Physically inacurrate light distribution in Octane Tweet

[abstrax]

Still not sure what you mean. There is no light leaking. This is what I get with your scene using the path tracing kernel (with up to 16 diffuse bounces):

Since the ground plane is very dark, bumping up it's brightness will make the underside of those extrusions brighter as they are mostly lit by the light reflected from the ground:

But as I said I don't know where your issue is. Can you point it out in the renderings above? Thank you.

[SSmolak]

I will do the same with default settings in other renderer to show difference. This too dark area under "balconies" is not acceptable in modern professional architectural visualizations. Sorry to say.

Increasing ground brightness is not the way because in most of real scenarios, ground is always darker.

But there are three workarounds to make it looks better in Octane - black body emission, additional fake ground invisible to camera with emission and AOV Diffuse Indirect mix with triplanar y- on surfaces. Anyway it should looks better by default.

As I know and saw, interiors done in Octane looks very professional in the terms of lighting. Probably because there is small closed area and light distribution is more equal.

[SSmolak]

This is example in Corona in different scene. Very dark lighting condition but Y- polygons have more noticeable brightness than in Octane in such situation. Even that they are not directly faced to ground but dark metal windowsill.

[abstrax]

Maybe create a simple scene and render it in Octane and then other render engines to compare. I am to 99% sure that there is no problem, only maybe some wrong expectations / setup.

[SSmolak]

Yes I will do that. The same scene in Octane and Corona. There is demo version of Corona so we will be able to check it.

[abstrax]

Yes I will do that. The same scene in Octane and Corona. There is demo version of Corona so we will be able to check it.

Thank you.

[SSmolak]

Using default lighting in Corona it looks even worse than in Octane. So this is my fault and false alarm I was wrong.

It turns out that more parameters are responsible for such lighting looks that I see in other FINAL renderers.

[SSmolak]

Corona scene attached. Octane is a winner here too.

[abstrax]

Corona scene attached. Octane is a winner here too.

In this example the "windows" are dark in Corona and white in Octane, which will reduce the amount of light reaching the extrusions a lot in the Corona render. So it's not a fair comparison.

In general think of it this way: If you look at a point on a perfectly white diffuse surface, try to "imagine" how much light can arrive at that point from all possible directions (spanning a hemisphere around the normal of that point). If the light from all possible directions is white (or 1), then the point will be shaded as white. If light from all possible directions is only 50% grey, then the point will be shaded as 50% grey. If the light from one half of the hemisphere is say 20% grey and from the other half say 60% grey, then the point will be shaded as 40% grey. Now if your surface is not perfectly white, then things get darker accordingly.

Another thing to keep in mind with exterior renderings is the contrast between the sky and the sun. Usually the sun is really bright and the sky is not. So unless your ground has a very light colour, any surface pointing down will be very dark compared to surfaces facing the sun. You can reduce the contrast by increasing the sky turbidity. But that goes only so far as the sun is really really bright.

[Dingbat]

This too dark area under "balconies" is not acceptable in modern professional architectural visualizations. Sorry to say.

Increasing ground brightness is not the way because in most of real scenarios, ground is always darker.

But there are three workarounds to make it looks better in Octane - black body emission, additional fake ground invisible to camera with emission and AOV Diffuse Indirect mix with triplanar y- on surfaces. Anyway it should looks better by default.

isn't this most likely "solved" with tone mapping?