Great.
It seems like photoephemeris.com, moon calc etc. use the refraction measure too, maybe just an average. Because when it says Moon a 0 degree it matches Sitrec with refraction ON
What in/out frame range did you use? That video has some sudden camera jumps
I just tried it on 93-254 and it seemed to work:
https://www.metabunk.org/sitrec/?custom=11433/Bat%20full%20timelapse/20260225_093540.js
I tried this on a around 400 frames of the Park City UT video, but running it a couple of times on the same section gave different results. For example the first with Vega identified, and second with Achird visible. But, perhaps that video is too...
Are you logged in with 3D buildings and the building source set to Google Photorealistic?
https://www.metabunk.org/sitrec/?custom=1/Mallorca%20eclipse%20totality/20260804_175704.js
It is now (2.114.0) on by default.
And it's quite dramatic at these zooms:
Refraction off (geometric ideal)
Refraction on (real life)
At the horizon, it's nearly 0.5 degrees. That does not sound like much, but it's about the diameter of...
Yes it's perfect now! The height is still depending of a lot of atmospheric things, so you have to tweak refraction pressure a bit to have a perfect match, but all in all extremely close.
But with refraction off, the height is way off so maybe it...
I just fixed some accuracy issues for precisely this type of thing. @ThomasH had a nice photo of the sunrise behind a building where the sun was horizontally off by half a degree and vertically off slightly. The exact DTLDs were known.
The fixes...
Astrometry.net often works very well for finding where in the sky we are looking, but is not great with frames from videos that are noisy or contain other moving objects.
So I created a star tracker that attempts to resolve what lights in the...
Quick update: I have now reached 84% recall rate compared to VASCO. I still have about 17K tiles to download and process, so the recall rate should hit at least 90%. That's a pretty good number.
The unpublished third-party list "V" (107K rows)...
This is fixed. The problem was to do with a mismatch between Google's 3D tiles and the base elevation. The tiles are high resolution, and the position of this camera on a curving slope meant the camera though it was underground.