There seems to be a problem with release 3.5 candidate #6 and specular
transparent image-maps. The specular highlight is visible where the
image-map is transparent.

Richard

Post a reply to this message

Richard Borsheim <rbo### [at] shawca> wrote:
> There seems to be a problem with release 3.5 candidate #6 and specular
> transparent image-maps. The specular highlight is visible where the
> image-map is transparent.

  Have you tested that:

  a) it doesn't happen with other pigments but just with image maps?
  b) it happens only in 3.5 and not previous versions?

-- 
#macro M(A,N,D,L)plane{-z,-9pigment{mandel L*9translate N color_map{[0rgb x]
[1rgb 9]}scale<D,D*3D>*1e3}rotate y*A*8}#end M(-3<1.206434.28623>70,7)M(
-1<.7438.1795>1,20)M(1<.77595.13699>30,20)M(3<.75923.07145>80,99)// - Warp -

Post a reply to this message

>   Have you tested that:
>
>   a) it doesn't happen with other pigments but just with image maps?

I'm not sure how that would be done. You would need to have a pigment with
transparent regions.

>   b) it happens only in 3.5 and not previous versions?

The model uses Mesh2/UV Mapping so it doesn't work in versions prior to 3.5,
and all the earlier betas of 3.5 have expired of course.

Richard

Post a reply to this message

"Richard Borsheim" <rbo### [at] shawca> wrote...
> There seems to be a problem with release 3.5 candidate #6 and specular
> transparent image-maps. The specular highlight is visible where the
> image-map is transparent.

This is a known feature that has existed for many versions of POV.  Think of
it like glass... you want the highlight to show up even though the glass is
clear.

You'd have to use a material_map or a texture_map with an image_pattern to
really make it work the way you want.

-Nathan

Post a reply to this message

Richard Borsheim <rbo### [at] shawca> wrote:
>>   a) it doesn't happen with other pigments but just with image maps?

> I'm not sure how that would be done. You would need to have a pigment with
> transparent regions.

  Easily. Specify a color_map with some of the values being rgbf 1

>>   b) it happens only in 3.5 and not previous versions?

> The model uses Mesh2/UV Mapping so it doesn't work in versions prior to 3.5,
> and all the earlier betas of 3.5 have expired of course.

  You should have tried the same thing with other objects. Your conclusion
that it's a problem with image_maps was too premature.
  As Nathan said, it's not a problem with image_maps, nor meshes. It's not a
problem at all. It works as it should.
  (Why? If not, how would you make clear glass with highlights?)


-- 
#macro M(A,N,D,L)plane{-z,-9pigment{mandel L*9translate N color_map{[0rgb x]
[1rgb 9]}scale<D,D*3D>*1e3}rotate y*A*8}#end M(-3<1.206434.28623>70,7)M(
-1<.7438.1795>1,20)M(1<.77595.13699>30,20)M(3<.75923.07145>80,99)// - Warp -

Post a reply to this message

> >>   a) it doesn't happen with other pigments but just with image maps?
> > I'm not sure how that would be done. You would need to have a pigment
with
> > transparent regions.
>   Easily. Specify a color_map with some of the values being rgbf 1

Fair enough. That doesn't sound too hard, but from my perspective the two
are not the same, though I can see that from POVRay's perspective they
likely are.

> >>   b) it happens only in 3.5 and not previous versions?
> > The model uses Mesh2/UV Mapping so it doesn't work in versions prior to
3.5,
> > and all the earlier betas of 3.5 have expired of course.
>   You should have tried the same thing with other objects. Your conclusion
> that it's a problem with image_maps was too premature.

I don't think I was premature at all. The renders looked wrong and I
isolated it to specularity. I think I put in a reasonable effort to
determine the cause. Since Mesh2 and UV mapped image-maps are new to POVRay
3.5 it is not unreasonable that I would come to the conclusion that it is a
new issue. I posted here in order to be of assistance.

>   As Nathan said, it's not a problem with image_maps, nor meshes. It's not
a
> problem at all. It works as it should.
>   (Why? If not, how would you make clear glass with highlights?)

