Help with working at astronomical scales [SOLVED]

[mrwig]

I'm currently attempting to render some photorealistic images of astronomical bodies. I assumed since Octane is an unbiased renderer that I should try to operate at actual scale whenever possible. However, I'm having issues with my lighting at astronomical scales. Using smaller scales has not helped, as it appears the ratio of distances may be too great to get a good render.

I've attached a sample file with a sphere standing in for Mars and a targeted octane light standing in for the sun. I take care to adjust my ray epsilon whenever I'm switching scales, but I've been unable to find a way to prevent distant light sources from creating a strange banding effect on geometry.

I've attached a screenshot to illustrate the issue I'm having. There are two lights in the scene, the active light represents the correct distance (approximately) from the Sun to Mars. This light renders with the banding. I moved the light 10x closer to the sphere and the banding went away, however, this is not at a real-life scale, which is what I'm trying to accomplish.

Does anyone have any useful tips or tricks for working at these large scales? Is it even possible? I'm sure there's some render theory I'm ignorant of causing this issue, but I'd like to learn how to get past these limitations.

[pxlntwrk]

Hi,
did you try to tweak "ray epsilon" feature in kernel settings?

[mrwig]

Hi,
did you try to tweak "ray epsilon" feature in kernel settings?

Yes, I adjusted it according to scale, then slid it from max to min a few times. There is more artefacting when it is not adjusted, but the screenshot I posted is as good as I can get it to look. Changing the light type to disc in the C4d light setttings helped fade the lines more, but again, the screenshot I posted is my best effort so far.

[aoktar]

Hi,
did you try to tweak "ray epsilon" feature in kernel settings?

Yes, I adjusted it according to scale, then slid it from max to min a few times. There is more artefacting when it is not adjusted, but the screenshot I posted is as good as I can get it to look. Changing the light type to disc in the C4d light setttings helped fade the lines more, but again, the screenshot I posted is my best effort so far.

If range of rayepsilon is not enough then switch to centimeter from meter and restart LV render. Propably float-point values are not enough for this scale.

[mrwig]

Thanks Aoktar, that makes sense.

I will try working with smaller units and see if I can get the correct ratios.

I'll post a follow-up after I experiment.

[pxlntwrk]

hi,
here a test with all divided by 1000

[pxlntwrk]

and here,

a test with Km replaced by Cm. and sun is as spherical Octane light, all accurately scaled.



But I do not know how to precisely adjust the sun light intensity?

[mrwig]

I've just been winging it with the sun intensity. I'd like to learn more about how to create as accurate a simulation as possible, but for me this is a good enough starting point.

Thanks to everyone who helped look into this for me. Working from pxIntwrk's modified file, I was able to get the setup to work on a 1:1,000,000 scale.

Mars radius was set to 3.39cm, the sun radius (targeted area light) was 695.7cm and I placed the light 228000 cm away, which is approximately correct for this scale.