Hi,

(This post in not off topic, yet *might* contain spoilers, very small ones,
so I'm going to put in some blank space)

...






...






...





...





...

I went to see "The Mummy" last night (good film btw, lots of fun, very like
an Indy Jones film), and I have to say I thought that the graphics were very
very good. Throughout the film, we see lots of computer generated dust
storms, sand formations, rocket trails, and other various vapour/cloud
effects. After the film was over and I had turned my brain back on, I tried
to work out how they did these specific effects.

There were times I thought when the sandstorms looked very like Povray's
media feature ( like the part where the biplane is being chased by a wall of
dust that has the priest's face in it. One of the characters starts shooting
at the dust, and I thought the little explosions in the dust looked very
media-esque). However, most of the other dust/cloud effects looked of much
higher quality than I personally have ever seen Povray accomplish.

Then I remembered seeing a webpage a year or two ago that gave examples of a
method of rendering clouds using things called voxels. These voxels not only
let you have realistic cloud shapes, but let them be shaded by light
sources, and be misty and insubstantial, so you could fly through them. The
cloud examples on the page were very impressive. Unfortunately, I have no
idea of the address of that page.

I did a hunt on the web last night for pages on voxels, and I cam across an
interesting one at
http://www.uni-koblenz.de/fb4/publikationen/gelbereihe/GelbeReihe.bib . On
this page, there is an abstract on a paper that reads :

   "In this report we present a volume rendering technique
    embedded into a customary ray tracing scheme which is able
    to visualize arbitrary particle densities perceived as
    realistic clouds of different types found in nature.
    Moreover, this technique can be abused to visualize
    phenomena like fire, steam, haze, dust or other gaseous
    effects in the atmosphere. Our volume rendering strategy
    utilizes the volume sampling method, i.e. it computes an
    image by successively sampling voxels along the current ray
    and it progressively adds the voxel's optical contribution
    to the final picture in accordance to the physical laws of
    scattering and absorption."

This got me wondering exactly how close in Povray's media engine to a voxel
engine? Those last two terms, "scattering" and "absorption" are already part
of Povray's media implementation. Is Povray's media a voxel engine? If it
isn't, would it be more useful and powerful if it was a voxel engine? As far
as I can make out, voxels give you much greater control over the shape of
your voxel container, and seem to give very smooth and accurate results.

If you want to see some examples of what voxels can do, take a look at
http://www.newtek.com/products/lightwave/hv2/hv2.html . This is a Voxel
plugin for Lightwave 3D (please don't kill me! I know it's a scanline
renderer, but it was the only visual example of voxels I could find), and I
think you will see how powerful voxels can be.

So in summary, my questions are:

1) Is Povray's media engine a voxel based engine?
2) If it is, then how come the images on the Lightwave 3D page look much
cleaner and more realistic than I have ever seen Povray do?
3) If Povray's media engine isn't a voxel based engine, what would be the
benefits of converting it to one (ie, do voxels have any benefits over the
current media implementation)?
4) Would it be difficult to implement this in a raytracer, or specifically
in Povray?

Thanks,

Equiprawn

------------------------------------------------
Wherever you go, there you are - Buckaroo Banzai
       http://m3.easyspace.com/equiprawn/
------------------------------------------------

Post a reply to this message

Saw the movie, it was fun to see.
Think I saw the same "voxel" wepage you did, had a few cloud examples.
Seems to me it was by a person making their own program to do that too.
POV-Ray could probably come close to it, in fact the few great examples
I've seen done were decent enough.
The biggest difference may be in render time most of all. And admittedly
maybe the present 'media' can not achieve the same results anyhow.
Simply unknown to me so far.
Peter Popov's adventuring into particle media is a good example too. At
least I think it was Peter.


