Lightnews — Scholar-powered news

Aleksey Shipilëv

@shipilev.bsky.social

1.5K followers 120 following 310 posts

In love-hate relationship with machines. Currently: OpenJDK, AWS. "Trust me, it's really me" backlink: https://shipilev.net/#social

shipilev.net

Posts Media Videos Starter Packs

Pinned

Aleksey Shipilëv @shipilev.bsky.social · Nov 16

Periodic reminder: builds.shipilev.net contains lots of useful binaries for JDK preview, debugging, and testing. Most if not all builds trigger and complete within 1 hour after the commit. While BlueSky remains cozy, reasonable requests can go here:

2 7 28

Aleksey Shipilëv @shipilev.bsky.social · 4d

I contend that only closet masochist perfectionists are truly happy in astrophoto hobby, because you need to get twenty things exactly right to get the perfect result. It is quintessentially man-vs-Universe story, often literally. So when we complain, that means we are actually having fun!

Aleksey Shipilëv @shipilev.bsky.social · 4d

16Kx16K image done. All of this for naught (scratch that) to confirm that this thing is absolutely diffraction-limited and gives no visible resolution improvement against the 9Kx9K image. Need larger aperture to get over 50 megapixel limit.

Aleksey Shipilëv @shipilev.bsky.social · 4d

Stitching mosaics by hand is pretty meditative, though. Like a puzzle where pieces do not exactly fit due to field curvature and exposure oddities.

1 3

Aleksey Shipilëv @shipilev.bsky.social · 4d

Grid Hollow Moon confirmed.

Aleksey Shipilëv @shipilev.bsky.social · 4d

Argh, this hobby is half frustration. The mosaics was miscaptured with clipped edge. Why could not it clip on the DARK edge? Argh. So close to 200 megapixels. Argh.

3 5

Aleksey Shipilëv @shipilev.bsky.social · 5d

My personal goal is to get 100+ megapixel full-disk Moon. This one is barely 60 megapixels large, and it already bumps against the hardware limits. Need much larger aperture SCT/RC for planetary work, oh no...

Aleksey Shipilëv @shipilev.bsky.social · 5d

3x Barlow pushes 80/560 refractor and 2.0μm pixel camera way below the diffraction limit for pixel-perfect resolution. You can still recover some by post-processing tricks. This photo is 9Kx9K, likely very printable! Downsampled 4x here for preview:

2 7

Aleksey Shipilëv @shipilev.bsky.social · 5d

If you ever find yourself unsure where you are in relation to the Universe, look up, find Saturn, and check it looks perfectly straight. ᵗʰᵉ ᵒᶠᶠᵉʳ ᶦˢ ᵗᶦᵐᵉ⁻ˡᶦᵐᶦᵗᵉᵈ ⁻⁻ ᶠᵒʳ ᵇᵉˢᵗ ʳᵉˢᵘˡᵗˢ ᵖᵉʳᶠᵒʳᵐ ᵗʰᵉ ᵖᵒˡᵃʳ ᵃˡᶦᵍⁿᵐᵉⁿᵗ ᵒᶠ ʸᵒᵘʳ ʰᵉᵃᵈ

Aleksey Shipilëv @shipilev.bsky.social · 5d

Using narrowband filters (Baader Solar Continuum 7.5nm in this case) does improve Moon contrast quite a bit, even though it pushes the exposure times way up. A doublet with heavy chromatic aberration gets fairly monochromatic light :P [Downsampled 4x here for preview:]

1 4

Aleksey Shipilëv @shipilev.bsky.social · 6d

Trials for high resolution full Moon mosaics are in progress. You don't need a cooled camera when it is already cold outside 🥶.

Aleksey Shipilëv @shipilev.bsky.social · 6d

Granted, doing 3x Barlow is way out of spec for my visual (OD 5.0) solar film. For this to work, one needs a brighter (OD 3.8, unsafe for visual) film or Herschel prism.

Aleksey Shipilëv @shipilev.bsky.social · 7d

Trying out Baader Solar Continuum 7.5nm filter. Pros: contrast improves quite a bit. Cons: narrowband means longer exposures, with 3x Barlow it goes into 50ms, which turns lucky imaging into REALLY LUCKY imaging. Left: what SDO sees. Right: what my scope sees.

Aleksey Shipilëv @shipilev.bsky.social · 8d

Also looked around my non-commercial software stack (GIMP, AutoStakkert, Siril, etc) that I used before switching to PixInsight, and donated to all those projects as well. Checked my SharpCap license is renewed. Beginners do not have to pay for software to enjoy their hobby.

Aleksey Shipilëv @shipilev.bsky.social · 8d

Gaaaah. Happened to me once, kinda: I have been using a mismatched exposure/gain master dark for half of my photos. It was messing up *stacking* accuracy on lots of deep-sky stuff. It was very cathartic to reprocess with a proper master dark. Looking forward to more beautiful stretching!

Aleksey Shipilëv @shipilev.bsky.social · 8d

This is happening. I have been using PixInsight for deep-sky and most of the solar/lunar work during the 45-day trial period, and it did not disappoint. Best €350 I spent on astrophotography stuff so far.

1 7

Aleksey Shipilëv @shipilev.bsky.social · 9d

Oh man, I agree, these *do* like Moires! Are these showing up only on BSky, or you see these patterns right in AsiFitsView outputs in your local image viewer too? Friends do not let friends resample with bad tools :)

Aleksey Shipilëv @shipilev.bsky.social · 9d

Astrophotography things I believed in half a year ago: expensive acquisition hardware. Things I believe in now: expensive post-processing software, plus post-processing skills. Same raw data, drastically different outcomes.

Aleksey Shipilëv @shipilev.bsky.social · 9d

Or, maybe posting full-res images triggers BlueSky to drastically drop the JPEG quality to compress it more, which exarcebates blocking? That might explain why I never seen it in my pictures: I always 2x downsample and compress at lower quality for these posts, hoping for verbatim reproduction.

Aleksey Shipilëv @shipilev.bsky.social · 9d

Huh. Looks too erratic and squarely to be a Moire? Also, would it imply the original image has some sort of periodicity? On the closer inspection, this looks more like JPEG recompression artifacts, maybe you got the dimensions "just right" that blocking got way worse than it should be?

Aleksey Shipilëv @shipilev.bsky.social · 9d

Upping the post-processing game:

1 4

Aleksey Shipilëv @shipilev.bsky.social · 9d

Nice shot! There are visible grid patterns, though: debayering or sharpening artifacts, perhaps?

1 2

Aleksey Shipilëv @shipilev.bsky.social · 10d

30 minute total exposure of M31 (Andromeda Galaxy) + M110 companion from the bright Potsdam city center.

1 5

Aleksey Shipilëv @shipilev.bsky.social · 10d

Me: "Astrophotography in the city sky is so frustrating, so much light pollution". Also me: <deploying the photo rig literally between two streetlights>

Aleksey Shipilëv @shipilev.bsky.social · 12d

You have your OCA on file, correct? Consider filing the PR with JRuby cases somewhere here: github.com/openjdk/leyd.... I suppose it would be a good target workload for "let's go full tilt with invokedynamic everywhere" case ;)

1 1

Aleksey Shipilëv @shipilev.bsky.social · 13d

Yeah, there are more performance improvements in the pipeline :) I would advise testing latest EA and latest nightly to properly gauge where a fairly complicated project like Leyden is going.