Definition of ray-tracing?
WILLIAMSON, JOHN EDWARD
I am wrting for _Stereoworld_. In particular, how does ray tracing
differ from the shading done by say, Autocad's Autoshade?
Thanks
John
Keith Mannist
the subject desired. The subject is generaly described in an x,y,z
coordinate system, with descriptions of the object surfaces in this space.
(Optical properties are also included in full ray tracing descriptions)
Robin Hayes
Sorry, I just couldn't let this whopper slide by.
Raytracing has nothing to do with the realism of the resultant image, nor
does it have anything to do with the description system used to store the
scene. (Although optical properties are included in most scene
descriptions used for raytracing.:)
Raytracing also called raycasting is a the method of producing images by
simulating the path of light backwards from the calculated position of the
viewer's eye, through each pixel on the screen to the objects in the scene,
and then to the lightsources in the scene. The color of each pixel is
calculated from the color of the lights and the color of the objects on
which are encountered on the path of the light ray.
A similar technique is called radiosity, but in radiosity rendering, the
light path is calculated forward from the light sources out, off the
objects to the eye. This technique can simulate more complex light
effects, but is MUCH more CPU intensive, since there is no way to tell if a
particular ray will be useful until you calculate it.
Amazing the things you learn in a computer graphics rendering class.
:)
-rdh
--
==================================================
Robin Hayes
I/S Assistance Center / Domain Adminstration
Information Services / Silicon Graphics Inc.
rob...@sgi.com / (415) 390-2456
==================================================
Samuel S. Paik
In article <C6znx...@odin.corp.sgi.com> rob...@escher.corp.sgi.com (Robin Hayes) writes:
>Sorry, I just couldn't let this whopper slide by.
>A similar technique is called radiosity, but in radiosity rendering, the
>light path is calculated forward from the light sources out, off the
>objects to the eye. This technique can simulate more complex light
>effects, but is MUCH more CPU intensive, since there is no way to tell if a
>particular ray will be useful until you calculate it.
This isn't radiosity, this is forward ray tracing, or ray tracing from
lights (as opposed to backward ray tracing, where you follow paths in
the reverse direction of light propagation).
Radiosity is a means of solving the "global illumination problem" by
propaging energy through an environment.
Sam Paik
Steven Collins
Just to add to the confusion, remember that ray-tracing may be used to
solve the radiosity formulation (or rather to estimate the form-factor
between 2 surfaces). On the other hand, diffuse reflection (which radiosity
attempts to model) may be captured using a traditional ray-tracer. Look at
"Radiance" by Greg Ward, for a great implementation of this.
Much of the confusion about ray-tracing arrises from the lack of distinction
between the use of ray-tracing (or more formally "ray-casting") as a method
for establishing occlusion (or penetration for volume renders) and the
Whitted Style ray-tracing algorithm used to model certain aspects of light
propogation within a scene. As an occlusion algorithm, ray-tracing (or
"casting") is very useful, but it is the lighting model used that is, in
general, insufficient. To capture spectral FX, for example, the spectrum
of light must be sampled, which requires tracing more rays. However, the
mechanism of ray-tracing doesn't change. All that changes is what information
the ray carries with it, and how to generate the rays in the first place.
steve
--
+-------------------------------------+----------------------------------------+
| Steven Collins, Visiting Researcher | email: st...@crl.hitachi.co.jp |
| Hitachi Central Research Lab. Tokyo | ph: (0423)-23-1111 fax: (0423) 27-7742 |
+-------------------------------------+----------------------------------------+
i_ca...@test.sub.su.se
>From: kman...@slee01.srl.ford.com (Keith Mannist)
>Subject: Re: Definition of ray-tracing?
>Date: 13 May 1993 19:42:08 GMT