I wouldn't necessarily say that it works as it should. It all depends on
your perspective. From my perspective it is more important to support
punch-through transparency with image-maps
(http://www.btinternet.com/~mike.pye/transparency4.jpg) but I recognize that
not everyone would agree.

Richard

Post a reply to this message

> > There seems to be a problem with release 3.5 candidate #6 and specular
> > transparent image-maps. The specular highlight is visible where the
> > image-map is transparent.
> This is a known feature that has existed for many versions of POV.  Think
of
> it like glass... you want the highlight to show up even though the glass
is
> clear.
> You'd have to use a material_map or a texture_map with an image_pattern to
> really make it work the way you want.
>
> -Nathan

So, I guess punch-through transparency
(http://www.btinternet.com/~mike.pye/transparency4.jpg) for image-maps is
not supported then? It's a pretty common technique in real-time rendering to
reduce polygon count, and for simplifying the modelling process.

Richard

Post a reply to this message

Richard Borsheim <rbo### [at] shawca> wrote:
> So, I guess punch-through transparency
> (http://www.btinternet.com/~mike.pye/transparency4.jpg) for image-maps is
> not supported then? It's a pretty common technique in real-time rendering to
> reduce polygon count, and for simplifying the modelling process.

  Are you sure that if you use phong highlights in a scanline-rendering
engine (eg. OpenGL) it will not do the same thing as POV-Ray?

-- 
#macro M(A,N,D,L)plane{-z,-9pigment{mandel L*9translate N color_map{[0rgb x]
[1rgb 9]}scale<D,D*3D>*1e3}rotate y*A*8}#end M(-3<1.206434.28623>70,7)M(
-1<.7438.1795>1,20)M(1<.77595.13699>30,20)M(3<.75923.07145>80,99)// - Warp -

Post a reply to this message

"Richard Borsheim" <rbo### [at] shawca> wrote in message
news:3d10062c@news.povray.org...
>
> So, I guess punch-through transparency
> (http://www.btinternet.com/~mike.pye/transparency4.jpg) for image-maps is
> not supported then? It's a pretty common technique in real-time rendering
to
> reduce polygon count, and for simplifying the modelling process.

Not as directly as you'd like. What you found should be the same as with any
pigment, when a finish which has specular highlighting added; the image_map
doesn't control finish attributes. More precisely, finish applies to pigment
as a whole, simply because transparency isn't a separate entity unto itself
I guess. When metallic is used, however, then any color does have affect on
the highlights. But again, that's actually due to finish controlling pigment
and not the other way around.

Maybe you understand all this now, but the problem you were describing
seemed to show that you did not before. For those people more acquainted
with POV-Ray this has become a common knowledge sort of thing. Or to be
fair, you might have been using a texture without any finish, in which case
it would certainly be a real problem since highlighting and diffusion
shouldn't be showing on completely transparent parts of a texture (rgbf 1 or
rgbt 1 only). You didn't say whether or not a finish were being used, so
that would be good to know, however you did say "specular image_maps". That
must mean you do have a finish set up.

Anyway, back to the item in question. Fact is, I think all you can do is
like Nathan suggested already. To map out the image via a texture not
pigment. For POV-Ray that would be the way to go about that punch-through as
you called it. The technique you told of isn't something I know about but I
get the idea.

Hope you don't take all the 'not a bug' stuff the wrong way. For instance,
Warp was pointing out the obvious. Some people expect highlights on clear
objects if specular or phong is specified in a finish, others don't.
Likewise, to see reflections if reflection is set, or not. Reflection
behaves this way too.

There isn't a special discerning of a finish within a singular (and
unmapped) texture with regards to pigment, unless you count varying
diffusion and other color changes due to incident angle of lights etcetera.

