Adaptive DMC Image Sampler VS DMC Sampler

[Morne]

Under the VRat tab, is the AA Adaptive image Sampler with Clr thresh of 0.005 and "use DMC sampler thresh" UNTICKED
the same as having it ticked and then in Settings Tab having adaptive amount 1.0 and Noise threshold 0.005?

If you set something in the one, does it affect the other. I've always used the one or the other, never both.

I'm confusining myself early in the year already

[Guest]

Sorry to hijack but I am reading through this stuff too and I have a few questions about each option. Firstly, why are they called the same? Its confusing :/ I have been reading through http://interstation3d.com/tutorials/...yfing_dmc.html
and at the "DMC Sampler options" section is where I get confused.

I understand that "DMC sampler" is always on which ever image sampler is used. As I understand it the "DMC sampler" is used to calculate anything that is a secondary ray, so reflection, refraction etc. One thing I am not sure of is how it initially samples each pixel. Does it sample within a pixel like "adaptive DMC image sampler"? So in summary if "adaptive subdivision" is used, primary rays from the camera are sampled from an edge of a pixel and secondary rays are sampled within a pixel.

If "adaptive DMC image sampler" is used then both primary and secondary rays are sampled within a pixel resulting in a much better image.

So leading on to DVP3D's question... If "use DMC sampler thresh" is ticked the threshold for generating neighboring samples is the same for both primary and secondary rays. If un-ticked then primary rays have their own threshold. In a nut shell unless you are hung up on getting the finer detail out of a primary ray to reduce rendering times then you could just control the threshold for the whole scene (primary and secondary) using the "DMC sampler".

[prima]

Firstly, why are they called the same? Its confusing :/

I agree.

Why don't use something, to help users go deeper, in the understanding of the vray dmc core? I am hoping someday to see something like a video, maybe by Vlado, commenting on a typical construction (I mean,from the materials settings to the rendering) of a photorealistic scene, in 3dsmax. Both for interiors and exteriors.

[pixela]

Because they are the same agent that control for noise.

[Morne]

ok so I still dont understand. Are we using both of these, or only one or the other?

[Guest]

Both I believe

[Macker]

Both do the same thing, but within seperate parts of the render engine; DMC is purely the algorithm they both use hence the same name; Deterministic Monte Carlo sampling.

The antialiasing DMC sampler will search for pixels next to eachother with high contrast; at which point it will "subdivide" those pixels (so a single pixel would become 4 "sub-pixels" for lack of better terminology) and repeat the process until one of the criteria is met that tells it to stop subdividing - either the noise threshold or the max subdivs. Once it has done this it will average out the colour of the 4 (or however many) sub pixels to provide a more smooth colour transition between light & dark.

The DMC Sampler within the VRay settings relates to GI noise (though it can also be linked to the antialising sampler via the "use DMC sampler threshold"), though this is where my knowledge gets a little hazy and will need Vlado to clarify.

[Guest]

Sorry to hijack but I am reading through this stuff too and I have a few questions about each option. Firstly, why are they called the same? Its confusing :/ I have been reading through http://interstation3d.com/tutorials/...yfing_dmc.html
and at the "DMC Sampler options" section is where I get confused.

I understand that "DMC sampler" is always on which ever image sampler is used. As I understand it the "DMC sampler" is used to calculate anything that is a secondary ray, so reflection, refraction etc. One thing I am not sure of is how it initially samples each pixel. Does it sample within a pixel like "adaptive DMC image sampler"? So in summary if "adaptive subdivision" is used, primary rays from the camera are sampled from an edge of a pixel and secondary rays are sampled within a pixel.

If "adaptive DMC image sampler" is used then both primary and secondary rays are sampled within a pixel resulting in a much better image.

So leading on to DVP3D's question... If "use DMC sampler thresh" is ticked the threshold for generating neighboring samples is the same for both primary and secondary rays. If un-ticked then primary rays have their own threshold. In a nut shell unless you are hung up on getting the finer detail out of a primary ray to reduce rendering times then you could just control the threshold for the whole scene (primary and secondary) using the "DMC sampler".

Can some one confirm this? Am I right about primary and secondary rays? I could be way off but I wasn't sure.

[vlado]

The secondary rays always start from wherever the primary ray hit a surface; so if the primary camera rays are within a pixel, then so are the reflections, and vice versa - if the image samples are at the edges, then the reflections are computed only for the points at the pixel edges.

Best regards,
Vlado

[Guest]

Thanks Vlado, very helpful.

[joconnell]

As a further clarification to this do we pretty much consider that the main dmc sampler deals with the noise from all of the samples bouncing around the scene until it's within the dmc threshold, and then passes the results back to the anti aliasing dmc sampler which does similar with the "2d" results in terms of bringing it within it's noise threshold?

As in the path would be:

Camera fires ray > secondary rays fired > Back to dmc sampler to check if results under threshold > More rays if needed > Results under dmc thresh > Fed back to AA sampler for 2d / edge / geometry sampling > Check if under AA noise threshold > More camera rays if needed > DMC loop > Back to AA sampler and finish if below thresh