 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
While developing the attached images I was manipulating different
lighting and finish effects to get a certain design result rather than
absolute photo realism.
Working with shadowless objects I was balancing the diffuse with the
reflection values in the finish of the reflecting plane. When I turned
fresnel "on" I got a result that both surprised and pleased me. It is
demonstrated in the second inmage. It surprised me because I thought the
fresnel function required variable reflection to be specified with a max
*and* a min value, but I was using just one value. Turns out the inverse
seems to be true, you can turn off all variable reflection by specifying
both max and min values with max = min.
This lead to running a few more tests which can be seen here:
http://www21.brinkster.com/jrcsurvey/fresnel/
Surprising to me also is the effect of ior = 1, now reflection seems to
be cancelled unless different mas and min values are put in explicitely.
Anyway, any preferences when comparing the attached images?
Can anyone help explain some of the effects I am seeing?
Post a reply to this message
Attachments:
Download 'fresnel.00001.jpg' (48 KB)
Download 'fresnel.00002.jpg' (57 KB)
Preview of image 'fresnel.00001.jpg'
Preview of image 'fresnel.00002.jpg'
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Jim Charter wrote:
> While developing the attached images I was manipulating different
> lighting and finish effects to get a certain design result rather than
> absolute photo realism.
>
> Working with shadowless objects I was balancing the diffuse with the
> reflection values in the finish of the reflecting plane. When I turned
> fresnel "on" I got a result that both surprised and pleased me. It is
> demonstrated in the second inmage. It surprised me because I thought the
> fresnel function required variable reflection to be specified with a max
> *and* a min value, but I was using just one value. Turns out the inverse
> seems to be true, you can turn off all variable reflection by specifying
> both max and min values with max = min.
>
> This lead to running a few more tests which can be seen here:
>
> http://www21.brinkster.com/jrcsurvey/fresnel/
>
> Surprising to me also is the effect of ior = 1, now reflection seems to
> be cancelled unless different mas and min values are put in explicitely.
>
> Anyway, any preferences when comparing the attached images?
> Can anyone help explain some of the effects I am seeing?
Fresnel reflection describes the reflection of light from a boundary between
two materials, one of which is probably air in your case. The index of
refraction (ior) equals 1 for air and e.g. about 1.5 for glass. ior=1 means
a boundary between air and air which means there isn't any. From a
non-existent boundary there is no reflected light. That's what you
observed.
The effect of reflection observed in you second image is the following. The
Fresnel equations describe the variation of reflectivity with the angle of
incidence of the light. Reflectivity is close to 1 for grazing incident
light and becomes less the steeper the angle of incidence is. Therefore,
the plane supporting your shoe seems bright close to the horizont and
darker close to the observer.
I'd suggest, you'd take an optics book and go through the basic chapters
again...
Alf
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Alf Burau wrote:
> Jim Charter wrote:
>
>
>>While developing the attached images I was manipulating different
>>lighting and finish effects to get a certain design result rather than
>>absolute photo realism.
>>
>>Working with shadowless objects I was balancing the diffuse with the
>>reflection values in the finish of the reflecting plane. When I turned
>>fresnel "on" I got a result that both surprised and pleased me. It is
>>demonstrated in the second inmage. It surprised me because I thought the
>>fresnel function required variable reflection to be specified with a max
>>*and* a min value, but I was using just one value. Turns out the inverse
>>seems to be true, you can turn off all variable reflection by specifying
>>both max and min values with max = min.
>>
>>This lead to running a few more tests which can be seen here:
>>
>>http://www21.brinkster.com/jrcsurvey/fresnel/
>>
>>Surprising to me also is the effect of ior = 1, now reflection seems to
>>be cancelled unless different mas and min values are put in explicitely.
>>
>>Anyway, any preferences when comparing the attached images?
>>Can anyone help explain some of the effects I am seeing?
>
>
> Fresnel reflection describes the reflection of light from a boundary between
> two materials, one of which is probably air in your case. The index of
> refraction (ior) equals 1 for air and e.g. about 1.5 for glass. ior=1 means
> a boundary between air and air which means there isn't any. From a
> non-existent boundary there is no reflected light. That's what you
> observed.
>
> The effect of reflection observed in you second image is the following. The
> Fresnel equations describe the variation of reflectivity with the angle of
> incidence of the light. Reflectivity is close to 1 for grazing incident
> light and becomes less the steeper the angle of incidence is. Therefore,
> the plane supporting your shoe seems bright close to the horizont and
> darker close to the observer.
