Post a reply to this message

In article <3A6C4C92.38C491C0@gmx.de>, Christoph Hormann 
<chr### [at] gmxde> wrote:

> How about photons? does it work with them like reflection?

Photons should just behave as if it were any other pigment. Having 
photons interact with it would probably require special handling in the 
photon code, and I am not familiar enough with that code to add it. I 
think it would have to create an invisible copy of the object at the far 
end of the portal and use that to capture photons that will come out of 
the portal.
So: it should be possible, though it wouldn't be like reflection, and I 
have no idea how to add it and am focusing on the basic features of the 
patch first. In priority, photons are right behind direct illumination 
through the portal, something I will look at later but may never get 
working (or may not get working before someone else does).


> Nice idea BTW, not only for the 'picture in a picture' scene but also for
> abstract stuff just like reflection, but with probably a lot less
> geometric limitations.  

Someone mentioned it might be good for "cubist" scenes...maybe put a 
bunch of randomly placed triangle portals in front of the camera.


> When are you planning to publish it? 

Um...no idea. I'd like to get camera portals working first...and figure 
out a way to explain how it works.

-- 
Christopher James Huff
Personal: chr### [at] maccom, http://homepage.mac.com/chrishuff/
TAG: chr### [at] tagpovrayorg, http://tag.povray.org/

<><

Post a reply to this message

In article <3a6c6276@news.povray.org>, "Jan Walzer" <jan### [at] lzernet> 
wrote:

> Now ... OK, what I did was just a short trick to show my idea of 
> "colortransformation" and this picture would be done with some kind 
> of post_process ...
> 
> What I meant is, if it would be possible, to make some kind of 
> color-transformation for the rays, passing through your portal, so 
> maybe I can make the other world, I see through the portal, appear 
> inversed, or in grayscale-only, or in other modified colors, but 
> still representing the scene ...

This still sounds like something that could be done for all 
pigments...what would make it specific to the portal pigment?


> the other question, yes, was in the direction, if it is only a 
> pigment, or pattern ... but you already answered this ... I just 
> thought, if it were a pattern, than we would be able, to take it as a 
> normal-map, or as a slope-map, making our other world appear only in 
> black&white ...

That should be possible with the pigment_pattern, but I don't have any 
idea what it would look like used as a normal...some kind of embossed 
look, probably.

-- 
Christopher James Huff
Personal: chr### [at] maccom, http://homepage.mac.com/chrishuff/
TAG: chr### [at] tagpovrayorg, http://tag.povray.org/

<><

Post a reply to this message

In article <3A6C4126.E2F07808@my-dejanews.com>, 
gre### [at] my-dejanewscom wrote:

> This is a really cool feature you've developed, Chris. But perhaps it's
> a limitation of my understanding in that I cannot grasp its use beyond
> type of application proposed by Fabien.  While I'd love to do anims
> where say there are TV monitors showing 2-3 extra views of the same
> action,  I still don't comprehend the concept of "portal object."

Well, I don't know what part you have trouble understanding, so...

It is a pigment, it only affects the color of a surface, but (depending 
on the type) it can vary depending on the angle. The type 0 portal has 
an effect similar to cutting out the area of the image occupied by the 
object and replacing it with a view rendered from a different camera 
position. If you move around, you can see through the portal into the 
other part of the scene, like looking through a window. It behaves like 
rays hit the object and then appear a certain distance away in a certain 
orientation.

BTW, I got the camera portals working...perspective cameras anyway. The 
effect is identical to what you would get if you rendered the image and 
used it as an image_map, but of course isn't limited by resolution. (it 
also doesn't have or need antialiasing, and doesn't do focal blur)
This will mostly be useful for video displays...though it might also be 
of help when modelling objects, to allow multiple camera positions per 
render.

I'm going to have to go through and implement the code for the other 
camera types one by one...this part of the patch will probably be buggy 
at first, since I rarely use anything other than perspective, and don't 
have any experience with most of the camera types.

-- 
Christopher James Huff
Personal: chr### [at] maccom, http://homepage.mac.com/chrishuff/
TAG: chr### [at] tagpovrayorg, http://tag.povray.org/

<><

Post a reply to this message

Chris Huff wrote:
> 
> This is just a test scene for the portal pigment.
[Snip]

Interesting, but I'm not very sure, whether there is need for hard-coded
patch for such feature. Of course I don't want to say,  that this patch
creation is totally in vain and nobody will use it, but there is
millions of possible patches, which could be all implemented in POV-Ray
but will be used very rarely. I'm not sure, that this is desired very
much, as in the end POV-Ray will end like some programs, having tons of
features, which are not used and resulting in program gowth and
slowness. 
[Note: In addition to such strong critics toward others shameless
self-promotion follows! Sorry, but I'm such egomaniac! :-)]
BTW, my patch (POVMan) has implemented SL's trace function (trace
accepts starting point and direction vector and returns color, which is
result of this trace). With simple shader:

surface portal(point campos=(0,0,0)){
  Ci = trace (campos, N);
  Oi = Os;
}

following portal picture was created:
Note2: current implementation of SL's trace function tends to alias
strongly, escpecially in such cases, where adjoin rays will diverge (as
with this scene), for only one ray is shooted for each sample. Turning
on antialiasing will ameliorate result a little bit, but better would be
to return integrated color over some sampling area. But this is not done
yet...

