3DWorld VS webgl-noise

You mean you want to draw the individual stars themselves? It's likely just a texture, probably white so that it can be colored differently for each star. This is how I've done it. Try a Google image search for "flare texture". You can find some of the ones I used here (flare*.jpg): https://github.com/fegennari/3DWorld/tree/master/textures

I remember it being called "wave surfing", where you start at the current pixel and draw a line to the light source, checking the elevation vs. the altitude value of the line (Y or Z) at every step. I have the code in my mesh_shadow_gen class here: https://github.com/fegennari/3DWorld/blob/master/src/visibility.cpp

This all sounds complex, but the system can be made to work well. I've written something similar myself, except with VBOs rather than SSBOs. My code can be found in these two files: https://github.com/fegennari/3DWorld/blob/master/src/buildings.h

C++ code for drawing spiders can be found in my GitHub project, in the spider_draw_t class: https://github.com/fegennari/3DWorld/blob/master/src/building_animal_draw.cpp

Filling the cracks properly generally gives better performance and fewer drawing artifacts compared to skirts. If you use indexed triangles and set them up in neat rows and columns, it's pretty easy to get right. No lookup tables or big switch cases, and it works with any number of LODs. See the crack_ibuf_t code starting around like 1800 here: https://github.com/fegennari/3DWorld/blob/master/src/tiled_mesh.cpp

I wrote a planet name generator that combines multi-character combinations. It may not do as good as a proper Markov Chain, but may produce names that sound better than your simple approach. If you're interested, the code can be found here: https://github.com/fegennari/3DWorld/blob/master/src/Universe_name.cpp

The code for this is in my voxel_manager::add_triangles_for_voxel() on line 497 here: https://github.com/fegennari/3DWorld/blob/master/src/voxels.cpp

Thanks. Yes, I was looking at this. The problem is that the texture atlases are in a different format than the ones I currently use. The MSDF versions don't have all of the ASCII characters and some characters have a different number of pixels. So it's not a drop in replacement for my system, which is using texture atlases like this: https://github.com/fegennari/3DWorld/blob/master/textures/atlas/text_atlas.png

I have some code for this here, but it's part of a large project and not easy to read/understand: https://github.com/fegennari/3DWorld/blob/master/src/free_obj.cpp

you may want to start with (4d cellular) ashima-simplex-noise functions of https://github.com/ashima/webgl-noise as basis function, as this is not convergent+iterative like other procgen functions (that are more like a convergent game-of-life). demos of them are in https://www.shadertoy.com/results?query=ashima&sort=newest&filter=

Oh nice. I think the GLM noise is actually based off the Ashima Arts + Stefan Gustavson GLSL noise https://github.com/ashima/webgl-noise

Oh I see what you mean now. Not saying https://github.com/KdotJPG/OpenSimplex2 isn't safe, just that https://github.com/ashima/webgl-noise/blob/master/src/noise3D.glsl isn't. Ty for clarifying, makes perfect sense now!

Several years ago when I was experimenting with all of this I found that OpenSimplex was significantly slower than regular Simplex (not sure what library I was using). Maybe it's improved since then? Anyway, I've since moved my noise computation to the GPU and speed is less of an issue for that particular project. I'm currently using the simplex noise function from GLM, which comes from https://github.com/ashima/webgl-noise. Do you have any idea how that compares with the various noise functions you've listed?