Equiprawn wrote:
> 
> Then I remembered seeing a webpage a year or two ago that gave examples of a
> method of rendering clouds using things called voxels. These voxels not only
> let you have realistic cloud shapes, but let them be shaded by light
> sources, and be misty and insubstantial, so you could fly through them. The
> cloud examples on the page were very impressive. Unfortunately, I have no
> idea of the address of that page.
> 
> I did a hunt on the web last night for pages on voxels, and I cam across an
> interesting one at
> http://www.uni-koblenz.de/fb4/publikationen/gelbereihe/GelbeReihe.bib . On
> this page, there is an abstract on a paper that reads :
> 
>    "In this report we present a volume rendering technique
>     embedded into a customary ray tracing scheme which is able
>     to visualize arbitrary particle densities perceived as
>     realistic clouds of different types found in nature.
>     Moreover, this technique can be abused to visualize
>     phenomena like fire, steam, haze, dust or other gaseous
>     effects in the atmosphere. Our volume rendering strategy
>     utilizes the volume sampling method, i.e. it computes an
>     image by successively sampling voxels along the current ray
>     and it progressively adds the voxel's optical contribution
>     to the final picture in accordance to the physical laws of
>     scattering and absorption."
> 
> This got me wondering exactly how close in Povray's media engine to a voxel
> engine? Those last two terms, "scattering" and "absorption" are already part
> of Povray's media implementation. Is Povray's media a voxel engine? If it
> isn't, would it be more useful and powerful if it was a voxel engine? As far
> as I can make out, voxels give you much greater control over the shape of
> your voxel container, and seem to give very smooth and accurate results.
> 
> If you want to see some examples of what voxels can do, take a look at
> http://www.newtek.com/products/lightwave/hv2/hv2.html . This is a Voxel
> plugin for Lightwave 3D (please don't kill me! I know it's a scanline
> renderer, but it was the only visual example of voxels I could find), and I
> think you will see how powerful voxels can be.
> 
> So in summary, my questions are:
> 
> 1) Is Povray's media engine a voxel based engine?
> 2) If it is, then how come the images on the Lightwave 3D page look much
> cleaner and more realistic than I have ever seen Povray do?
> 3) If Povray's media engine isn't a voxel based engine, what would be the
> benefits of converting it to one (ie, do voxels have any benefits over the
> current media implementation)?
> 4) Would it be difficult to implement this in a raytracer, or specifically
> in Povray?
> 

-- 
 omniVERSE: beyond the universe
  http://members.aol.com/inversez/homepage.htm
 mailto://inversez@aol.com?Subject=PoV-News

Post a reply to this message

Equiprawn wrote:
> 
> Hi,
Hi. 

For a little info on rendering particle systems see:
http://freespace.virgin.net/hugo.elias/

For many discussions on related topic see the RTNews by Eric Haines. There
is a good search engine on his site that can help you pin point the relevant
articles on this topic.

http://www.acm.org/tog/resources/RTNews/html/

-- 
Ken Tyler
  
mailto://tylereng@pacbell.net
http://home.pacbell.net/tylereng/links.htm

Post a reply to this message

Equiprawn wrote:

> So in summary, my questions are:
> 
> 1) Is Povray's media engine a voxel based engine?
 Well, very close.  The "intervals" keyword is the number of "voxels"
 that a ray passes trough.
 (but it is rather a series of slices of a given distance, instead
  of fixed size voxels (voxel means "volume element", as Pix-el)

> 2) If it is, then how come the images on the Lightwave 3D page look much
> cleaner and more realistic than I have ever seen Povray do?
 Because they have computers 100 times more powerful than what we
 can afford, and they are able to crank up the quality settings...

> 3) If Povray's media engine isn't a voxel based engine, what would be the
> benefits of converting it to one (ie, do voxels have any benefits over the
> current media implementation)?
  The most needed improvement for media is changing the sampling method
  (or giving the choice) from monte-carlo to another, quicker (as halos)
  samplig sheme.

> 4) Would it be difficult to implement this in a raytracer, or specifically
> in Povray?
  See above...

Fabien.

Post a reply to this message

> > 2) If it is, then how come the images on the Lightwave 3D page look much
> > cleaner and more realistic than I have ever seen Povray do?
>  Because they have computers 100 times more powerful than what we
>  can afford, and they are able to crank up the quality settings...

I'm sorry Fabien, but I'm going to have to dissagree with you. I have tried
Lightwave on a regular PC, and more than quality settings it is the superior
algorithms and renderers they use. They have people being payed to do these
things (R&D), while freelance programmers find papers by scientists
(mathematicians, physicists, etc.) and then implement them, modify them, and
don't always end up with the best result. The ones being payed normally do a
better job (with the obvious exception of Microsoft's programmers), specially
since they have better tools and resources to work with.

