Started by Dune, March 04, 2022, 11:02:29 AM

Previous topic - Next topic


I got intrigued by this question:,29893.0.html
Apparently a procedural starfield doesn't work (just blinking and randomly appearing stars), so I tried a spherical image map from the web, but the resolution was way too low. So I made a 8000x4000 spherical image of a simple procedural starfield. Took only 3mins, but this resolution is a bit on the low side too.
Rotation is simply done by transform shader by X and Y. It will probably take some experimentation to get the 'rounded' movement. I kept is very basic. The moon is a rotated simple shape, added to the starfield image.


And the second one.
It's likely possible to add different starfields and have them rotate at different speeds for some fake depth too.


Actually the first one looks a bit like snow. It's the procedural one, right?
The second one is much better. I wouldn't use a second starfield. Stars are so far away, that I don't think, you could spot some depth in there.


Are you using voronoi for your stars? PFs have varaition noises and stuffs and I don't know if you can actually fully disable such a effect on them (hence the issues with warping roads and stuff due to the 3D nature of the noise and all the variation of it).

Also remember you need to use Smooth AA, which doesn't have edge reconstruction. Otherwise, shapes will sparkles/flicker, etc. Kinda like you'd doo for compositing and stuff you wouldn't use AA with effects, and instead apply any of those effects like hard edges in post on the whole composited scene.


Yeah, first one is a direct PF (voronoi lowest octave indeed) for stars. That 'moon' could go at different speed, but perhaps not the stars.

This was more to satisfy my curiosity about moving a starfield, nothing more, so I'm not refining. But thanks for the info.


I think I may have figured that one out actually. Specifically rotating with x and y seems to be the key, together they actually behave like azimuth/elevation. I was trying to use y and z before and apparently the y transformation is applied before z, which is wrong for an off axis rotation. Here's what I'm working on. Still tweaking...