 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Hi,
I want to use POV-Ray to simulate an physicalic correct x-ray image. To do that
I use completely transparent bodies (filter 1) and the physicalic correct fade
distance for the used materials. To test this, I create two boxes on top of each
other. The code looks like this (don't
wonder about the small values, I need to work with millimeters):
//----------------------------
#include "colors.inc"
#declare Nothing = texture {pigment {color White filter 1} }
#declare Metall1 = interior{ fade_power 1000 fade_distance 8.8865E-5}
#declare Metall2 = interior{ fade_power 1000 fade_distance 3.01368E-4}
light_source { <0,0,0.1> color rgb <1,1,1> }
camera {orthographic location <0,0,-0.1> look_at <0,0,0> angle 5 }
background { color White }
box { <-2E-3, -0.5E-3, 0>, <2E-3, 0.5E-3, 0.03E-3>
texture {Nothing}
interior {Metall1}
}
box { <-0.5E-3, -2E-3, 0.03E-3>, <0.5E-3, 2E-3, 0.06E-3>
texture {Nothing}
interior {Metall2}
}
//----------------------------
When you try this code you will notice, that the boxes are not transparent on
their touching area. If you "push" the upper box a little bit higher so they
don't touch each other any more (1E-6 is enough), then you get the expected and
correct image.
//----------------------------
#include "colors.inc"
#declare Nothing = texture {pigment {color White filter 1} }
#declare Metall1 = interior{ fade_power 1000 fade_distance 8.8865E-5}
#declare Metall2 = interior{ fade_power 1000 fade_distance 3.01368E-4}
light_source { <0,0,0.1> color rgb <1,1,1> }
camera {orthographic location <0,0,-0.1> look_at <0,0,0> angle 5 }
background { color White }
box { <-2E-3, -0.5E-3, 0>, <2E-3, 0.5E-3, 0.03E-3>
texture {Nothing}
interior {Metall1}
}
box { <-0.5E-3, -2E-3, 0.03E-3+1E-6>, <0.5E-3, 2E-3, 0.06E-3+1E-6>
texture {Nothing}
interior {Metall2}
}
//----------------------------
Has anybody an idea why this happens and how I can avoid that effect without
"pushing" bodies around?
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Sounds like a coincident surface issue. One thing you can do is use merge{}
with touching or overlapping objects, however this does tend to add to the
render time, sometimes significantly
-tgq
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"Trevor G Quayle" <Tin### [at] hotmail com> wrote:
> Sounds like a coincident surface issue. One thing you can do is use merge{}
> with touching or overlapping objects, however this does tend to add to the
> render time, sometimes significantly
>
> -tgq
Thank you for your quick response. I tried to merge the two boxes but then it
seems like the touching area is not effected by the fade distance any more -
this area gets comletely white, which is not correct. The problem is, that every
used material has to use it's own fade distances to generate a correct x-ray
image.
Other ideas to solve this problem with the coincident surfaces?
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Am 16.02.2011 14:54, schrieb MrS1203:
> Other ideas to solve this problem with the coincident surfaces?
No. It /is/ a known problem, and the only known solution is to avoid
coincident surfaces wherever they're visible.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
clipka <ano### [at] anonymousorg> wrote:
> Am 16.02.2011 14:54, schrieb MrS1203:
>
> > Other ideas to solve this problem with the coincident surfaces?
>
> No. It /is/ a known problem, and the only known solution is to avoid
> coincident surfaces wherever they're visible.
Ok, thank you. Then I have to seperate these surfaces with a "push". Not very
nice but it's bearable.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> Hi,
>
> I want to use POV-Ray to simulate an physicalic correct x-ray image. To do that
> I use completely transparent bodies (filter 1) and the physicalic correct fade
> distance for the used materials. To test this, I create two boxes on top of each
> other. The code looks like this (don't
> wonder about the small values, I need to work with millimeters):
>
> //----------------------------
> #include "colors.inc"
> #declare Nothing = texture {pigment {color White filter 1} }
> #declare Metall1 = interior{ fade_power 1000 fade_distance 8.8865E-5}
> #declare Metall2 = interior{ fade_power 1000 fade_distance 3.01368E-4}
>
> light_source {<0,0,0.1> color rgb<1,1,1> }
> camera {orthographic location<0,0,-0.1> look_at<0,0,0> angle 5 }
> background { color White }
>
>
> box {<-2E-3, -0.5E-3, 0>,<2E-3, 0.5E-3, 0.03E-3>
> texture {Nothing}
> interior {Metall1}
> }
>
>
You need to avoid coincident surfaces, so, you need to fudge your
objects dimentions a little so that ther surfaces are very close without
actualy touching.
A side note:
You don't need to use such small values, even working in mm. A POV-unit
can be any "real world" unit you want, from Parsects (or more) down to
Armstrong and smaller... All you need to do is to make your model
assuming that 1 unit = 1 mm.