Well, have I beat this into the ground enough?  :-)

bob h

Post a reply to this message

> > So, I guess punch-through transparency
> > (http://www.btinternet.com/~mike.pye/transparency4.jpg) for image-maps
is
> > not supported then? It's a pretty common technique in real-time
rendering to
> > reduce polygon count, and for simplifying the modelling process.
>
>   Are you sure that if you use phong highlights in a scanline-rendering
> engine (eg. OpenGL) it will not do the same thing as POV-Ray?

This is standard practice when modelling for games. Certainly Direct3D
supports it and correctly renders it (in my experience). It is an incredibly
effective technique (as long as you don't look too closely :-) ) For example
the holes punched through the side of the wheel supports in this image
(http://www.amabilis.com/gallery/Alex%20Grigny%20de%20Castro%20-%20Sojourner
%20-%20Large.jpg) would not be renderable with POVRay. They would look like
they have a shiny glass surface over them.

Richard

Post a reply to this message

Richard Borsheim <rbo### [at] shawca> wrote:
> This is standard practice when modelling for games. Certainly Direct3D
> supports it and correctly renders it (in my experience). It is an incredibly
> effective technique (as long as you don't look too closely :-) )

  Then how would you render clear glass (with highlights) with such engine?

> For example
> the holes punched through the side of the wheel supports in this image
> (http://www.amabilis.com/gallery/Alex%20Grigny%20de%20Castro%20-%20Sojourner
> %20-%20Large.jpg) would not be renderable with POVRay.

  Why not? There have been several mentions about image_pattern in this
thread. You can use it to get that effect.

-- 
#macro M(A,N,D,L)plane{-z,-9pigment{mandel L*9translate N color_map{[0rgb x]
[1rgb 9]}scale<D,D*3D>*1e3}rotate y*A*8}#end M(-3<1.206434.28623>70,7)M(
-1<.7438.1795>1,20)M(1<.77595.13699>30,20)M(3<.75923.07145>80,99)// - Warp -

Post a reply to this message

> Maybe you understand all this now, but the problem you were describing
> seemed to show that you did not before.

I think that I knew that (or should have), but didn't apply it to this
particular case. My thinking was "100% translucency=Non-existence".

> You didn't say whether or not a finish were being used, so
> that would be good to know, however you did say "specular image_maps".
That
> must mean you do have a finish set up.

I did have a finish set up. It works fine if there is no specular set.

> Anyway, back to the item in question. Fact is, I think all you can do is
> like Nathan suggested already. To map out the image via a texture not
> pigment. For POV-Ray that would be the way to go about that punch-through
as
> you called it. The technique you told of isn't something I know about but
I
> get the idea.

I think I understand, but I don't think this could be applied in this
particular case since the POVRay file is "generated" and that would require
some steps that would be akward.

The solution/work-around is to only use punch-through modelling on objects
that don't require specularity/reflection etc.

> Hope you don't take all the 'not a bug' stuff the wrong way. For instance,
> Warp was pointing out the obvious. Some people expect highlights on clear
> objects if specular or phong is specified in a finish, others don't.
> Likewise, to see reflections if reflection is set, or not. Reflection
> behaves this way too.

I completely understand the "not a bug" stuff. I'm a developer myself and
know the "feature vs. bug" issue well. I did try to phrase it politely (but
obviously failed). I guess I'll have to post a request for a new feature:
"100% translucency = non-existence". Though, I suppose not many "real"
POVRay users will be interested in such a feature.

> Well, have I beat this into the ground enough?  :-)

Yup. Good job! Thanks.

BTW anyone know why if you use atmosphere you need to have something behind
the atmosphere or the "visible light" is not visible? Should I post that
somewhere else?

Richard
PS Here is something rendered with POVRay and created with my program (3D
Canvas) that I thought was particularly nice:
http://www.train-sim.com/dcforum/User_files/3ccaa5291beb7bb4.jpg

Post a reply to this message

> > This is standard practice when modelling for games. Certainly Direct3D
> > supports it and correctly renders it (in my experience). It is an
incredibly
> > effective technique (as long as you don't look too closely :-) )
>
>   Then how would you render clear glass (with highlights) with such
engine?

