Question about computing time.

[ROUBAL]

Hi, I have noticed something about computing time, and I'd like to get an explanation :

I'm working on a very big scene : a large landscape with some vehicles, a character and many detailed props. When the camera is over the lanscape and catches in its field everything or a large part of the scene, the rendering is fast (around 13.5 fps). On the opposite, when the camera looks in close up at a small area of the scene, involving few props (very detailed, but they are also present in the large view), the computing time is huge, reducing the speed to 0.8 fps or less.

I would have thought that more objects in the camera field would lead to a longer rendering time...

Any explanation ?

[matej]

In both of your cases Octane takes into account the whole scene, ie. the objects that are out of camera view are not ignored for calculation. Taking this into account, the render speed should be the same.

My speculation: the pixels that represent the background (HDRI, sunsky) are not sampled (at the same rate) as opposed to those that represent geometry. So a close-up which consists of all geometry, renders slower.

[kubo]

I would say that if all the materials are difusee you wouldn't see a difference (try setting the clay mode and see if speed is similar in both cases) but for what I understand is not the same calculating, let's say, 100 times 3x3 pixels worth of reflections and refractions when it's a global view than a big huge area of complex reflections and refractions in the close ups.
Also I think that lots of alpha maps on close ups make it harder to calculate than when far away.

[ROUBAL]

Thanks ! very interesting answers.

And, I will try the clay mode.