[zj]

Thanks for reply. I am testing and modifiying the template to make it production ready. The shader and document will be posted soon.

[NoRB]

Ok, i wait your new shader ! :P
So, I look again your cloud. I think the shader is good. But you must work the particles system. I think two or three expressions for u in the particles system is not a problem !

NoRB

[zj]

Since I am just testing the shader, simplest geometries is used, so I can render the image and fine-turn the shader faster. In production, geometries will be much more complex, so let me get a cool machine first.(I am using PIII 550MX now.)

[NoRB]

zj ? Are u here ?
I dont know how contact u, so if u read this post ....
I try your translucence shader but i only have a black rendering with object.
I use rat 5.0.
Do u have an idea about my problem ?
Thx

NoRB

[zj]

Contact me via

fredbach@163.com

The translucence shader depends on shadow buffers to take effect. Try older tutorials about depth map based ambient occlusion and ibi first.

The model must cover a closed volume, a flat piece of surface is a no-no, and make sure the surface noramls pointing outward.

I will update the shader and document soon.

[NoRB]

Ok. I'm going to try your older tutorials.
If i ve a problem i contact you with a mail.
Thx for your answers !

NoRB

[zj]

Hi, so long not updated. Just having no time. But I am trying shadow buffer- driven displacement now. Now I have only some test images, and tutorial and template is on their way.

[zj]

[NoRB]

Cool !
Images are great. I think this shader is going to be great !!!
I work on maya for the moment but i hope can work again on RMan quickly !

NoRB

[zj]

Hi, I have just uploaded the files about how to render such a thing. The test animation needs some time to be available still.Link: http://zj.deathfall.com/arbitrary_displacement.htm

[sabotage3d]

looking cool very cool
the technic with shadow buffers with Translucence is interesting but these many lights the dome
is very slow and it is old stuff look the multishadow technic of the russians
it is based on blur and depth map http://www.sfx.ru/stuff/subscatter/subscatter.htm
http://www.sfx.ru/stuff/multishadow/multishadow.htm
You have made many great templates but some of them are not practical for usage
and have bugs but I am your great fan keep going
I am using RAt 5.5.1 Maya 5
is there other way to use shadow buffer without dome with 280 lights

[sabotage3d]

http://www.renderman.org/RMR/Books/index.html#Sig ok this is
great technics with the siggraph 2003 Christophe Hery - Implementing a Sking BSSRDF
Erin Tomson - Human Skin for "Finding Nemo"
the technics are really awsome ....
Shadow buffers are old

[sabotage3d]

the images and code
look the displacement is very unsharped in the SSS version and smoothed

[Edited on 21/9/2003 by sabotage3d]

[sabotage3d]

the imagesa and code
color Scatter(color Cbase;
point P; /* The point being shaded */
point Porig; /* P before displacement */
normal N;
float Kd;
string mapname;
float scatterWidth;
color scatterTint;
output float __overrideDerivs;/* Message passing to the light shader
*/
output float __derivA /* Use this area for the filterwidth*/
{
color Idiffuse = Diffuse(P,N);
color Iscatter = 0;
varying point Pworld = transform("world",Porig);
varying normal Nworld = normalize(ntransform("world",N));
/*
* Read in any maps that haven't been read
* Keep pointers to them for future calls
* Free any cached information from the previous grid.
*/
ScatterDSOReadMap(mapname);
/*
* For each grid point, search the octree and create
* a cache of nearby samples. Return the grid point's
* index into this cache. At the same time, build a
* queue of nearby samples that still need to be shaded.
*/
float grid_index = ScatterDSOAddGridPoint(Pworld,
scatterWidth);
/* Irradiate all samples in the queue, filling
* the grid with more samples if necessary.
*
* prman will run the light shader on all grid points
* in this grid, even if called on only one grid point.
* If we were to serialize this portion of code (by
* running only on the first grid point, it would thus
* be extremely inefficient.) To thwart this, we'll always
* fill an entire grid. If there aren't enough samples
* in the queue, we'll use any old samples we can find.
* (cause we'll probably need to shade them eventually)
*/
varying point Psample = point(0,0,0);
varying normal Nsample = normal(0,0,0);
varying float Asample = 0;
varying color Isample = color(0,0,0);
__overrideDerivs=1;
while(ScatterDSOGetSample(Psample,Nsample,Asample) ) {
__derivA = Asample*length(transform(?world?,?current?,
vector(1,1,1));
Isample = Diffuse(Psample,Nsample);
ScatterDSOSaveIrradiance(Isample);
}
__overrideDerivs=0;
/*
* Accumulate illumination from the nearby
* samples, using the cache to avoid searching
* the octree again.
*/
Iscatter = ScatterDSOScatter(grid_index,Pworld,
Nworld,scatterWidth,scatterTint);
return Kd*Csurf*mix(Idiffuse,Iscatter,scatterAmount);
}

