POV-Ray : Newsgroups : povray.beta-test : Subsurface Light Transport and mesh2 Server Time
23 Dec 2024 00:55:21 EST (-0500)
  Subsurface Light Transport and mesh2 (Message 1 to 10 of 13)  
Goto Latest 10 Messages Next 3 Messages >>>
From: FlyerX
Subject: Subsurface Light Transport and mesh2
Date: 25 Mar 2012 04:58:12
Message: <4f6ede24$1@news.povray.org>
Hello,

Windows 7-64bit, POV-Ray 3.7 64bit RC5

I have been having some issues with the new Subsurface Light Transport. 
If the mesh is made up of several sub-meshes the seams between them are 
darkened.

Using a test model of a head about 7x11x9 inches, made up of several 
sub-meshes (either union or merge), 25mm per unit, translucent color of 
0.8, and ior of 1 (there was not much change with other values). I get 
the following:

http://imgur.com/5mw16

The skin does not have any image maps, bump, specular, ambient or 
reflection. There is one spherical area light and no radiosity. The dark 
edge is alleviated somewhat as the translucent color is decreased but it 
is always visible.

If I merge the mesh outside of POV-Ray into a single mesh with a single 
material for the complete head skin then the dark seams disappear. But I 
cannot merge the mesh outside POV-Ray since that would destroy the 
separate UV mapping that each mesh uses. Does POV-Ray 3.7 work with a 
union or merged mesh2 group and SSLT? The test model is open at the 
bottom but I also tested this with a closed sphere with two materials 
and it gave me the same dark edge between the materials.

Also the edges of the polygons are slightly visible on the forehead. 
That was easily fixed by subdividing the mesh before rendering.

FlyerX


Post a reply to this message

From: clipka
Subject: Re: Subsurface Light Transport and mesh2
Date: 25 Mar 2012 09:50:46
Message: <4f6f22b6$1@news.povray.org>
Am 25.03.2012 10:58, schrieb FlyerX:

> I have been having some issues with the new Subsurface Light Transport.
> If the mesh is made up of several sub-meshes the seams between them are
> darkened.

I've never seen anything like this before with PoseRay output. (I very 
much presume you're using PoseRay for generating the POV-Ray mesh2 
objects yourself, right?)

> If I merge the mesh outside of POV-Ray into a single mesh with a single
> material for the complete head skin then the dark seams disappear. But I
> cannot merge the mesh outside POV-Ray since that would destroy the
> separate UV mapping that each mesh uses.

Theoretically it should be possible to merge into a mesh2 with multiple 
textures without loss.

> Also the edges of the polygons are slightly visible on the forehead.
> That was easily fixed by subdividing the mesh before rendering.

That's a known problem, and difficult to fix without some significant 
internal changes to surface normal handling. At present, subdividing is 
indeed the only known workaround.


Post a reply to this message

From: Stephen
Subject: Re: Subsurface Light Transport and mesh2
Date: 25 Mar 2012 10:58:13
Message: <4f6f3285$1@news.povray.org>
On 25/03/2012 2:50 PM, clipka wrote:
> Am 25.03.2012 10:58, schrieb FlyerX:
>
>> I have been having some issues with the new Subsurface Light Transport.
>> If the mesh is made up of several sub-meshes the seams between them are
>> darkened.
>
> I've never seen anything like this before with PoseRay output.

I have but unfortunately I cannot find any examples. It does seem to be 
dependent on the Poser models.

-- 
Regards
     Stephen


Post a reply to this message

From: FlyerX
Subject: Re: Subsurface Light Transport and mesh2
Date: 25 Mar 2012 16:17:17
Message: <4f6f7d4d@news.povray.org>
On 3/25/2012 8:50 AM, clipka wrote:
> Am 25.03.2012 10:58, schrieb FlyerX:
>
>> I have been having some issues with the new Subsurface Light Transport.
>> If the mesh is made up of several sub-meshes the seams between them are
>> darkened.
>
> I've never seen anything like this before with PoseRay output. (I very
> much presume you're using PoseRay for generating the POV-Ray mesh2
> objects yourself, right?)

Yes, this is a PoseRay output scene. The issue happens with all the 
models that have several materials on the same mesh. PoseRay breaks the 
model mesh by materials.