All you need to di is to remove those E-3 from your dimentions, or add
+3 to the expoment.
Alain
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Am 17.02.2011 02:25, schrieb Alain:
> A side note:
> You don't need to use such small values, even working in mm. A POV-unit
> can be any "real world" unit you want, from Parsects (or more) down to
> Armstrong and smaller... All you need to do is to make your model
> assuming that 1 unit = 1 mm.
You probably mean Angstrom?
Well, given that visible light has a wavelength of ~400-800 nm =
6000-8000 Angstrom, rendering results probably won't be too realistic at
those dimensions :-P
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> Am 17.02.2011 02:25, schrieb Alain:
>
>> A side note:
>> You don't need to use such small values, even working in mm. A POV-unit
>> can be any "real world" unit you want, from Parsects (or more) down to
>> Armstrong and smaller... All you need to do is to make your model
>> assuming that 1 unit = 1 mm.
>
> You probably mean Angstrom?
>
> Well, given that visible light has a wavelength of ~400-800 nm =
> 6000-8000 Angstrom, rendering results probably won't be too realistic at
> those dimensions :-P
If you use hard gamma rays as a light source and false colour it...
When working in microscopic scale, NO imaging is going to be realistic,
even a photo. Usualy, when going under the micrometer scale, colours
tend to loose any meaning. The same also appens to highlights soon after
that.
In the photos we see, any colour is from extremely concentrated dies and
post processing. Under a microscope, without dies, any cell looks prety
much colourless transparent, with the occasional speck of colour.
Alain
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Alain <aze### [at] qwertyorg> wrote:
> You need to avoid coincident surfaces, so, you need to fudge your
> objects dimentions a little so that ther surfaces are very close without
> actualy touching.
>
> A side note:
> You don't need to use such small values, even working in mm. A POV-unit
> can be any "real world" unit you want, from Parsects (or more) down to
> Armstrong and smaller... All you need to do is to make your model
> assuming that 1 unit = 1 mm.
> All you need to di is to remove those E-3 from your dimentions, or add
> +3 to the expoment.
Thanks for your response. My plan is to import objects from another software
into POV-Ray. From this software I only get a huge amount of points, which have
to meshed in POV-Ray. To change the value of one coordinate (such as the
z-value) before meshing those points is no problem. To change the whole
dimension of the objects is sadly not possible, the objects are not symmetric.
Or is there any way to shrink an object which is made out of meshed points?
I have to use these small values because I get them from the other software
where I have to use SI-units. To test my approach I used small values too. It
would be possible to multiply all coordinates with 1000 to get "normal" values.
But I don't see any advantages if I would do that. Or do I overlook something?
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> Alain<aze### [at] qwertyorg> wrote:
>> You need to avoid coincident surfaces, so, you need to fudge your
>> objects dimentions a little so that ther surfaces are very close without
>> actualy touching.
>>
>> A side note:
>> You don't need to use such small values, even working in mm. A POV-unit
>> can be any "real world" unit you want, from Parsects (or more) down to
>> Armstrong and smaller... All you need to do is to make your model
>> assuming that 1 unit = 1 mm.
>> All you need to di is to remove those E-3 from your dimentions, or add
>> +3 to the expoment.
>
> Thanks for your response. My plan is to import objects from another software
> into POV-Ray. From this software I only get a huge amount of points, which have
> to meshed in POV-Ray. To change the value of one coordinate (such as the
> z-value) before meshing those points is no problem. To change the whole
> dimension of the objects is sadly not possible, the objects are not symmetric.
> Or is there any way to shrink an object which is made out of meshed points?
If you multiply all the coordinates of an object by a constant value,
you effectively scale that object. If the object is not at the origin,
the object will also move radialy.
Once the mesh have been created, you can scale it like any other object.
A scale like: scale<1, 1, 0.999999> will cause almost no distortion,
shrink by 1 millionth along the Z axis, and is enough to prevent a
coincident surfaces.
>
> I have to use these small values because I get them from the other software
> where I have to use SI-units. To test my approach I used small values too. It
> would be possible to multiply all coordinates with 1000 to get "normal" values.
> But I don't see any advantages if I would do that. Or do I overlook something?
>
>
>
>
Working in mm with units of 1 m, not realy problematic unless you also
have some rather small values. By small, I think under 0.001mm or less.
If you use to small values, you can begin to have floating point
precision and rounding errors.
It can manifest by areas that are suposed to be lighted been dark or
areas that are suposed to be shadowed been lighted. It also can manifest
itself by having light "leaking" from some corners.
If that appens, then, yes, you may need to scale everything up. Place
all your objects into one big union and scale that union up as needed.
BUT. If all your values are within some reasonable range, there should
be no real problem. Reasonable usualy been a range of about 4 order of
magnitude.
Just try to not put your light to far away, like over 1000m away when
working in mm. If you need a very distant light, use the parallel
attribute and place it closer. "parallel" simulate a light situated at
an "infinite" distance.
Alain
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
|
|
