I started these Sub Surface Scattering tests to see different combinations of Half Extinction Distances and their effect on the calculations of Sub Surface Scattering. Each of the twenty-seven images are captures of a white wedge. The left side of the wedge is ten centimeters thick, while the right side is one millimeter thick.

The text at the top of each image shows the Half Extinction Distances, the image under that is the actual render, the small bar under that image is the Hue of the image above for each horizontal pixel, and the bar at the bottom is the average of the vertical pixels for ever horizontal pixel from the render. If you go through and look at all of the images, you may notice that some of them have a gradient. This only occurs when all of the Half Extinction Distances are different in Value. For renderings where two Half Extinction Distances are of equal value, there will be an almost instant change in the Hue; six of these have noisy changes in their Hue.

The Sub Surface Hue of the Main surface, or thick areas (on the left side of the wedge) is governed by the ratio of the Half Extinction Distances for their respective channels; Half Extinction RGB(60, 30, 15) corresponds to a Hue of RGB(255, 127, 0). The thinest area (the right side of the wedge) has its Hue governed by the complement to the Half Extinction RGB channels; if the main surface was Blue in Hue, the secondary surface would be Yellow in Hue.

The three renderings that are Gray have Half Extinction Distances of exactly the same value.

I am currently trying to find out what influences the time it takes to render Sub Surface Scattering. I've created the following Data Table using the Half Extinction Distances and Time to Render as variables:

Click to view attachment

It appears that the Difference Between Half Extinction Distances has affects rendering time. Click to view attachment

HED stands for Half Extinction Distances, and the Ratio is The Difference Total over the Total of the Half Extinction Distances.

I ran a trial where I stepped the Channels down by 10 starting from 100 in each channel and these are the results I got:

10:
Click to view attachment

20:
Click to view attachment

30:
Click to view attachment

40:
Click to view attachment

50:
Click to view attachment

60:
Click to view attachment

70:
Click to view attachment

80:
Click to view attachment

90:
Click to view attachment

100:
Click to view attachment

I like 50 the best even though there's a large anomaly on the head. Thinner areas do not appear to get lit when using higher half extinction distances. All of these were set at 16 Passes and rendered at 30 seconds for 100 half extinction distances to 2 minutes for 10 half extinction distances.

Here's a motion blur with 16 passes from 10 to 100:
Click to view attachment

Motion Blur 36 Passes from 5 to 36 (3:23):
Click to view attachment

Matt, this very valuable! Thank you!

I'm not quire sure why it's the complement. With a very thin piece on the right side, I would have expected very bright gradients of the primary color. Something more like this:

Click to view attachment

So for now, to get a nice effect with incredibly fast render times, use Half Extinction Distances of the same value; it will come out as the diffuse color, or decal color.

thanks Matt!

Nice tests, Matt!

H

I was surprised to see that you have used motion blur in these static images. Does motion blur act as a sort of gausian blur in static images? I thought that it would only have an effect if there was motion.

The SSS attributes are changing on a sub frame level. Creating a motion blur with 36 passes blends 36 different images with different values together. This way, every thickness of geometry can be shown in one frame.