You probably wouldn't, but it would depend on the engine. You might use a
high level of translucency (say 99%) for glass and have a special case for
100% translucent which means don't render it. I just did a little test with
the engine I use and it just makes the highlight proportional to the
translucency which of course makes it impossible to create glass (without
faking it)

> > For example
> > the holes punched through the side of the wheel supports in this image
> >
(http://www.amabilis.com/gallery/Alex%20Grigny%20de%20Castro%20-%20Sojourner
> > %20-%20Large.jpg) would not be renderable with POVRay.
>
>   Why not? There have been several mentions about image_pattern in this
> thread. You can use it to get that effect.

I read the posts in the wrong order. See my other post.

Richard

Post a reply to this message

in news:3d10c362$1@news.povray.org Richard Borsheim wrote:

> BTW anyone know why if you use atmosphere you need to have something
> behind the atmosphere or the "visible light" is not visible? Should I
> post that somewhere else?
> 

In the doc, section "6.9.1 Atmospheric Media", in the note.

Ingo

Post a reply to this message

"Richard Borsheim" <rbo### [at] shawca> wrote...
> I guess I'll have to post a request for a new feature:
> "100% translucency = non-existence". Though, I suppose not many "real"
> POVRay users will be interested in such a feature.

FYI, it probably won't ever get added.  It's generally not good practice to
have a "trigger value" like that.

The more likely feature to be added in the future is a "finish map" which
will allow the finish to be directly controlled by a pattern.  Until then, a
"texture map" (already mentioned) is the way you can achieve this result.
It would look something like this:

texture {
  image_pattern {
    <your image file>
    use_alpha // so that it uses the alpha (transparancy) channel
  }
  texture_map {
    [0.0 texture {<your image texture with highlight>} ]
    [1.0 texture {<clear, no-highlight texture>} ]
  }
}

-Nathan

Post a reply to this message

You know what I want? Aside from a finish_map, but that would solve this
problem?

I want the ability to specify a "transparency_map" for the entire texture,
separate from the pigment, normal, or finish. Something that tells how much
whatever's underneath shows through.

This would be a slightly easier way of mapping textures, only it would have
more advantages. For instance, right now, transparency can only be specified
in the pigment - but what if you have an image_map for your pigment, and
want to change the transparency? At the moment, I believe you'd have to
average the pigment with a transparent one. But with a transparency{} block
of some sort, you could do it more easily.

It would just be nice to have a way to influence the transparency of the
entire texture all at once.

 - Slime
[ http://www.slimeland.com/ ]

Post a reply to this message

Slime <slm### [at] slimelandcom> wrote:
> I want the ability to specify a "transparency_map" for the entire texture

  Does this suffice?

#macro TextureWithTransparency(OriginalTexture, TransparencyFunction)
  texture
  { function { TransparencyFunction(x,y,z) }
    texture_map
    { [0 OriginalTexture]
      [1 pigment { rgbt 1 }]
    }
  }
#end

// Example:
#declare MyTexture =
  texture
  { pigment
    { granite color_map { [0 rgb <.5,.3,.1>][1 rgb <1,.6,.3>] }
    }
  }

#declare Transparency = function { pattern { bozo scale .25 } }

camera { location <1,2,-5> look_at 0 angle 35 }
light_source { <100,200,-50>, 1 }
plane { y, -1 pigment { checker rgb 1, rgb .5 } }

sphere
{ 0,1
  TextureWithTransparency(MyTexture, Transparency)
}


-- 
#macro M(A,N,D,L)plane{-z,-9pigment{mandel L*9translate N color_map{[0rgb x]
[1rgb 9]}scale<D,D*3D>*1e3}rotate y*A*8}#end M(-3<1.206434.28623>70,7)M(
-1<.7438.1795>1,20)M(1<.77595.13699>30,20)M(3<.75923.07145>80,99)// - Warp -

Post a reply to this message

Hmm. Actually, that's not bad.

 - Slime
[ http://www.slimeland.com/ ]

Post a reply to this message