About Zenith

Github Page

Images: Pan-STARRS astronomical survey

Pan-STARRS1 home page
Making earth’s rotation visible requires a field of view on the order of arcminutes. For a browser window ~1k pixels wide, PanSTARRS meets this critical need with a resolution of 0.26 arcseconds/pixel.
1000px x 0.26”/px = 260” = 4.33’

PanSTARRS makes this observatory feasible with the necessary:

• resolution
• broad sky coverage
• of visible light

That said, it’s not the ideal image set for our purpose. While it has the resolution and coverage, PanSTARRs' primary mission was moving object detection (near-earth asteroids, etc.) and deep field sensitivity, not providing presentable images of known stars. Hence one tradeoff is frequent oversaturation. Any major star blows out the sensor and shows color distortions. See below.

Object names: SIMBAD database

https://simbad.cds.unistra.fr/simbad/

The SIMBAD astronomical database provides basic data, cross-identifications, bibliography and measurements for astronomical objects outside the solar system.

We query SIMBAD for objects based on the current field of view. Zoomed in this deep, it's all obscure. Getting the Horsehead Nebula will be like winning the lottery.

Tiling & Overlays - leaflet.js.

Leaflet is typically used for interactive maps. Though the celestial map is a sphere and motion is rotational, at this extreme zoom level, we can treat each tile as a rectangle, and the motion as simply linear. Leaflet handles layering PanSTARRS images, the coordinate grid, and SIMBAD names. Impressively SIMBAD crosshairs align with actual stars in the PanSTARRS images, down to the pixel, while coming from completely independent sources.

Language: Javascript. all client-side, no server component (zero work for daserver :)

Image processing

PanSTARRS raw images are not pretty. We have 2 steps

• white edge removal - some tiles are incomplete and have white sections along an edge. (if you see narrow white lines on screen, that's a remnant where the white edge removal was too conservative).

• noise filter - a simple threshold function, with super aggressive settings.

Before:

After:

Unsolved problem: oversaturation.

Sensors from this telescope survey oversaturated with even medium brightness stars. See the green blobs in the center of stars in the “before” image, which are then mishandled by the noise filter.

Still working to find a solution. attempts to process near monochromatic pixels (nearly pure green, pure red) and make them white, also drain the color from tiny red stars. Currently exploring topology-based solutions - “green blobs surrounded by white”, etc. such algorithms catch some but not all of the bad pixels, looking like a lazy coloring-book job.

Vibe coded in a day with claude.

Installation concept: project on a ceiling

Inspiration:

That first time you looked through a medium power telescope, with no equatorial mount, to see an object drift out of view. Someone, perhaps a teacher, told you "that object didn't move. We did. You're seeing the Earth turning."

Privacy:

This website requests your location, in order to show what's traveling over your particular location. That information never leaves your computer. You can decline, and it will default to showing the the view overhead Stonehenge. Depending on your device's settings, the location request itself may be suppressed - ie, your browser may not offer the option of sharing your location.

Next steps:

• More engaging description- offer interesting facts about what’s passing
• Server component : cache - not only reduce image retrieval, can cache the processed images
• Optimize performance : image filters are perfect application for WebGL / WebWorkers.
• Planets, solar system objects. requires tracking live objects, and supplying imagery
• Future imagery. PanSTARRS is just barely usable at this maxxed out level. New astronomical surveys coming in the next few years will have greater resolution where this application will thrive.