>
>> If I merge the mesh outside of POV-Ray into a single mesh with a single
>> material for the complete head skin then the dark seams disappear. But I
>> cannot merge the mesh outside POV-Ray since that would destroy the
>> separate UV mapping that each mesh uses.
>
> Theoretically it should be possible to merge into a mesh2 with multiple
> textures without loss.
>
>> Also the edges of the polygons are slightly visible on the forehead.
>> That was easily fixed by subdividing the mesh before rendering.
>
> That's a known problem, and difficult to fix without some significant
> internal changes to surface normal handling. At present, subdividing is
> indeed the only known workaround.


Post a reply to this message

From: FlyerX
Subject: Re: Subsurface Light Transport and mesh2
Date: 25 Mar 2012 16:19:52
Message: <4f6f7de8$1@news.povray.org>
On 3/25/2012 8:50 AM, clipka wrote:
> Am 25.03.2012 10:58, schrieb FlyerX:
>
>> I have been having some issues with the new Subsurface Light Transport.
>> If the mesh is made up of several sub-meshes the seams between them are
>> darkened.
>
> I've never seen anything like this before with PoseRay output. (I very
> much presume you're using PoseRay for generating the POV-Ray mesh2
> objects yourself, right?)
>
>> If I merge the mesh outside of POV-Ray into a single mesh with a single
>> material for the complete head skin then the dark seams disappear. But I
>> cannot merge the mesh outside POV-Ray since that would destroy the
>> separate UV mapping that each mesh uses.
>
> Theoretically it should be possible to merge into a mesh2 with multiple
> textures without loss.
>
>> Also the edges of the polygons are slightly visible on the forehead.
>> That was easily fixed by subdividing the mesh before rendering.
>
> That's a known problem, and difficult to fix without some significant
> internal changes to surface normal handling. At present, subdividing is
> indeed the only known workaround.

Here is the material I used on the skin:

#declare PR_DIFFUSE=pigment {color srgb <0.93,0.71,0.54> }
#declare F1=finish{specular 0 //Max highlight intensity: 0 -> 1
                    roughness 0.02203444 //surface roughness: 0.0005 -> 1
                    phong 0 phong_size 0 //Phong specular disabled

                    ambient rgb <0.00,0.00,0.00>
                    diffuse 0.6
                    reflection{0 } conserve_energy
                    subsurface {translucency 0.8}
}
material{
          texture{uv_mapping
                  pigment{PR_DIFFUSE  }
                  finish{F1}
                 }
  interior{ior 1}
         }

***and here are the global settings

global_settings {
subsurface {}
mm_per_unit 25
}


Post a reply to this message

From: clipka
Subject: Re: Subsurface Light Transport and mesh2
Date: 25 Mar 2012 19:35:53
Message: <4f6fabd9$1@news.povray.org>
Am 25.03.2012 22:19, schrieb FlyerX:

> interior{ior 1}

Not that I think it matters, but why ior 1? That would be air. Realistic 
skin should be something around 1.3.


Post a reply to this message

From: clipka
Subject: Re: Subsurface Light Transport and mesh2
Date: 25 Mar 2012 20:37:48
Message: <4f6fba5c$1@news.povray.org>
Am 25.03.2012 22:19, schrieb FlyerX:
> Here is the material I used on the skin:
>
> #declare PR_DIFFUSE=pigment {color srgb <0.93,0.71,0.54> }
> #declare F1=finish{specular 0 //Max highlight intensity: 0 -> 1
> roughness 0.02203444 //surface roughness: 0.0005 -> 1
> phong 0 phong_size 0 //Phong specular disabled
>
> ambient rgb <0.00,0.00,0.00>
> diffuse 0.6
> reflection{0 } conserve_energy
> subsurface {translucency 0.8}
> }
> material{
> texture{uv_mapping
> pigment{PR_DIFFUSE }
> finish{F1}
> }
> interior{ior 1}
> }

OK, here's the deal: To avoid SSLT seams between multiple objects, they 
currently must /share/ a common interior - it's not sufficient to have 
interiors with identical parameters, or even instances of the same 
#define'd interior. The only way to achieve this is to specify the 
interior in the parent CSG rather than the individual primitives.

Thus, to get the desired results, you need to:

- REMOVE any interior statements from the material.
- ADD the interior statement to the union or merge.
- For each part that needs a different ior (e.g. eyelashes or teeth) add 
an individual interior statement.


Post a reply to this message

From: FlyerX
Subject: Re: Subsurface Light Transport and mesh2
Date: 25 Mar 2012 23:40:52
Message: <4f6fe544$1@news.povray.org>
On 3/25/2012 7:37 PM, clipka wrote:
> Am 25.03.2012 22:19, schrieb FlyerX:
>> Here is the material I used on the skin:
>>
>> #declare PR_DIFFUSE=pigment {color srgb <0.93,0.71,0.54> }
>> #declare F1=finish{specular 0 //Max highlight intensity: 0 -> 1
>> roughness 0.02203444 //surface roughness: 0.0005 -> 1
>> phong 0 phong_size 0 //Phong specular disabled
>>
>> ambient rgb <0.00,0.00,0.00>
>> diffuse 0.6
>> reflection{0 } conserve_energy
>> subsurface {translucency 0.8}
>> }
>> material{
>> texture{uv_mapping
>> pigment{PR_DIFFUSE }
>> finish{F1}
>> }
>> interior{ior 1}
>> }
>
> OK, here's the deal: To avoid SSLT seams between multiple objects, they
> currently must /share/ a common interior - it's not sufficient to have
> interiors with identical parameters, or even instances of the same
> #define'd interior. The only way to achieve this is to specify the
> interior in the parent CSG rather than the individual primitives.
>
> Thus, to get the desired results, you need to:
>
> - REMOVE any interior statements from the material.
> - ADD the interior statement to the union or merge.
> - For each part that needs a different ior (e.g. eyelashes or teeth) add
> an individual interior statement.

Thanks. It works now. I just added a material{interior{ior 1.3}} at the 
end of the grouped skin meshes. Is union or merge better for SSLT? What 
about translucency color? is there a range that gives better results for 
human skin?

I used ior of 1 just to make sure that no other effects were giving me 
the dark edges.

FlyerX


Post a reply to this message

From: clipka
Subject: Re: Subsurface Light Transport and mesh2
Date: 26 Mar 2012 00:24:41
Message: <4f6fef89$1@news.povray.org>
Am 26.03.2012 05:40, schrieb FlyerX:

> Thanks. It works now. I just added a material{interior{ior 1.3}} at the
> end of the grouped skin meshes. Is union or merge better for SSLT?

When using mesh2 objects, it shouldn't make any difference

> What
> about translucency color? is there a range that gives better results for
> human skin?

I'm currently using translucency <2.0,1.5,0.8>; if I got the math right, 
that's somewhere in between the two measured "skin" data sets listed in 
That Original SSLT Paper. I do like the effect, at least with the 
high-quality DAZ texture I'm currently using.

Make sure to use the right mm_per_unit setting. I'm using 1000 mm per 
unit, and scaling the PoseRay output by an empiric factor of 1.8/78.


Post a reply to this message

From: FlyerX
Subject: Re: Subsurface Light Transport and mesh2
Date: 26 Mar 2012 06:04:04
Message: <4f703f14$1@news.povray.org>
On 3/25/2012 11:24 PM, clipka wrote:
> Am 26.03.2012 05:40, schrieb FlyerX:
>
>> Thanks. It works now. I just added a material{interior{ior 1.3}} at the
>> end of the grouped skin meshes. Is union or merge better for SSLT?
>
> When using mesh2 objects, it shouldn't make any difference
>
>> What
>> about translucency color? is there a range that gives better results for
>> human skin?
>
> I'm currently using translucency <2.0,1.5,0.8>; if I got the math right,
> that's somewhere in between the two measured "skin" data sets listed in
> That Original SSLT Paper. I do like the effect, at least with the
> high-quality DAZ texture I'm currently using.
>
> Make sure to use the right mm_per_unit setting. I'm using 1000 mm per
> unit, and scaling the PoseRay output by an empiric factor of 1.8/78.

Now with a texture, specular and a bump:

http://imgur.com/wkmWc

Seems to work well. I get some bright dots around the ear shadow. I am 
using +am2 +a0.1 +r3.

clipka, I checked the paper you refer to and I still uncertain about the 
translucency values. Is the value you are using the reduced scattering 
vector in Fig. 5?

regards,

FlyerX


Post a reply to this message

Goto Latest 10 Messages Next 3 Messages >>>

Copyright 2003-2023 Persistence of Vision Raytracer Pty. Ltd.