Cineversity

Am still confused by this.  I understand the multipliers near the bottom and i think i understand the shading subdivision min, max and error threshold (they work like the antialiasing quality settings in the standard renderer).  But what does the sampling subdivision actually do?

And whats the difference between a shading subdivision and a sampling subdivsion?  And from help file, why does upping the shading subdivision min and max have no effect if the sampling subdivision is zero. 

In short it would be great to know what the sampling subdivision and shading subdivision is doing inside the renderer.

Thanks

Part-2 is really where I cover what is going on under the hood of the renderer. http://www.cineversity.com/vidplaytut/the_physical_renderer_-_part_2

The Sampling subdivision is more or less your base subdivision through out the scene, this value affects anti-aliasing along the edges of objects and also DoF and Motion Blur.
The Shading subdivisions mostly affect surface effects such as textures, blurry reflection/refraction and AO.

A pixel is always going to be an averaged result of all of its subdivisions.
so in stage a we have an image that is of a blue sphere and a smaller yellow sphere.
The resulting image will be 1px by 1px.
in both the top and bottom the pixel has subdivisions, the top fewer than the bottom.
in column b) the objects have filled the subdivided areas edge to edge based on how much of the object resides in that subdivision.
in column c) the result from b) is averaged out to display a pixel to us.
Although we started with the same initial scene the divisions change what values get averaged to provide the final result.

And from help file, why does upping the shading subdivision min and max have no effect if the sampling subdivision is zero.

I’m not to sure where this is from, raising the values in the shading/min/max will cause changes, but can only produce results that are somewhat clean, you need to still raise the sampling subdivision to get the best results.

thanks for that patrick, that helps, i was just getting confused as to what the difference was between sampling and shading subdivisions.

As for my “And from help file, why does upping the shading subdivision min and max have no effect if the sampling subdivision is zero.” I probably poorly referenced it.  I got it from right clicking on sampling subdivisions and choosing show help.  Theres an image and a explanation from tim clapham saying :

“Sampling Subdivisions values that are too low cannot be compensated for by higher-quality shading parameters (Scene by Tim Clapham).”

However with your explanation above i can see why this is so now.  The shading subdivisions cannot compensate for sampling subdivisions because its the sampling subdivisions that control edges, dof and motion blur.

Cheers for the help

The sampling sub-divisions are not just for edge detection/DoF and motion blur…they can have an effect across the surface of objects in your scene.
For example, set up a scene with a cube a floor and an area light, make sure the light casts shadows. If you render with everything set to 0 and identical grain distribution checked you get a really grainy image. Raising the min/max values in the shading subdivisions produces the exact same grain pattern, but it appears softer.
That is because the shading subdivisions are finding a better result for each pixel because it is putting more samples into each pixel, this causes a smoother transition from pixel to pixel…but the base noise pattern is the same because the sampling subdivisions are to low.
that is why you must find a good balance between the 3 settings (sampling, shading min, and shading max.)

ah i see now, thanks very much patrick!