I was reading some post, recent and not-so-recent, on using media to
simulate SSS. I tried it out on some basic shapse, and was fairly
impressed, so I decided to try it on one of the [mediocre] head models I
was working on. Hmm... 5 hours later, I have a skin that looks great at
first glance, but looks worse and worse the more you look at it. The left
is with SSS, the right is without. Both are the same model, the same color,
etc. I have learned this:
1. To use this method with meshes, the resolution must be HIGH. No
smooth_triangles here. You can see that the top of the head of my model has
only a few quads, and with normal rendering it looks fine. but the SSS,
with media inside the mesh, is banded and rough.
2. You really need two models: one for the media, and one for the inside
opaque layer. The *plus* of this is that you can increase the media's depth
in areas of semi-transparent cartilage, like the ears, nose, and eyelids.
This will give an added sense of realism, and even less need for other
complex texturing, in image maps, etc.
3. This takes forever to render with fancy lighting, i.e. radiosity, area
lights, etc. The good news is, it shows up more when in harsh lighting (a
point light more than an area light, for example).

This would not be great for everyday renders, and probably pretty bed for
animations, but for stills where realistism is important, I was very happy
with how my crumby slapped-together test came out.

(You can see where a custom inner form is important. in this image, I just
scaled the head down a percent, and made it opaque. The double shadows on
the nose, and the wierd color of the lips is the result of this. A custom
model for the inside, and a higher resolution outer skin would make this
much better.)

Anyway, I thought it looked a little softer, a little better.

Comments?

-S
5TF!

Post a reply to this message

Attachments:
Download 'head2.jpg' (25 KB)

Preview of image 'head2.jpg'