>
> I'd suggest, you'd take an optics book and go through the basic chapters
> again...
Thanks for replying
So I take it if you give the finish
reflection { .96 fresnel on } with interior { ior 1.5 }
then .96 is taken as the high reflectivity value as angle of incidence
approaches 0 and the low reflectively value, at angle of incidence 90,
is derived from the ior value?
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Jim Charter wrote:
<snip>
> Thanks for replying
>
> So I take it if you give the finish
> reflection { .96 fresnel on } with interior { ior 1.5 }
> then .96 is taken as the high reflectivity value as angle of incidence
> approaches 0 and the low reflectively value, at angle of incidence 90,
> is derived from the ior value?
Jim,
frankly spoken, I am not so familiar with SDL. The only thing I can do for
sure is to help you understand the Fresnel reflection.
The angle of incidence is usually measured between the direction of the
incident light and the surface normal. Therefore, a 90 degree angle of
incidence means light parallel to the surface, 0 degree angle of incidence
means light parallel to the suface normal, i.e. perpendicular to the
surface. The Fresnel reflection coefficients describe the amount of light
being reflected off a surface, depending on the angle of incidence. In two
cases the values for the Fresnel coefficients can be determined easily.
The first case is the 90 degree angle of inciedence. This is not really a
reflection, since the light is parallel to the surface. In this case no
light enters the surface and the reflectivity equals 1, i.e. all light is
"reflected".
The second case is the 0 degree angle of incidence. In this case some light
enters the surface and some light is reflected. The amount of reflected
intensity is
R = | (n-1)/(n+1) |**2,
where n ist the index of refraction (ior). This is easy to calculate for
transparent media. For absorbing media, n becomes a complex number but the
formulae do not change. For all intermediate angles the expression for the
Fresnel coeffitients becomes more difficult and the reflection coeffitients
are different for different polarization states of light.
In my opinion, the Fresnel reflection should be well-defined by just giving
the ior. Since povray cannot handle abssorbing media by simply taking a
complex ior, it might be neccessary to give an additional parameter.
However, did you consult the povray manual?
I am not sure, if this does help you.
Alf
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Jim Charter wrote:
<snip>
> Thanks for replying
>
> So I take it if you give the finish
> reflection { .96 fresnel on } with interior { ior 1.5 }
> then .96 is taken as the high reflectivity value as angle of incidence
> approaches 0 and the low reflectively value, at angle of incidence 90,
> is derived from the ior value?
Jim,
maybe this will help you:
<http://www.povray.org/documentation/view/3.6.1/348/>
Alf
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Alf Burau wrote:
> Jim Charter wrote:
>
> <snip>
>
>>Thanks for replying
>>
>>So I take it if you give the finish
>>reflection { .96 fresnel on } with interior { ior 1.5 }
>>then .96 is taken as the high reflectivity value as angle of incidence
>>approaches 0 and the low reflectively value, at angle of incidence 90,
>>is derived from the ior value?
>
>
> Jim,
>
> maybe this will help you:
>
> <http://www.povray.org/documentation/view/3.6.1/348/>
>
> Alf
Thanks Alf, yes I was familiar with that page of the manual. It is not
a life and death thing I am just blathering. But you see my confusion
right? The manual doesn't really say what happens if both "variable
reflection" is invoked ie. a min_value and a max_value are specified,
and fresnel is on. It treats treats these nore like they are exclusive
situations. I, meanwhile had thought they were dependent on each other.
My long time assumption was that fresnel was some sort of falloff
curve that was scaled to "fit between" the two values. So I did not
expect to see variation when I specified only one value.
So I guess it is like this:
Always the "standard" ( presumably linear ) function is used if non zero
reflection is specified with the single value specified taken as min =
max = value. If two values are there, the slope of the falloff is max/min.
If fresnel is specified it overrides the standard function except in the
case of ior = 1, then ? If one reflection value is specified, the
fresnel equation which returns zero reflectivity seems to take
precedence, but if two values are specified, the standard function
seems to be used to return a reflectivity value but a lower one than it
would if fresnel is off. I just wondered if all this is folded into
some universal formula combining the two functions, that I am too dense
to see.
Anyway I will research fresnel as you suggest and probably come to look
back on this whole post with embarrassment. And I hadn't realized that
it related so directly to ior, I just thought it was some sort of curve
that happened to model reflectiveity falloff with increased accuracy.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
|
|
