RenderMan & RISpec >> Modeling slabs

by guanaco » Fri, 09 Jun 2006 09:19:29 GMT

Hi, I am trying to model a thin slab of material (the RIB is below). My
approach is use a cube of scaled (7.5, 10, 0.1) bilinear patchs. The
camera is looking at the center of one of the slab faces (XY plane).
Inside the slab shader I am printing the shading point P. I noticed
that I am getting repeated points differing only in z-coordinate: i.e.
P: [-2.812 4.792 17.950] and P: [-2.812 4.792 18.050], for example.

My question is if that is normal (maybe is the way Renderman works) or
I am having a problem with normals.

BTW if anyone knows where I can download models, I would appreciate. I
have looked in the Stanford and the Aim-at-Shape repositories. I am
interested something thin, like a sheet of paper.

========= RIB =========

version 3.03
Display "slab.tif" "file" "rgba"
Format 400 300 1

# Aliasing
ShadingRate 1
PixelSamples 8 8
PixelFilter "sinc" 4 4
Hider "hidden" "jitter" [0]

Projection "perspective" "fov" 45

# Camera
# Look From: 0 5 18
# Look At: 0 5 0
Rotate 180 0 1 0
Translate -0 -5 -18

WorldBegin

LightSource "ambientlight" 0 "intensity" [0.3] "lightcolor" [1 1 1]

# Front spotlight
LightSource "spotlight" 1
"lightcolor" [1 1 1]
"intensity" [60]
"from" [0 6 15] "to" [0 6 0]
"float coneangle" [0.15]
"string ShadowMapName" ["paper_shadow_front.tdl"]
"string NormalMapName" ["paper_normal_front.tdl"]

# Vertical plane
AttributeBegin
TransformBegin
Translate 0 5 0
Scale 7.5 10 0.1

Surface "mySurface"

SolidBegin "primitive"
TransformBegin
Scale 1 1 1
Patch "bilinear" "P" [0.5 0.5 0.5 0.5 0.5 -0.5 0.5 -0.5 0.5 0.5 -0.5
-0.5]
Rotate 90 0 0 1
Patch "bilinear" "P" [0.5 0.5 0.5 0.5 0.5 -0.5 0.5 -0.5 0.5 0.5 -0.5
-0.5]
Rotate 90 0 0 1
Patch "bilinear" "P" [0.5 0.5 0.5 0.5 0.5 -0.5 0.5 -0.5 0.5 0.5 -0.5
-0.5]
Rotate 90 0 0 1
Patch "bilinear" "P" [0.5 0.5 0.5 0.5 0.5 -0.5 0.5 -0.5 0.5 0.5 -0.5
-0.5]
Rotate 90 0 1 0
Patch "bilinear" "P" [0.5 0.5 0.5 0.5 0.5 -0.5 0.5 -0.5 0.5 0.5 -0.5
-0.5]
Rotate 180 0 1 0
Patch "bilinear" "P" [0.5 0.5 0.5 0.5 0.5 -0.5 0.5 -0.5 0.5 0.5 -0.5
-0.5]
TransformEnd
SolidEnd

TransformEnd
AttributeEnd

WorldEnd


RenderMan & RISpec >> Modeling slabs

by aghilesk » Fri, 09 Jun 2006 10:50:03 GMT


Hello,

A possible explanation is that you are using a renderer that shades in
SIMD (such as 3delight, prman or renderdotc). This means that it prints
"P" for an entire grid at once. So each row of a grid will have
effectively only one changing coordinate (in your case 'Z' for the
sides of the slab).

- Aghiles
www.3delight.com

RenderMan & RISpec >> Modeling slabs

by aghilesk » Fri, 09 Jun 2006 10:50:16 GMT

Hello,

A possible explanation is that you are using a renderer that shades in
SIMD (such as 3delight, prman or renderdotc). This means that it prints
"P" for an entire grid at once. So each row of a grid will have
effectively only one changing coordinate (in your case 'Z' for the
sides of the slab).

- Aghiles
www.3delight.com

RenderMan & RISpec >> Modeling slabs

by Hector » Fri, 09 Jun 2006 14:59:41 GMT

Thanks Aghiles, in fact I am using 3delight. Then, what happen with a
middle point? e.g. P: [-2.812 4.792 18.0]?

Similar Threads

1. smallest slab enclosing n points

Hi,
   In our ML research we have come across a problem where it would help
if we can find out the smallest slab enclosing n points (** IN HIGH
DIMENSIONAL SPACE, NOT 2/3 DIMENSIONAL **) . Are there any efficient
solutions to this problem? I googled on web but ended up getting many
papers for 2/3 dimensional cases.

Any pointers to relevant literature??

Regards
Pannaga

2. Please help: how to project 3D model to 2D model

3. Low res Model to Hi res Model texture Coordinate Mapping

Is there anyone that can suggest an algorithm for mapping a Low res
model texture coordinates to an Hi_res model texture coordinates??

Thanks in advance
Luca

4. OT: Hand model or scan of existing model.

5. Polygonal Modeling vs. Spline Modeling

6. Max user converting to Lightwave for modeling................flying boat model

7. DirectX .x file format or 3DS for models?

Hello,

I am rewriting a lot of my code and porting it from OpenGL to
DirectX.  I have been using Lib3DS to parse the model and load it into
vertex arrays... this has worked out well in OpenGL.

Well, now that I am using DirectX I was hoping I could use some of the
experience everyone has had with .x models?  I will have a fairly high
poly count for what I'm rendering (I've barely been able to reach
30fps in OpenGL).  What is your opinion of .x file formats... I've
heard that they can be quite bloated and you should choose a different
format such as OBJ or 3DS.  Does anyone have experience with this that
they can comment... would it be worth going to .x files since it's
natively supported by D3D or should I stick with my 3DS
implementation.

Comments are welcome... thanks for reading :)

8. DirecX8.1 beginners steps - The model gets cut off when camera goes up