Post a reply to this message

Attachments:
Download 'manportal.jpg' (15 KB)

Preview of image 'manportal.jpg'

In article <3A6DB06D.7D620ADA@aetec.ee>, Vahur Krouverk 
<vah### [at] aetecee> wrote:

> Interesting, but I'm not very sure, whether there is need for hard-coded
> patch for such feature. Of course I don't want to say,  that this patch
> creation is totally in vain and nobody will use it, but there is
> millions of possible patches, which could be all implemented in POV-Ray
> but will be used very rarely. I'm not sure, that this is desired very
> much, as in the end POV-Ray will end like some programs, having tons of
> features, which are not used and resulting in program gowth and
> slowness. 

Well, the basic portal types should be quite easy to do with a shader, 
and a shader would be much more flexible, but the syntax can be much 
simpler than an equivalent shader...the simplest version will just be 
"portal_pigment"... ;-)
This shader apparently does the "object normal" type, which uses the 
normal of the object to determine the direction of the ray...how complex 
would it be to take the direction of the original ray and apply an 
arbitrary transformation to it? That is what the default portal pigment 
does...
Also, my patch can handle cameras as well as simple portals...I don't 
think you can do that with a shader, you would have to write shaders for 
each camera type and pass them the parameters directly. Maybe have a 
macro generate the shader, if that is possible. And there is a 
possibility of doing direct lighting and photon interaction, which I 
don't think are possible with shaders unless you write your whole scene 
with some system of shaders.

-- 
Christopher James Huff
Personal: chr### [at] maccom, http://homepage.mac.com/chrishuff/
TAG: chr### [at] tagpovrayorg, http://tag.povray.org/

<><

Post a reply to this message

Can a portal object be infinite?  If so, portals can be used to vary the
effective topology of space.  Simple example: place two pairs of
parallel portal-planes to make the `torus' field of many games.

I've long wanted to try pseudo-hyperbolic space, where (for example)
five plane-quadrants meet at a point.  Such an illusion would be
improved, i think, if the portal had an optional `refraction' setting.

-- 
Anton Sherwood  --  br0### [at] p0b0xcom  --  http://ogre.nu/

Post a reply to this message

In article <3A8ED9D6.19D97662@pobox.com>, Anton Sherwood 
<bro### [at] poboxcom> wrote:

> Can a portal object be infinite? 

It's just a pigment, so it can be applied to any object.


> If so, portals can be used to vary the effective topology of space.

I'm not sure about this...it depends on what you mean. Portals just let 
you see into another area of space, if you have a cylinder poking into a 
portal object, it won't come out on the "other" side, it will just poke 
through the portal object. You could probably come up with a two-way 
portal (using two portal objects) that does something that looks like 
the object emerges on the other side.

> Simple example: place two pairs of parallel portal-planes to make the 
> `torus' field of many games.

I think the most you could do is an effect that looks like a "hall of 
mirrors", where you see infinite (ok, max_trace_level) repetitions of 
the scene when you look to the side.


> I've long wanted to try pseudo-hyperbolic space, where (for example)
> five plane-quadrants meet at a point.  Such an illusion would be
> improved, i think, if the portal had an optional `refraction' setting.

That is a good idea, I will have to put that on the list.

-- 
Christopher James Huff
Personal: chr### [at] maccom, http://homepage.mac.com/chrishuff/
TAG: chr### [at] tagpovrayorg, http://tag.povray.org/

<><

Post a reply to this message

> Anton Sherwood <bro### [at] poboxcom> wrote:
> > If so, portals can be used to vary the effective topology of space.

Chris Huff wrote:
> I'm not sure about this...it depends on what you mean. Portals
> just let you see into another area of space, if you have a cylinder
> poking into a portal object, it won't come out on the "other" side,
> it will just poke through the portal object. . . .

Er, yes, I wasn't thinking of objects passing through the plane;
I'd fake that with two copies, of course.

-- 
Anton Sherwood  --  br0### [at] p0b0xcom  --  http://ogre.nu/

Post a reply to this message

I wrote:

> I've long wanted to try pseudo-hyperbolic space, where (for example)
> five plane-quadrants meet at a point.  Such an illusion would be
> improved, i think, if the portal had an optional `refraction' setting.

hm - i think this calls for tan a = r tan b, rather than realistic
refraction (sin a = r sin b).  and note that it applies equally both
ways through the portal.

-- 
Anton Sherwood  --  br0### [at] p0b0xcom  --  http://ogre.nu/

Post a reply to this message