[zj]

Hi sabotage3d,

Thanks for your reply. You are right. Many shaders of mine are just for tutorial use and not production-proven since I have no chance to really use them. But I am working on my demo reel to put them into practice.
About the translusence shader, it is just some extension of the depth map-based occlusion shader from Andrew Whitehurst, http://www.andrew-whitehurst.net/amb_occlude.html
It is not really a surface-light interaction, and I am looking forward to a new and more efficient version of it. Now it has a fatal bug- can not handle very thin piece of geometry.
Thank you for the links. I will check the documents, but try to find some Russian-to-English translator first.
The technique presented in SigNotes by Pixar is based on in-house cache file format and DSO, I think, which are definitely beyond my still at this time. I will find some walkaround by using tools available now. But the point-cloud cache file for translusency is not so promising since the shader is usually applied to characters which have a lot of deformations. So I think the shadow map techniques will still be more practicle, as long as much less lights are used.

[sabotage3d]

thank you for your reply I will translate some of the russian characters ok this multishadow is by
JinDOs and it is for production is very fast but there is no tutorial but he achieved great results
ok I will explain how to use it in the multisahdow zip theres three files
tkTexturesMultiSh.slim - this is a shadow like the zjshdmap it can be seen when turning on the expert menus in slim and in float I think
jplastic.slim - this is the SSS shader It can simlutae GI and SSS very fast
objVarDo.mel - mel- it is working with maya 5 but you have to put it into your scripts dir
and call it into the script editor with objVarDo;
but before setup a scene with any kind of light and turn on in maya shadows depth map sahdows
then select the light make a shelf objVarDo; and in the light extra attributes will appear
many options then use the jplastic slim to any kind of object but I can't still get good results maybe
the main thing is the incad. channel the filter of the shadow is min and the resultion is 1024
the tweaking is very hard I think I can't still get absorbing coefficient like the ZJ one but
it is with only one and the shadow buffer 80 minutes waiting
I think there's new technics And ZJ is there way to make the siggraph 2003 fast BSSRDF
I think you have used some of this papers for the zjshdmap but I can't still
use it I can't see the shadow pass I make the coordsys and the spot light the simple scene
and use expression [coordsys ThenameoftheMotrcoordsys] but what else to do
nothing happend when render anyway I like SSS

[zj]

I write a more efficient translucence shader based on shadow buffer technique. It use 7 shadow maps and diffuse scatter. I think it is very promising to be a fast sss rendering.

[zj]

Each spot light is attached to domespot light shader and shadow map generator. The coordsys represents the light direction. The shader is still working in progress.

[titus]

I want to see the behaviour of your shader in motion.

[mesh]

Hi, ZJ!
Looks very promising.
Isn't it possible to keep only 1 coordinate system with 6 shadow map generators without extra lights?
This is just supposition. Couldn't it be done like in slim SoftShadow map generator?

BTW, JinnDOS in his MultiShadow method used 1 shadow map, tricky shifted and blured inside shader to achive very soft shadows. He doesn't use this shadow for SSS. JPlastic is just "backlight scattering" surface.

best regards,
mesh