 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Following plane with bozo normals looks good as long bump_size is low (less
than 5). When I set bozo to high value as 50 then reflection should come
very blurry and should look similar to radiosity-effect. But in POV-Ray,
reflection on high bump_sized normals looks more shinny than blurry...
WHY?
Thanks for reply... best regards, zuegs
my scene:
camera {
location <0, 2, -10>
look_at 0
}
plane { y, 0
finish { reflection 1.0 ambient 0.0 diffuse 0.0 specular 0.0 }
normal { bozo 50 scale 0.001 }
}
box { <-1, 0, -1>, <1, 5, 1>
pigment { color rgb <1, 1, 0> }
finish { ambient 1.0 diffuse 0.0 specular 0.0 reflection 0.0 }
}
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Hello all,
like to rethrow this theme to hopefully get some comments from you, or from
a developper...
The above behavior source is in "lightning.cpp". The variables meaning are:
- Jitter_Raw_Normal: is the normal vector of the surface.
- Jitter_Normal: is the normal vector of the disturbed surface
(normal{bozo})
- Ray->Direction: is the vector of incomming light
- New_Ray->Direction: is the vector of the reflected light
-----------
01 VDot(n,Ray->Direction, Jitter_Normal);
02 n *= -2.0;
03 VAddScaled(New_Ray.Direction, Ray->Direction, n, Jitter_Normal);
04 /* Nathan Kopp & CEY 1998 - Reflection bugfix
05 if the new ray is going the opposet direction as raw normal, we
06 need to fix it.
07 */
08 VDot(n, New_Ray.Direction, Jitter_Raw_Normal);
09 if (n < 0.0)
10 {
11 /* It needs fixing. Which kind? */
12 VDot(n2,New_Ray.Direction,Jitter_Normal);
13 if (n2 < 0.0)
14 {
15 /* reflected inside rear virtual surface. Reflect Ray using
16 Raw_Normal */
17 VDot(n,Ray->Direction,Jitter_Raw_Normal);
18 n *= -2.0;
19 VAddScaled(New_Ray.Direction, Ray->Direction,
n,Jitter_Raw_Normal);
20 }
21 else
22 {
23 /* Double reflect NRay using Raw_Normal */
24 /*VDot(n,New_Ray.Direction,Jitter_Raw_Normal); - kept the old n
25 around */
26 n *= -2.0;
27 VAddScaledEq(New_Ray.Direction, n, Jitter_Raw_Normal);
28 }
29 }
30 VNormalizeEq(New_Ray.Direction);
-------------
My problem happens when condtion on line 13 gets true, this happens when
incomming ray is behind the disturbed surface Jitter_Normal. The programmer
decided to reflect incomming ray on non-disturbed surface what creates the
fully mirror effect.
The more the normal is disturbed (big bump_size) so bigger is the
probability that incomming ray can be behind disturbed normal.
Why not apply Double reflection in this case too as on line 23-27?
Should I better post this in "bugreports"?
Please help me understand...
zuegs
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Hi all...
i can't solve this myself...
....anyone a idea on above...
Please, please, please
zuegs
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Hello Zuegs,
I'm not actually answering your question here, but I'm hoping that if I can
clarify the problem then someone with more knowledge than me may step in and
comment.
I've had a look at your code and rendered it with various alternative bozo
settings.
I've looked at your description of the problem and when you say:
>
> When I set bozo to high value as 50 then reflection should come
> very blurry and should look similar to radiosity-effect. But in
> POV-Ray, reflection on high bump_sized normals looks more
> shinny than blurry...
>
I think you mean that as you pass from bozo 0 to bozo 5 the reflected edges
of your box become progressively more blurred. As you pass from bozo 5 up to
bozo 50, you get a greater distribution of reflected light across the
reflecting surface, but the reflected edges of your box start to become well
defined again, until you get really quite crisp and distinctive reflections
of the edges of the box.
I've increased the scale of the finish and zoomed in on a portion of the
reflection to illustrate this effect. I've also added a little blue sphere
where the edge of the box meets the surface, so everything below the blue
dot is a reflection of the box above the blue dot.
camera {
location <0, 2, -10>
look_at <1,0,-1>
angle 15
}
plane { y, 0
finish { reflection 1.0 ambient 0.0 diffuse 0.0 specular 0.0 }
normal { bozo 15 scale 0.3 }
}
box { <-1, 0, -1>, <1, 5, 1>
pigment { color rgb <1, 1, 0> }
finish { ambient 1.0 diffuse 0.0 specular 0.0 reflection 0.0 }
}
sphere {<1,0,-1>,0.02 pigment {color <0,1,1>} finish {ambient 1.0}}
If you reduce this to bozo 3, you get little or no straight edges to the
reflected image, whereas above bozo 3 you get progressively more straight
edge to the reflection. This is what I think you're referring to in your
second note where you say:
>
> The programmer decided to reflect incomming ray on
> non-disturbed surface what creates the fully mirror effect.
>
I don't fully follow the POV-Ray source code you included in your second
note, but it seems to me that, if the ray strikes the surface where the bozo
normal is close to perpendicular to the actual surface, but leaning more
towards the camera than the box, that the code treats it as if it is exactly
perpendicular and therefore produces a 'flat' reflection, rather than
perform the 'real' calculation of the shape of the reflection. Whereas if
it's close to vertical but leaning more towards the box, it performs a more
detailed calculation, resulting in no straight edge.
When the bozo setting is relatively low, this approximation is fine and as
it increases, the approximation becomes more apparent.
Did I correctly describe the effect you're having problems with?
Regards,
Chris B.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Hi Chris B
I'am very pleased for your reply, thanks.
