ZBrushCentral

What is sub pixel displacement?

ZBrush Usage Questions
#1

DPSubPix. Sub-pixel displacement. Pixologic, the practical guide, the internet, I get the same answer from everywhere: higher value is more accurate. But I would like to understand what exactly this value represents.

I imagine that perhaps for each vertex in the base mesh, a certain number of samples must be measured, which are distances from the base mesh to the final mesh…and perhaps DPSubPix refers to the number of additional samples for each vertex? But, I don’t know, this is just a wild guess.

#2

As far as I understand it the Tool > Displacement > DPSubPix slider subdivides the tool further when calculating the displacement map. This extra subdivision is not reflected in the tool’s subdivision levels, it is only in effect during the displacement map calculation.

You have probably read that you should subdivide as many times as you can to get a good quality displacement map, eventhough you are not using the extra subdivision levels to model detail. Sometimes you simply do not have enough ram to subdivide anymore and then the DPSubPix slider is useful.

#3

Ok. So based on what you’re saying, if the disp map image looks smooth, then higher values of DPSubPix can’t help make it better…because all it does is subdivide the model more, and if the model is already smooth enough more smoothing is just a waste of memory resources.

#4

Stuh,

the way I read Tveyes explanation is…

it doesn’t affect the subdivision but the calculation of the displacement…it doesn’t subdivide the model itself. I could be wrong but that was my take on his reply.

#5

He said that it subdivides it further, but did not mention anything about the actual calculations being different. If he is correct, then subdividing 3 extra times after you’re done would have the same effect on the final disp map as setting DPSubPix to 3.

There are two possibilities. Either DPSubPix effectively smoothes the model further, or it causes the displacements for the existing mesh to be calculated with greater precision.

Here is a simple experiment which will definitively tell you which answer is correct. Start with a very low-res mesh, and then subdivide it once, so that it is still pretty low poly, and all the poly edges are very visible. Then create a disp map between them. This map should not be smooth, it should have a lot of sharp lines in it showing the poly edges on the final mesh. From this point, crank up DPSubPix and calculate a new disp map to compare to the first.

If the new map appears to have smoothed out all these creases, then we know that DPSubPix is as TVeyes described, simply simulating increased levels of smoothing. This would make the displacements LESS accurrate in this case. If, on the otherr hand, it results in greater precision of calculations due to increased sampling, then the sharp edges in the disp map should become MORE pronounced instead of more blurred.

I would have done this experiment already myself but ZB was crashing so I’ve given it a rest for the night.

#6

My post in this thread should help answer questions as to how to determine the best DPSubPix setting.