Excavating the Deep Cut

[WAS]

Interesting. One of the main problems with this scene is maintaining the density of the cloud cover. If the landscape is properly exposed the clouds are almost completely blown out. Are you suggesting that I should *decrease* the albedo of the cloud cover, to make the clouds darker to begin with?

The renderings are 32-bit exr, so changing soft clip shouldn't have any effect.

The opposite. As than the clouds transmit more light to the surface more like a normal overcast day. This allows populations; people, buildings, and vegetation shadows to be more visible. Like it is looking into a forest or what not. Denser, darker clouds means more overall shadow on the ground, to hide other shadows, kinda muddling the realism.

I'm out on a walk now under overcast and rain, take this shot of the forest and a clearing.

[sboerner]

Thanks, Jordan. That's given me some ideas.

Here is a new rendering with the fill light removed. I increased the density and coverage of the top cirrus layer to soften the sunlight, then increased the camera exposure value (now at +2.5 stops) to compensate. I also increased max rays to 8, which seems to improve the shadow detail at little cost in rendering time.

All of the cast shadows are much more diffuse, including the one cast by the left wall. The rendering has picked up some noise, no surprise there, so that will need to be dealt with. (AA is now at 9 and max paths at 25.)

I used a luminance mask to isolate the sky and bring it back into gamut, but otherwise there's very little post work here.

[WAS]

Fantastic! This is coming along fast. The lighting looks leagues better. The forest has it's deep shadows, the grass on the ground has it's blades more defined due to internal shadowing, and the wall shadow looks richer.

[Hannes]

I agree. Much better to my taste!

[Kadri]

Yes looks better indeed.

[luvsmuzik]

Smoke, dust, steam, clouds, reflection in puddles;... Tophats, waistcoats, overalls, caps and boots ...Wheelbarrows, hammers, chisels, sledges, carts, horses...Terragen has progressed to a new level with this project. A lifetime of struggle building this canal accomplished in a few short months...AWESOME render and project!

[sboerner]

Thanks, everyone. I have to agree that the lighting is much improved. I guess I needed a push – lesson learned. Should now just be a matter now of fine-tuning the renderer.

(And thanks for the kind words, Luvs. It's been a good project, for sure.)

[Dune]

Just awake, but astonished by the difference. Great work, Steve!

[j meyer]

Great work, indeed.
And nice improved lighting now.

[sboerner]

Thanks – and I appreciate the constructive criticism. I agree that it's a big improvement.

Running tests now to optimize the sampling. Because of the shadows and cloud layers this one is going to need higher settings than I normally use. I need the machine for other tasks, too, so it will probably will take a few days.

[Dune]

I guess your reflections in the canal will take quite some time, in relation to the clouds. That's what I noticed yesterday in a PT render. Usually for (not too) wet mud and finely displaced ground I use non-RT reflections, which are fast and certainly good enough. No difference to RT-reflections, especially under vegetation and such.
But I believe that PT really calculates all as, well, paths (comparable to RT), so that takes far more time. Certainly if clouds are involved.
So maybe it's much faster to do a GI cache of the whole, render the top half with clouds and the lower half without, or a simple v2 overcast sky, instead of several deep v3 cloud layers. You might miss some cloud reflections in open pools, but on ground it won't make much difference. Theoretically that is. You may have problems with the sunlight though, you'd have to test it.
On the other hand... you can also take a short vacation

[WAS]

I guess your reflections in the canal will take quite some time, in relation to the clouds. That's what I noticed yesterday in a PT render. Usually for (not too) wet mud and finely displaced ground I use non-RT reflections, which are fast and certainly good enough. No difference to RT-reflections, especially under vegetation and such.
But I believe that PT really calculates all as, well, paths (comparable to RT), so that takes far more time. Certainly if clouds are involved.
So maybe it's much faster to do a GI cache of the whole, render the top half with clouds and the lower half without, or a simple v2 overcast sky, instead of several deep v3 cloud layers. You might miss some cloud reflections in open pools, but on ground it won't make much difference. Theoretically that is. You may have problems with the sunlight though, you'd have to test it.
On the other hand... you can also take a short vacation

Pretty sure PT ignores whether ray trace is checked or not, and it is technically ray traced, well path traced.

[Dune]

Yes, that's what I think, and that's why it's probably slow, especially with lots of small displacements. Would be cool if some sort of non-RT/PT reflections could be made in PT mode, that give a reasonable 'reflection' and be as fast as in deferred rendering.

[Matt]

Yes, that's what I think, and that's why it's probably slow, especially with lots of small displacements. Would be cool if some sort of non-RT/PT reflections could be made in PT mode, that give a reasonable 'reflection' and be as fast as in deferred rendering.

These "reflection" paths shouldn't be much slower than the diffuse paths, and typically they will be faster if the roughness is above about 0.2. If you were to replace those with the reflection approximation from the Standard renderer it would wash out the detail in the shadows and make the image look a lot more like the Standard renderer, but you'd still be paying in render times for PT on the diffuse lighting. I doubt that would be a good quality/time ratio, and you might as well use the Standard renderer instead.

[Dune]

Ah, that's good to know, Matt. Thanks. All worries taken away