PS: Please don't flame. I am in no way implying that competent, highly-capable
programmers like the ones that do POV are doing a bad job, simply that they
don't have the time and resources that they big boys do. Thank you for
understanding. Go POV Team! Go Unofficial-Patch Team!

--
Anthony L. Bennett
http://welcome.to/TonyB

Graphics rendered
by the Dreamachine.

Post a reply to this message

> 2) If it is, then how come the images on the Lightwave 3D page look much
> cleaner and more realistic than I have ever seen Povray do?

Can you provide a URL to what you think are the best images on that site?
Perhaps we could point you to some similiar POV-Ray images to look at.

As far as your other questions are concerned, I don't have a clue.  I always
thought that voxel was just a fancy term for particle - like a 3D version of a
pixel.  It's difficult to compare a raytracer like POV-Ray to what they use for
movies.  I would guess that the particle effects in the movies aren't even
rendered and that they just store the positions of particles and then apply the
effect in post.

I've seen tutorials on using voxels to create blobby effects in LW, something I
think POV blobs can do much better.

-Mike

Post a reply to this message

Was recently doing some render comparisons using POV-Ray 3.02
'atmosphere' and 3.1 'media' as compared to another persons tests on
Animation Master, Bryce and Imagine of their respective volumetrics. I'd
have to say each has their ups and downs, while also having a better and
worse overall status (this probably wasn't a real good test btw). Most
notable is that POV-Ray would appear to be the slowest amongst these and
closest to Imagine which was superior it seems.
Such comparisons are not empirical to be sure (like what are the right
settings for each case?), however I'd have to guess the resulting
"ratings" are good evidence of real status though.
The Volumetrics Study which was used to base things upon is at: 
http://ironsaga.freeservers.com/vol-revu.html
Jim P. at AOL initiated the idea to do POV-Ray also,
and I'll post the 3.02 & 3.1 comparitive images at p.b.i.


TonyB wrote:
> 
> > > 2) If it is, then how come the images on the Lightwave 3D page look much
> > > cleaner and more realistic than I have ever seen Povray do?
> >  Because they have computers 100 times more powerful than what we
> >  can afford, and they are able to crank up the quality settings...
> 
> I'm sorry Fabien, but I'm going to have to dissagree with you. I have tried
> Lightwave on a regular PC, and more than quality settings it is the superior
> algorithms and renderers they use. They have people being payed to do these
> things (R&D), while freelance programmers find papers by scientists
> (mathematicians, physicists, etc.) and then implement them, modify them, and
> don't always end up with the best result. The ones being payed normally do a
> better job (with the obvious exception of Microsoft's programmers), specially
> since they have better tools and resources to work with.
> 
> PS: Please don't flame. I am in no way implying that competent, highly-capable
> programmers like the ones that do POV are doing a bad job, simply that they
> don't have the time and resources that they big boys do. Thank you for
> understanding. Go POV Team! Go Unofficial-Patch Team!
> 
> --
> Anthony L. Bennett
> http://welcome.to/TonyB
> 
> Graphics rendered
> by the Dreamachine.

-- 
 omniVERSE: beyond the universe
  http://members.aol.com/inversez/homepage.htm
 mailto://inversez@aol.com?Subject=PoV-News

Post a reply to this message

Very nice stuff on that webpage. It looks to me to be a mixture of
isosurface-type objects and media. Or just isosurface objects. They mention
"particles" constantly on the webpage. I would assume they are actually clumps
which are placed (and maybe merged together) to make the effects. Very cool
stuff indeed.

