Indigo Renderer Forum

Simple scene with shadows cast on a SSS sky.

Here is a daytime version, kind of crude mapping, etc. Just displaced a single square (only four vertices) with 15 or 16 subdivisions.

Woah, my mind is blown. Awesome.

Hey you have tamed the atmosphere medium, very nice ! Or is it a custom atmosphere ?

Nice one! The background looks like a surrealist painting, awesome!

Oh my god, the sky, the planet..... incredible

Are those planets in the distance... really planets? Giant spheres ludicrously far away?

How I did it:

The sky is SSS applied to a big cylinder centered by the camera...about 2000 Blender Units high and 80,000 across. Homogeneous medium, so it does not drop off with height, multiplied 0.008 (or 0.0008) to a scattering color of r = 0.4, g = 0.4, b = 1.0, no absorption at all. Heyney Greenstein 0.6, exponent 1.

Note that some of the big towers stick up to the top of the medium, giving dramatic shadows.

Renders fast - the pictures posted were rendered in 90 minutes on an 8 core i7. Used regular path tracing, with no bidirectional or MLT -- this seems to work the best -- MLT gives colored artifacts. Got rid of fireflies by rendering with supersampling of 4 or 5.

Terrain is a single blender "plane" with only four vertices, adaptive subdivision of 16 times. Height field is a bitmap exported from Mojoworld, textured with an mage map exported from Mojoworld. Big square maps, 6000 or 8000 pixels wide.

What's funny is that Indigo renders these scenes (with scattering media) just as fast or faster than Mojoworld's rather crude rendering!

Next step will probably be to use ISL to get the density to drop off. Have experimented with ISL clouds, white, dense scattering for the clouds, and blue, thin scattering for the sky -- this is slow, and somehow it all renders as white, not blue and white -- a bug???

Used "Extra Atmospheric" sunlight.

Actually a typo -- scattering was r = 0.2, g = 0.4, b = 1.0.

Another comment -- noticed that mesh subdivision of "smooth" Blender meshes seems to mess up and distort the UV's. A bug? If the displaced mesh is not smooth, it does not mess up the UV's, but it seems to disconnect the polygons from each other.

That's why I used a single four-vertex plane for the landscape.

The "planets" in the sky are big spheres a long way away.

Hi Sweetcement,

Very nice renders! Thanks for settings explanation!

I dont know which exporter u use but Ive got same problem with UV maps using latest blendigo. It should work fine with older 3.0.14.

Blazing setup ! I was not expecting a custom atmosphere to produce such shades, nice work really.

About the ground, try to smooth all edges before you export a divided version, that should sew the faces together. Assuming that the "smooth" currently troublesome is the shape smoothing (ie smooth the edges to sew them and leave shape smoothing alone, furthermore for a rocky landscape).

Another caveheat is that viewdependent subdivisions are disabled in most, if not in all cases due to realtime navigation. It's a must have for landscapes.

Can't wait for more !

edit: check blendigo 3.2.2.0: * fixed UV indexing !