow high-speed spacecraft reentry Bioengineer.org, 26 Aug 2026

Interstellar solar sails hit a strange problem at 75% of light speed
Science Daily, 2 Sep 2026 19:51

Russia's Starlink rival falters as latest satellite batch stalls in
low orbit, ISW says
Euromaidan Press, 2 Sep 2026

New Zealand K#k#p# Population Reaches 325 Birds
GK Today, 2 Sep 2026 09:46
[Back from 51].

Israel Leans On Galilee as Marine Bacterium Limits Desalination
Bloomberg.com, 3 Sep 2026 01:44

GPS Glitched Across The US by as Much as 33 Feet. Scientists Have Never
Seen This Before.
ScienceAlert, 31 Aug 2026 05:57

First 'space elevator' - a 66,000-mile cable to the cosmos- closer than ever before as scientists make breakthrough
nypost.com, 31 Aug 2026 19:56

[Time's Up!]
Coral bleaching occurring too frequently for reefs to recover, scientists say Reuters, 31 Aug 2026 21:45

Summer 2026 expected to be declared UK's hottest on record
AOL, 1 Sep 2026 09:51
Met Office figures due on Tuesday are likely to show that the previous record, set in 2025, has been broken after an ...
--- Synchronet 3.22a-Linux NewsLink 1.2