"Method Tracing" in async-profiler: github.com/async-profil...

Trying out off-axis guiding instead of guidescope: hitting the focus on both cameras in the same train is such an nuisance, especially with flatteners. But it tracks well once you figure it out. Here, M11 with 64x30s subs, mostly for higher SNR:

The 50x50cm canvas print is here, and it does look pretty nice!

30 minutes light from M16 through SV220 narrowband filter. AM5N is also apparently capable of tracking for 300s exposures without much problem, even with a not well matched guidescope, should be even better when OAG arrives.

Dual narrowband filters for OSC cameras FTW. On the left: ~40 minutes integration with plain UV/IR cut. On the right: ~20 minutes integration with SV220, in much poorer seeing. Fewer stars, better contrast, quite a bit more detail.

Sun Hα, 8x disk + limb mosaic, 6700x6700px (~57cm at 300dpi), remastered from 2 week old capture. This is the best Sun Hα shot I have so far. Feels like squeezing the final performance drops from Coronado PST, which is ostensibly visual only instrument, ha! I mean, Hα!

Sun Hα, full-disk, processing in negative colors. This looks rather nice when printed. (forgot to post it last month)

99% Moon shots from before the Lunar eclipse day.

The government does not want you to know that Blood Moon is actually alive and bleeding through its craters. You can only see it with specialized hardware and a bit of software post-processing. All astrophotographers are on it, and part of the global cabal to k NO WAIT I APOLLOGI

Had 20+ people ask me about stuff in 2 hours. So I decided: if I go back to Potsdam Luisenplatz on Sunday night for lunar eclipse observations, it will be for full visual with binoviewer, matching eyepieces, etc. Weather permitting, it is going to be awesome.

"Gentlemen, you can’t fight in here! This is the Department of War!"

First success at 8x Sun Hα mosaic: 5.7K disk in good detail (downsampled here). Coronado PST is a visual instrument, so brightness is not uniform across the field, which gives lots of trouble with compositing. But it almost works, if you are very patient!

More Eagle Nebula, with a little more exposure from last night.

A day worth of massaging Leyden into training with slower/optimizing compiler and delegate the untrained compilations to faster/not-that-optimizing compiler at runtime. This would be nice to replace all those -XX:TieredStopAtLevel=1 suggestions for much more efficient config.

Welp, TIL that if you crank R-B white balance settings for ZWO camera, it does not only affect the preview, but the freaking *raw capture* would not have anything interesting in green channel at all. GAAAH, WHY, ARGH.

RepairCafe: Saturday, Aug 30, 11:00..13:00 (shorter closing time) at Potsdam SLB. Come bring your curiosity and your broken devices. We can tinker together, maybe successfully, maybe not, but it should be fun either way. We speak German, English, Russian.

Let's go, first ever nebula on this imaging rig. No real set up, 30 minutes integration time through an open window from inside my apartment in Bortle 6 skies.

On this photo: amateur German astrophotographer cannot go into a German restroom without his own German equatorial mount.

First attempt at Sun mosaic in white light: 20x panel mosaic with imperfect stitching gives 8K full Sun disk image. I made a few acquisition mistakes, which means it could be even better! Ohhhh... The compressed / downsampled preview here does not do it justice:

First and not very successful Moon 8x mosaic. Mis-configured the mosaic planner, so incomplete disk was captured. But this theoretically allows 7K pixel wide disks! With more shots and 2x resampling, up to 14K pixels wide? Gotta try it next. Downsampled to 2K here for preview:

Largest capture (8Kx4K frames, 120 GB) with lots of high-quality lucky frames allows for 2x resampling during stacking, taking 100+ cpu-hours. This is the highest resolution full-disk Sun I got in white light so far, measuring 4K pixels in diameter. Down-sampled here for preview:

I hope you are not sitting inside with a computer on a warm summer Saturday? (On this photo: a computer, now outside)

Impressed how good the cable management looks when you have distribution box and curly cables. (Not shown here: another two connections to mount.) Can now power the whole rig with one power cable and connect all devices to imaging laptop with a longer USB 3.0 cable.

This morning I suddenly realized the Hubble Space Telescope has the public data archive, where one can download, for example, the 18000x18000 perfectly aligned R/G/B FITS of M42 and practice astrophoto postprocessing workflows on it. What time to be alive!

A bit of luck with two hour clearing in the clouds, stable equatorial mount, medium-sized refractor, monochrome camera, basic guiding software, 30 GB of raw data, a bit of post-processing, et voila: Sun surface granulation is clearly visible, along with some nice Sun spots.

