Physical Starlight and Atmosphere Addon for 2.8 (v1.1)

here’s an update

I decided that round earth will take too long finalize so next release will be still based on flat earth. The changes so far:

  • I set up a git and project for tracking changes and issues
  • general cleanup of python files and code
  • I have now a UX expert helping me out
  • UI is moved to World tab
  • I decided that having “Azimuth” and “Elevation” variables is a bit dumb, so I removed them
  • Sun can now be rotated by SunPosition addon or any other tool
  • Starmap can now be added in UI
  • node groups are hidden
  • sky and sun radiance is now accurate (requires to lower the camera exposure)
  • fog automatic addition/removal

what is being worked on right now:

  • atmosphere presets system
  • clouds
  • moon
  • renderfarm fix
  • bake to HDRI button

Sunvariables are gone? I bought your add-on but didn’t play much with it yet but it seems like azimuth and elevation are good for positioning the sun. When you say ‘use another addon’ you mean we can’t change the position of the sun within your addon anymore?

it means that instead using value, now we can use sun rotation.

IMO personally, I like sun position using empty relative to camera, so I can position it based on x,y axis.


Nice… and can along bake to HDRI button also come bake HDRI animation?
Would love to use it in other engines :wink:

it seems to be more intuitive just rotating sun lamp in 3D viewport or use ‘sun position’ addon where you specify date and time of day. i can, of course leave the azimuth elevation where it is :slight_smile: it is not hurting anyone (except of cluttering the UI)


Please leave it, don’t want use or install another addon.

1 Like

I agree, please leave it or at least make it optional to show/hide

alrighty. it is back :slight_smile:


When can we expect the update?


Blender 2.82 will have sun position addon, which looks quite convenient.


@martinsh Hi there, do you have any news about the next update?


And something interesting to check out. Maybe even good for motivation or an inspiration…


belated time for an update

  • I have been investigating why sun position is not changing during render. it is taking surprisingly long time to figure out.

here are screenshots of the current progress


Looks great! Feel free to publish a beta to BM if you feel like it… even if its a little broken.

I would love to play with it setting, and yes the cloud!. Any time frame ?

Looks great so far. I’m assuming the clouds will be lit eventually as well?

@martinsh when i enable atmosphere in the default scene Memory increases from 23.7 MiB to 553 MiB. Is it normal?

yeah, could be because of the included starmap in node setup. I have replaced it with a blank texture to be user replaceable for the upcoming release.


Sounds great, I have been rerouting current texture as new input and plugging the stars there

Also, related to animation problems, been for longer time moving shaders with #frame because keyframes may not just do anything, and also I haven’t felt like investigating if its working now

Thinking that #frame pi math makes me feel like digital clocksmith :robot:

It’s really nice to see someone taking atmospheric modelling seriously in blender. Over the years, starting with Bruneton, I’ve read quite a bit on the subject. But I’ve never taken the leap of actually making my own implementation. So big kudos for actually doing it!

However there is a problem I see in your model that is shared by every other atmospheric scattering model I’m aware of: the horizon opposite the sun has a strong color gradient at sunrise/sunset. I’ve never seen anything like this IRL and I usually see the sun rise everyday.

Maybe you could enlighten me on this as I haven’t looked at your code or anything, but I think this issue may be related to treating the sun lights color temp as a variable that only changes with sun angle. In actuality the sun color temp is a function of how much air mass a ray has traveled through. So the color of the sunlight that is scattering opposite the sun should be quite a bit different than the light that is hitting the observer/center of the scene, which in turn will be a different color temp than the light that is scattering in the direction of the sun. Is your model doing anything like this?

Additionally what are you doing for clouds? Some raymarched solution like Guerrilla Games came up with?