Samuel Benge


Equiprawn wrote:

>

<snip!>

> If you want to see some examples of what voxels can do, take a look at
> http://www.newtek.com/products/lightwave/hv2/hv2.html . This is a Voxel
> plugin for Lightwave 3D (please don't kill me! I know it's a scanline
> renderer, but it was the only visual example of voxels I could find), and I
> think you will see how powerful voxels can be.
>
> So in summary, my questions are:
>
> 1) Is Povray's media engine a voxel based engine?
> 2) If it is, then how come the images on the Lightwave 3D page look much
> cleaner and more realistic than I have ever seen Povray do?
> 3) If Povray's media engine isn't a voxel based engine, what would be the
> benefits of converting it to one (ie, do voxels have any benefits over the
> current media implementation)?
> 4) Would it be difficult to implement this in a raytracer, or specifically
> in Povray?
>
> Thanks,
>
> Equiprawn
>
> ------------------------------------------------
> Wherever you go, there you are - Buckaroo Banzai
>        http://m3.easyspace.com/equiprawn/
> ------------------------------------------------

Post a reply to this message

Attachments:
Download 'us-ascii' (1 KB)

On Thu, 22 Jul 1999 13:24:54 +0100, "Equiprawn" <equ### [at] tinetie> wrote:

>I went to see "The Mummy" last night (good film btw, lots of fun, very like
>an Indy Jones film), and I have to say I thought that the graphics were very
>very good. Throughout the film, we see lots of computer generated dust
>storms, sand formations, rocket trails, and other various vapour/cloud
>effects. After the film was over and I had turned my brain back on, I tried
>to work out how they did these specific effects.

  Be on the lookout for computer graphics magazines such as CGW (Computer
Graphics World) and industry trade journals that you might be able to find
at larger booksellers. Computer-generated effects in film get a fair share
of print space these days.

  ILM (Industrial Light and Magic) did the effects for "The Mummy". Some of
the software was proprietary such as I Sculpt, written by ILM employees for
use in "SW:The Phantom Menace", while other software is pretty much
straight-off-the-shelf like Softimage 3D.

  To give an idea of the processing power involved for "TPM", according to
an article in the June 1999 issue of CGW, ILM used eight to ten 32-processor
Origin 2000 workstations, each equipped with between 8 and 12GB of RAM. At
night they converted another 200 SGI O2 machines on the premises to the
render farm to aid in image generation.

  ILM uses a variety of software depending on the job at hand. George Lucas'
philosophy is to use the minimum effort at the cheapest cost to get the shot
completed, although this may be misleading since "TPM" was so ambitious that
the 'minimum' is better than most films' max effort :)

  George Lucas and LucasFilm, Ltd. were successful in their bid to acquire
the use of a parcel of property in the Presidio, the former U.S. Army base
in San Francisco, California. ILM is tentatively slated to be moved to the
proposed industrial park that will be constructed at this site. If LucasFilm
conducts tours of this facility after the move (may be several years away) I
will check it out and give a report, if anyone's interested. By that time,
they should be hard at work on Star Wars Episodes II and III.

-- 
Alan
--------------------------------------------------------------------
http://www.povray.org - Home of the Persistence of Vision Ray Tracer
news.povray.org - where POV-Ray enthusiasts around the world can get
together to exchange ideas, information, and experiences with others
--------------------------------------------------------------------

Post a reply to this message

On Thu, 22 Jul 1999 12:42:53 -0500, Bob Hughes <inv### [at] aolcom>
wrote:

>Saw the movie, it was fun to see.
>Think I saw the same "voxel" wepage you did, had a few cloud examples.
>Seems to me it was by a person making their own program to do that too.
>POV-Ray could probably come close to it, in fact the few great examples
>I've seen done were decent enough.
>The biggest difference may be in render time most of all. And admittedly
>maybe the present 'media' can not achieve the same results anyhow.
>Simply unknown to me so far.
>Peter Popov's adventuring into particle media is a good example too. At
>least I think it was Peter.

I wasn't home these days so... and I spent the last five or so days
trying to cope with the field around a mesh, which will (eventually)
be used to model arbitrary repulsor and attractor objects. I think
I'll start with mesh emitters now. Just the right job for the
barycentric coordinates stuff, thanks a lot, Jerry!

But how on Earth am I going to get all these into one? And the syntax
of the particle system definition file, easy yet readle by POV, is
really killing me. I'll leave the parser for the end.

Gee, I just hope I don't give up when I'm half through it [done it a
zillion times :( ]...


Peter Popov
ICQ: 15002700

Post a reply to this message