> Alain <kua### [at] videotronca> wrote:

>>> Image to
>>>
>>>
http://news.povray.org/povray.newusers/thread/%3Cweb.52d0b932f649b333171348a80%40news.povray.org%3E/
>>>
>>> Best regards,
>>> Michael
>>>
>>
>> It looks like your textures have to much ambient, making reliefs
>> difficult to see.
>> The default value is 0.1, and it's often a good idea to use some smaller
>> values down to zero.
>> If some areas are then to dark, you can use a shadowless fill light,
>> possibly having a different tint.
>>
>>
>> Alain
>
> As all I know ambient is set to zero using radiosity. The pigments are red, blue
> and yellow from colors.inc simply. There are some radiosity artefacts present

In this case, I'd suggest using some actual light_source instead of a 
plain background.
Another solution would be to use a light_dome with some pattern, like a 
simple gradient, so that the radiosity illumination is no longer uniform.

>
> Just for the fun of it I rendered a small animation with "undulated"
> electrodes...
>
> I used Alain's approach modifying only frequency and amplitude of the sinus wave
> for the electrodes. The main trick was to rebuild the electrodes as an
> isosurface as Alain proposed. Using Chris Hormann's ISO-CSG-library it was done
> quickly. For the box object at the ground I used the function
>
>     function { fn1(x,y+sin(clock-sqrt(x*x+z*z)*5)/4,z)}
>
> with
>
>     #declare fn1 = IC_Box(<-20,-1,-10>,<20,1,10>)
>
> from the ISO-CSG library too. I exchanged the radiosity lighting here with a
> small area light since the radiosity artefacts within the animation were
> annoying completely.

For that, you need to adjust the radiosity settings.
Some suggestions:

Reduce pretrace_end to 0.01, 0.005, 0.0025 or even less, as long as it 
still larger than a single pixel.
Try pretrace_end 2/image_width to use 2 pixels tiles on the final 
pretrace step.

Reduce low_error_factor. The default of 0.5 is often to large, especialy 
for animations.

Use the radiosity sample direction pool: count 100 1000. Use a large 
second value, 10 to 100 or more times larger than the base count value.

Slightly increase error_bound, but slightly decrease minimum_reuse.

>
> If someone is interested in the code of this little exercise I will put it here
> under a cc-0 licence.
>
> Best regards,
> Michael
>

Alain

Post a reply to this message