Repair Cafe: Saturday, July 26, 11:00..15:00 at Potsdam SLB. (Surprise! No summer break for us!). Come bring your curiosity and your broken devices, we can tinker together. We speak German, English, Russian.

Me: Enjoys astrophotography, complains about the clouds. Also me: Literally works for a cloud company.

German children books are teaching the way local economies work. Easter bunnies cry about skilled worker shortage, enlist some farm animals to "help" with their work. Hired animals proceed to do it... arguably poorly. But the Easter bunnies are happy and rest instead. 10/10.

On this picture: you are an amateur astrophotographer having a convenient weekend free time window, but you live in E̸u̸r̸o̸p̸e̸ G̸e̸r̸m̸a̸n̸y̸ B̸e̸r̸l̸i̸n̸ Potsdam.

The Turing test is woefully inadequate in the times when people are able to see Jesus face in a toast. There is a considerable overlap between sufficiently smart toasters and insufficiently smart toaster users. Especially when toaster is well trained, and users are not.

If this astrophotography thing does not work out, I can always switch to bird^W insect watching. This little bug is about 1 arcminute wide, roughly the angular size of International Space Station.

Plunging deeper into astrophotography, decided to get a refurbished laptop to drive the hardware and isolate personal data. Came with Windows preinstalled. Why not, if most astrosoftware is native. German seller, so it came with German language pack. Why not, if I understand it.

Who spent 3 hours in the sun yesterday, capturing 320+ GB of Sun footage in good seeing, but somewhat strong wind, only to discover at the end of the session that mount head was not properly screwed in and wobbling? <--- This guy.

Extremely fun Sun today. First attempt at compositing disk/limb/proms shots, what a day to randomly try one!

PSA: If you are running your servers outside the plane of ecliptic (not advisable), expect some of the workloads to mysteriously fail.

Star test at random point in the sky. Enough to YOLO some things without extra guiding. Comparing with star atlas, apparent magnitude <13.5 is resolvable with 5 min total integration.

English is egregious at backward compatibility.

Repair Cafe: Saturday, June 28, 11:00..15:00 at Potsdam SLB. (The last one before summer break, I think.) Come bring your curiosity and your broken devices, we can tinker together. We speak German, English, Russian.

First light for second-hand ASI533MC camera. The sensor is huge, so it requires Barlow lenses to have any reasonable disk size, which unfortunately forces to bump the exposure, which fights the atmosphere. It does resolve things well, though!

Vacation: get up late, stretch, enjoy some coffee, feel relaxed, mark all atomic ops as std::memory_order_relaxed, have a nice breakfast, push to production, go for a walk, do not look at the phone and screaming alerts, read a book, visit a friend who isn't on call, watch sunset.

More Friday morning Sun. One of 10+ shoots in multiple modes, again feels pretty close to the limit what my current setup can do with full-disk shots. Next stop: 2x Barlow and mosaics.

Astrophoto "raw" SER files are basically uncompressed video. So for archival, compressing with just about anything helps. Actual lossless video codecs are hard to beat. ZSTD is super-fast. XZ/LZMA provides best compression, at the cost of slow speeds. GZIP is pretty meh.

Sunny day, spent 20 minutes on white light Sun captures. Some acrobatics was needed. 0.8x reducer fits disk into frame, but camera requires adapter extensions to reach focus. The image is probably a bit over-sharpened and definitely 45deg rotated. Better preparation next time.

On fun side of astrophoto, you have two major components: the actual capture (choosing and using the gear, seeing, luck) and the post-processing (choosing stacking, sharpening, compositing, coloring options, patience). And you can go and re-process stuff you captured months ago:

Unsolicited parenting advice: If your kids are playing Minecraft, deploy a local server on an isolated VM, and set up daily backups for it. Maybe also dogfood your own JDK builds and one of the concurrent GC you helped develop, if you are into that kind of deeper DIY thing...

Some more Moon shots from yesterday. Feels close to the limit for unprepared 15-minute capture session with non-cooled scope, haphazard alignment, through the apartment open window, at twilight. Can still extract a lot out of it:

JDK 25 is now in rampdown for release in Sep 2025. It is very likely to become another LTS release. I cannot stress this enough: if you are planning to use JDK 25 this year, you have to start testing it now, so that we have enough time to work out the issues, if any.

Finish your work day, open a bedroom window, make a few shots of early summer Moon.

Apologies for bad German weather on (longer) weekends for the last several weeks. I'll try to avoid planning my astrophoto sessions, hopefully that would help. Argh.

RepairCafe: Saturday, May 31, 11:00..15:00 at Potsdam SLB. Come bring your curiosity and your broken devices. We can tinker together, maybe successfully, maybe not, but it should be fun either way. We speak German, English, Russian.