Distortion on applying texture with plate carrée projection to sphere

Greetings !

While working on my new Planetarium project, I stuck on following problem: after applying texture to sphere the poles has noticeable distortion:

The NASA official site states about this projection:

The maps are presented in plate carrée projections using either celestial (ICRF/J2000 geocentric right ascension and declination) or galactic coordinates. They are designed for spherical mapping in animation software. The oval shapes near the top and bottom of the star maps are not galaxies. The distortion of the stars in those parts of the map is just an effect of the projection.

Is there a way to remove this distortion and ‘heal’ the universe ?
Maybe different sphere subdivision will help ?

1 Like

Icosphere, three.js docs with high detail usually works well.

1 Like

Thank you, I’ve tried it, but unfortunately without success - distortions still exist, but of different kinds.

It is generally not possible to map a 2d image onto a sphere and have no distortions.

There’s another approach called “Cube sphere,” vertices of a subdivided cube projected onto a sphere, which I use to render planets, using six textures (a cube map) instead of one. That’s about as good as it gets but still not completely free of distortion.

1 Like

download a higher quality exr from NASA SVS | Deep Star Maps 2020

convert it to something smaller using https://gainmap-creator.monogrid.com/en/

use anisotropic filtering

texture.anisotropy = renderer.capabilities.getMaxAnisotropy()

Here is a R3F version that worked out ok.
https://editor.sbcode.net/ccab2014a862eb361978bb6145c8d58f1131c73b

{A4347608-D9E2-4359-B5AC-0052ECCED74E}

3 Likes

Thank you so much - this method works fine !

It depends on projection. I was able to apply texture with equirectangular projection almost without distortion with:

texture.mapping = THREE.EquirectangularReflectionMapping;

Here is example: Gallery

If you have come up with a projection that maps a single 2d image onto a sphere without distortion then congratulations: file a patent, you’re gonna be very rich.

I assumed the reason for your original issue was that the geometry of a standard sphere has a singularitiy at the poles. I would love to understand how a higher quality texture plus anisotropic filtering was able to solve that fundamental issue.

1 Like