While there’s no denial to the quality of your work, I have to disagree on this statement of yours:
Originally posted by The Ripper
Human skin is very difficult to achieve so if you don’t have the right tools such as a very good SSS shader plugin (like the one they used on the gollum in LOTR) and if you don’t use radiosity to light it… I’m sure your character/skin won’t look very realistic.
Using full-blown radiosity or any other Global Illumination solution in a production enviroment is close to impossible IMHO, or at least it’s a very stupid thing to do. Even for video resolution (600-720 pixel wide), render times are still far too long in every renderer I know of, especially if you want reasonable antialiasing and motion blur as well; not to mention displacement mapping (which is pretty much a must above 1K resolutions nowadays).
Now we’re pretty far away from Weta, Pixar, ILM or even smaller teams like the Blizzard guys, but let me use our last project for a small example:
We’ve never had any layers with render times longer than 2-3 minutes per frame, and whenever it was possible, we’ve tried to go under 1 minute (and luckily we had no crowd shots or other extremely demanding stuff).
It is quite possible to do, even when using Maya, and I don’t think we should be ashamed of the results. So this means absolutely no ray tracing, and also as few lights as possible; what’s more, we’ve rendered camera projection maps to replace the shaders + lighting whenever it was possible. And with all this, we were still barely able to stay on schedule, combined with the idiocies of the client. Also, this has allowed us to render each sequence at least 2 times, so that we had enough headroom to correct any problems and glitches (of which there are many no matter what you do).
Of course larger houses with bigger render farms and longer schedules have a different threshold, but they have similar workflow and similar goals.
Production means that you MUST make render times as fast as possible. A few - or even a few dozen - spotlights will ALWAYS be faster than the current GI solutions. So whatever you want to render, it is reasonable to state that it’ll render faster without GI. And I’ve never had a project where we were able to say, ‘so, it renders very fast, let’s add something more demanding to the scene’… 
There are of course clever ways to integrate some GI into the production pipeline, with the more common being ambient occlusion and reflection occlusion. New technologies should make other things possible in the future, but I don’t expect to see faster GI - there’s only faster hardware coming, but so far, scene complexity has increased at a similar rate.
Finally, let me also add that Gollum was not rendered with radiosity I don’t know about enviroment based lighting. But I guess we can both admit that the results were fantastic and convincing.
Of course still images are a different matter: you can wait more for a single frame, and you won’t get many antialiasing artifacts from high frequency details (like skin bump maps). I expect that you’d have to bump up your antialiasing for animations, which would increase render times considerably…
[B]
I’m also pretty sure that those gradients on luminosity, color, translucency,… to fake the sss effect don’t work at all.
[/B]
It is quite possible that SSS works better, but I wouldn’t say that the gradient/falloff/facing ratio stuff couldn’t be ‘good enough’, especially if one takes the time to render separate passes. Don’t underestimate the possibilities of a biiiiig shader network
:applause:
