With the changes to the SSLT code in RC4 there is now only one 
controlling vector, translucency, instead of two, scattering and 
absorption. What if any is the relationship between translucency, 
scattering and absorption?
Scattering and absorption made sense to me and the reference pdf gave 
values for some materials. So how do we go about using translucency?

-- 
Regards
     Stephen

Post a reply to this message

On 02/27/2012 01:13 PM, Stephen wrote:
> With the changes to the SSLT code in RC4 there is now only one
> controlling vector, translucency, instead of two, scattering and
> absorption. What if any is the relationship between translucency,
> scattering and absorption?
> Scattering and absorption made sense to me and the reference pdf gave
> values for some materials. So how do we go about using translucency?
>
I find the new method much more simpler ... I started with the scene 
file that was added ~scenes/subsurface/subsurface.pov and played around 
with it changing things and observing the affect. It's a simple scene 
and renders fairly quickly. I then graduated to a little bit more 
complicated object then started playing with that ... move the light, 
and/or change the camera view. Hint cylinders make a better test object 
(the edges) than say rounded (sphere). Don't forget to play with the 
scaling (mm_per_unit: smaller value makes it more transparent at the 
edges, larger lessens the affect) ... keep it simple at first until you 
get the hang of it, then graduate to more complicated shapes.

BTW: RC5 is on deck because of the bug report you lodged. clipka's fix 
also fixed a couple of other issues I found during testing, so it won't 
be prone to crashes due to improper usage ... a bit more robust. Don't 
get discouraged, it's a cool feature and well worth the seemingly 
tedious learning curve ... lastly the use of color vectors with a zero 
component is still problematic ... zero diffuse as well IIRC

Post a reply to this message

On 02/27/2012 07:26 PM, James Holsenback wrote:
> On 02/27/2012 01:13 PM, Stephen wrote:
>> With the changes to the SSLT code in RC4 there is now only one
>> controlling vector, translucency, instead of two, scattering and
>> absorption. What if any is the relationship between translucency,
>> scattering and absorption?
>> Scattering and absorption made sense to me and the reference pdf gave
>> values for some materials. So how do we go about using translucency?
>>
> I find the new method much more simpler ... I started with the scene
> file that was added ~scenes/subsurface/subsurface.pov and played around
> with it changing things and observing the affect. It's a simple scene
> and renders fairly quickly. I then graduated to a little bit more
> complicated object then started playing with that ... move the light,
> and/or change the camera view. Hint cylinders make a better test object
> (the edges) than say rounded (sphere). Don't forget to play with the
> scaling (mm_per_unit: smaller value makes it more transparent at the
> edges, larger lessens the affect) ... keep it simple at first until you
> get the hang of it, then graduate to more complicated shapes.
>
> BTW: RC5 is on deck because of the bug report you lodged. clipka's fix
> also fixed a couple of other issues I found during testing, so it won't
> be prone to crashes due to improper usage ... a bit more robust. Don't
> get discouraged, it's a cool feature and well worth the seemingly
> tedious learning curve ... lastly the use of color vectors with a zero
> component is still problematic ... zero diffuse as well IIRC

OH yeah ... forgot to mention the to don't forget the interior {ior 
whatever} part ... bundle that with the rest of the texture definition 
as a material

Post a reply to this message

On 28/02/2012 12:26 AM, James Holsenback wrote:

> I find the new method much more simpler ...

Simplicity is not always best. Think about how many people complain 
about M$ windows and iTunes.

> I started with the scene
> file that was added ~scenes/subsurface/subsurface.pov and played around
> with it changing things and observing the affect. It's a simple scene
> and renders fairly quickly. I then graduated to a little bit more
> complicated object then started playing with that ... move the light,
> and/or change the camera view. Hint cylinders make a better test object
> (the edges) than say rounded (sphere).

A good suggestion. It looks like I will need to run a series of 
experiments instead of just being able to convert my existing scenes.

Don't forget to play with the
> scaling (mm_per_unit: smaller value makes it more transparent at the
> edges, larger lessens the affect) ... keep it simple at first until you
> get the hang of it, then graduate to more complicated shapes.
>

Bah! Kiddy stuff. :-P
It is generally with meshes that I have been using it with.

> BTW: RC5 is on deck because of the bug report you lodged.

That is fine, Chris sent me an evaluation copy to test. :-)

> clipka's fix
> also fixed a couple of other issues I found during testing, so it won't
> be prone to crashes due to improper usage ... a bit more robust.

That is good.

> Don't
> get discouraged, it's a cool feature and well worth the seemingly
> tedious learning curve ... lastly the use of color vectors with a zero
> component is still problematic ... zero diffuse as well IIRC

That is understandable. A totally black surface will not have any 

hard to imagine but what do I know? ?_?

> OH yeah ... forgot to mention the to don't forget the interior {ior
> whatever} part ... bundle that with the rest of the texture definition
> as a material

It is easy to notice if you have forgotten it. The render is too fast. :-)

What I really wanted is an understanding of what the translucency values 
mean in RL.
To my way of thinking a value of <1, 1, 1> would mean that the material 
is completely translucent and all the colour components are allowed to 
penetrate unhindered. Why can we use a value > 1?

Yes some experimenting is needed.




-- 
Regards
     Stephen

Post a reply to this message

On 02/28/2012 06:10 AM, Stephen wrote:
> Bah! Kiddy stuff. :-P
> It is generally with meshes that I have been using it with.

simpler shapes ... walk before ya run ;-)

>> Don't
>> get discouraged, it's a cool feature and well worth the seemingly
>> tedious learning curve ... lastly the use of color vectors with a zero
>> component is still problematic ... zero diffuse as well IIRC
>
> That is understandable. A totally black surface will not have any

> hard to imagine but what do I know? ?_?

not just rgb 0 ,.... rgb <1,0,0> should be rgb <1,0.01,0.01> .... etc

Post a reply to this message

On 28/02/2012 11:23 AM, James Holsenback wrote:
> On 02/28/2012 06:10 AM, Stephen wrote:
>> Bah! Kiddy stuff. :-P
>> It is generally with meshes that I have been using it with.
>
> simpler shapes ... walk before ya run ;-)
>

I know, I know. But... ;-)

>>> Don't
>>> get discouraged, it's a cool feature and well worth the seemingly
>>> tedious learning curve ... lastly the use of color vectors with a zero
>>> component is still problematic ... zero diffuse as well IIRC
>>
>> That is understandable. A totally black surface will not have any

>> hard to imagine but what do I know? ?_?
>
> not just rgb 0 ,.... rgb <1,0,0> should be rgb <1,0.01,0.01> .... etc
>



I have just made a cylinder mesh and am about to start experimenting.



-- 
Regards
     Stephen

Post a reply to this message

Am 27.02.2012 19:13, schrieb Stephen:
> With the changes to the SSLT code in RC4 there is now only one
> controlling vector, translucency, instead of two, scattering and
> absorption. What if any is the relationship between translucency,
> scattering and absorption?
> Scattering and absorption made sense to me and the reference pdf gave
> values for some materials. So how do we go about using translucency?

There were various major drawbacks with the parameterization based on 
scattering and absorption.

First, while they allowed to plug in physical properties of a material 
directly, sources for such data is scarce (that one pdf is the only 
source I'm currently aware of), and tweaking those properties to mimic 
any other desired material - even just a similar material with different 
color, or a material with same color but stronger or weaker SSLT effect 
- is non-trivial to say the least.

Second, those two parameters fully determined the material's color, 
leaving the classic pigment parameter with no effect; this resulted in 
more maintenance effort if you wanted test renders without SSLT to use 
similar colors.

And last not least, with those parameters it was a major challenge - if 
not impossible - to make use of classic image textures (like, say, Poser 
skin textures) together with SSLT.


The new parameterization makes it straightforward to achieve a 
particular effective color by simply specifying that color as a pigment; 
this leaves you with only one RGB parameter to toy around if you want to 
mimic a particular material, automatically keeps SSLT and non-SSLT color 
of a material in sync, and enables you to change the material color - or 
apply standard image textures - in a pinch.

The new parameterization also imposes no added limitations: Any material 
you could model with the old parameterization can also be modeled with 
the new one.


The translucency parameter specifies the so-called "mean free distance" 
of a light ray in the material (in mm), i.e. the average distance a 
light ray travels before it is either absorbed or scattered. This can be 
more than 1.0 mm in case of very "waxy" materials, and could 
theoretically be as large as infinity (for a perfectly transparent 
material).

I've only recently heard of Poser figures coming with translucency maps, 
which obviously need to specify values in the range 0..1; it would be 
interesting to know what parameterization is used there, so we could 
adapt POV-Ray to easily make use of them as well; the SSLT syntax is 
therefore still in flow.

Post a reply to this message

On 28/02/2012 4:44 PM, clipka wrote:
>
> The translucency parameter specifies the so-called "mean free distance"
> of a light ray in the material (in mm), i.e. the average distance a
> light ray travels before it is either absorbed or scattered. This can be
> more than 1.0 mm in case of very "waxy" materials, and could
> theoretically be as large as infinity (for a perfectly transparent
> material).
>
Thanks Christoph I understand it better now that I know what the values 
related to. I can get a mental image of what is happening.

that too much to ask for?

> I've only recently heard of Poser figures coming with translucency maps,
> which obviously need to specify values in the range 0..1; it would be
> interesting to know what parameterization is used there, so we could
> adapt POV-Ray to easily make use of them as well; the SSLT syntax is
> therefore still in flow.

I've not heard of that. Interesting!
-- 
Regards
     Stephen

Post a reply to this message

clipka wrote:

> The translucency parameter specifies the so-called "mean free
> distance" of a light ray in the material (in mm), i.e. the average
> distance a light ray travels before it is either absorbed or
> scattered. This can be more than 1.0 mm in case of very "waxy"
> materials, and could theoretically be as large as infinity (for a
> perfectly transparent material).

It probably depends on photon wavelength as well, although I'm not
sure that effect is strong enough to impact the visual perception of
an object with translucency.

Post a reply to this message

Am 28.02.2012 20:28, schrieb Christian Froeschlin:
> clipka wrote:
>
>> The translucency parameter specifies the so-called "mean free
>> distance" of a light ray in the material (in mm), i.e. the average
>> distance a light ray travels before it is either absorbed or
>> scattered. This can be more than 1.0 mm in case of very "waxy"
>> materials, and could theoretically be as large as infinity (for a
>> perfectly transparent material).
>
> It probably depends on photon wavelength as well, although I'm not
> sure that effect is strong enough to impact the visual perception of
> an object with translucency.

The wavelength-dependency of the mean free distance is indeed quite 
significant in some materials; that's why translucency is an RGB parameter.

Post a reply to this message