Yes I think you fully understood my problem, thanks again.
source-code:
if the bozo-normal is leaning towards the box so that the camera gets behind
the normal/surface, the code treats this by reflecting it as surface is
'flat'.
The more the camera is down (near surface with quite parallel view) the more
often it happens that the camera gets behind the bozo-normal. So surface
looks flat... (the lower the camera less bozo is needed to get this
artefacts)
It would be great if one of the programmers could have a look on this
because i think that the current implementation of reflection leads to
artefacts (wrong reflection) in this case and this seams to be a bug for
me.... am i wrong???
best regards
zuegs
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"zuegs" <zue### [at] bluewin ch> wrote in message
news:web.43da25b14d3eab5823b7640@news.povray.org...
> Hi Chris B
>
> I'am very pleased for your reply, thanks.
> Yes I think you fully understood my problem, thanks again.
>
> source-code:
> if the bozo-normal is leaning towards the box so that the camera gets
> behind
> the normal/surface, the code treats this by reflecting it as surface is
> 'flat'.
> The more the camera is down (near surface with quite parallel view) the
> more
> often it happens that the camera gets behind the bozo-normal. So surface
> looks flat... (the lower the camera less bozo is needed to get this
> artefacts)
>
> It would be great if one of the programmers could have a look on this
> because i think that the current implementation of reflection leads to
> artefacts (wrong reflection) in this case and this seams to be a bug for
> me.... am i wrong???
>
> best regards
> zuegs
>
>
I'm not sure I'd call it a bug, because it seems to me to be related to the
behaviour of a part of the surface that, if the surface was bumpy for real
rather than a flat surface with distortion of the normals, would be out of
sight of the camera. Because it is really just a flat surface, the point is
visible and the programmer had to define a reflective behaviour which, in a
real environment, would be undefined.
There are probably lots of scenes that would be significanlty altered if the
default behaviour were to be changed, but you could suggest an optional
setting to handle this condition in the way you described.
Also you might try fiddling with the reflection setting which you can use to
reduce the reflection for low angles by using it in the reverse fashion to
that described in the documentation. ie. by telling it to reflect highly at
high angles and less at low angles.
The following example takes the bozo to 50 using the second reflection
parameter to compensate for the effect you're having a problem with.
It's a bit of a cheat, but it may serve as a workaround pending a better
solution.
camera {
location <0, 2, -10>
look_at 0
}
plane { y, 0
finish { reflection {3.0 0.5} ambient 0.0 diffuse 0.0 specular 0.0 }
normal { bozo 50 scale 0.001 }
}
box { <-1, 0, -1>, <1, 5, 1>
pigment { color rgb <1, 1, 0> }
finish { ambient 1.0 diffuse 0.0 specular 0.0 reflection 0.0 }
}
Regards,
Chris B.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Hi Chris B
Thanks for your reply!
I see that it is difficult to define a reflection behavior that doesn't
exist in real environment. Having read some publications on this topic over
weekend I found that suggesting a perfect handling for describted condition
wouldn't be possible.
But while reviewing source-code I feeled that the "else handling" on second
condition (line 23-27 in code above) would give a better result as current
handling (line 15-19).
So i suggest to just to remove second "if" and let the "else"-code in place:
11 /* It needs fixing. Which kind? */
23 /* Double reflect NRay using Raw_Normal */
24 /*VDot(n,New_Ray.Direction,Jitter_Raw_Normal); - kept the old n
25 around */
26 n *= -2.0;
27 VAddScaledEq(New_Ray.Direction, n, Jitter_Raw_Normal);
I meen to double reflect in every case when "New_Ray.Direction" is behind
"Jitter_Raw_Normal". So the resulting normal stays bumpy but just flips on
good side of surface. Ok, this is not the most accurate and physical
solution but it gives a much better result than just a flat reflection.
> There are probably lots of scenes that would be significanlty altered if the
> default behaviour were to be changed, but you could suggest an optional
> setting to handle this condition in the way you described.
I think, for most scenes with small bumpy there will be no significantly
change because they will rarely match condition that "New_Ray.Direction" is
behind "Jitter_Raw_Normal". One of the reason, that noone noticed this
"reflection-problem" yet.
As I was not yet able to compile pov-ray myself, i coded a simple raytracer
with current and new reflection handling. I posted a picture here:
http://www.rowen.ch/pov/povref1.jpg
On left-side is the default povray handling and at the right-side is the new
handling. Up-side without aliasing, down-side with aliasing.
As you can see, with new handling the reflection stays bumpy/blurry and
gives a more realistic result for me.
I would feel very happy if you could tell me what you think about...
.... with best regards
zuegs
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"zuegs" <zue### [at] bluewinch> wrote in message
news:web.43dd45544d3eab55f1e9ae40@news.povray.org...
> Hi Chris B
>
> ... snip ...
>
> http://www.rowen.ch/pov/povref1.jpg
>
> On left-side is the default povray handling and at the right-side is the
> new
> handling. Up-side without aliasing, down-side with aliasing.
>
> As you can see, with new handling the reflection stays bumpy/blurry and
> gives a more realistic result for me.
>
> I would feel very happy if you could tell me what you think about...
> .... with best regards
>
> zuegs
>
Hi Zuegs,
I'm with you. I think your version looks better.
I'm not sure what the correct route would be to propose this as a change to
the POV-ray source code.
I see that there are newsgroups on povray.programming and on
povray.unofficial.patches.
Also I think Mega-POV includes unofficial patches, so it may be worth
contributing to that.
Because it potentially affects any surface with a pronounced normal and
where the line of sight of the camera is close to parallel to the surface, I
suspect there'll be more chance of adoption if it's a parameter controlled
change, as this would reduce the risk of interfering with existing scenes.
Good luck,
Chris B.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
|
|
