I'm Manuel Bastioni from makehuman.org project.
We have found a stable, easy and robust way to have a good sss effect in human skin.
Take a look:
I want use this method in Pixie too (in automated way), so we can include it in next makehuman release.
I'm trying to bake the model, following the instruction here:
but I've only a black texture :-(
Instead boring you with my files, I prefer ask you a zipped archive with a working example files, like
-sphere.rib (a file that save points cloud from a UVmapped polygonal sphere)
-ptc2tiff.rib (a file that load the points cloud and render it as 2D texture)
I can't wait to test this sss with Pixie, please help me! Have sss characters is fundamental.
Anyway, these are files I've done.
To try them, compile the shaders, and render step1.rib and step2.rib.
I can't understand the reason of black result :-(
There may have been a bug in pointCloud support in a previous release (which, if I'm right, is now fixed).
Can you upload the shaders in zip or tgz format (I don't have unrar or whatever). It might just be a bug with zero-normal lookups.
In terms of SSS, there is a plan to provide a hierarchical SSS tool. This would use baked out worldspace point cloud files. It would then perform SSS on the file and spit out another point cloud which you can read back in with SSS contribution.
There is now a Ptc API to allow you to read point clouds, so it's totally possible to write SSS code and run it on ptc files, if you wanted to process them in some other way rather than via a texture.
The SSS results you have look good. It is almost certainly faster than what I had planned (though that will support internal blocking geometry / highly translucent objects).
Here is the zip (sorry for rar and for my late reply! ):
To be sure it was not a makehuman geometry bug, this is another experiment, with a more simple poly (only the head, without eyes , toungue and teeth. Better use this for experiments.
It's a pixie bug unfortunately, relating to lookups with zero normals.
It'll be fixed soon. Sorry about this.
I thought at first this was a Pixie bug. But I don't think that's the case.
It seems that (in the first example at least) the 'st' range is outside 0-1. That's no big deal, but you'll have to replace the flat polygon with
Polygon "P" [ -1 -1 0 1 -1 0 1 1 0 -1 1 0 ] "st" [ -1 -1 1 -1 1 1 -1 1 ]
To create a texture which covers the increased -1 1 range.
Hope that helps
It work perfectly now! Pixie is impressive and very quick.
Thanks for this great sw!
If you want, you can use this as baking example file in your wiki:
In this weekend I hope to work on a good skin shader using this method, so we can use Pixie in next reelase of MH.
PS: I've copied the wrong polygon from this tutorial:
PS2: I'll inform you about skin shader progress. Finally with sss we are competitive...
PS3: can you fix the zero normal bug in next 2 months?
Thanks for your feedback, and thanks for the tutorial material.
The polygon was OK in the tutorial, because the st coordinates on the mesh were in 0-1 range. It's OK if they're not, but the range of the polygon would need to be different.
Over time I'd like to have a tool to automate this render-back process.
Please do keep us updated on the shader. Looks good.
There was no zero normal bug in the end (well, actually there was, but we fixed it a while ago)