Original Post
Whats the best way to go about rendering materials with both Transmittance and Reflectance properties? At the moment I have defined my materials to be one of the other, like this
What If i wanted to render plastic? which would have some portion of each? What I want to do is just compute all 3 and then add them up with their particular ratio. But then I would have to have materials that get created like this
material->setDiffuseRatio( 0.25f );
material->setSpecularRatio( 0.25f );
material->setTransparencyRatio( 0.50f, indexOfRefraction );
Is there a better way?
Second, what happens when I perform the tracing of a single ray and it hits multiple perfectly reflectivesurfaces before the number of bounces allowed reaches 0? What should I return? Say I have two mirror balls next to each other and the ray gets into a situation where it just keeps bouncing between the 2. I can solve the infinite recursion by setting the maximum number of bounces to fall out, but what should the pixel colour be?
Third, in my current system, you dont see the light reflected on the objects, because i only compute the diifuse lighting. Am I right in thinking that to add this I have 2 options.
1. Use the phong type specular reflection model and restrict myself to point, spot and directional lights.
2. Use area lighting
Where when using area lighting I have the lighting for a particular point perform ray/surface intersection for each area light, doting the points normal and the light surface point normal. But then how do I compute the amount of light that surface is recieving? The previous compuation doesnt account for the fact that bigger area lights should have more of an effect, ie that the hemisphere above the point being lit sees more of the area light if it is bigger or closer and less if it is smaller and further away. Im not sure how using attenuation would work in this case or if its even valid to do so. Should I instead perform multiple ray samples and then times that by the surface area of the hemisphere? something like
Its likly that if the area light is small then the rays could very well all miss it, producing incorrect results
Trace( ray, bounces )
{
material = intersection(ray);
if( material == SPECULAR )
{
return Trace( reflectededray, --bounces )
}
else if( material == REFRACTIVE )
{
return Trace( refractedray, --bounces )
}
else if( material == DIFFUSE )
{
return calculateDiffuse();
}
}What If i wanted to render plastic? which would have some portion of each? What I want to do is just compute all 3 and then add them up with their particular ratio. But then I would have to have materials that get created like this
material->setDiffuseRatio( 0.25f );
material->setSpecularRatio( 0.25f );
material->setTransparencyRatio( 0.50f, indexOfRefraction );
Is there a better way?
Second, what happens when I perform the tracing of a single ray and it hits multiple perfectly reflectivesurfaces before the number of bounces allowed reaches 0? What should I return? Say I have two mirror balls next to each other and the ray gets into a situation where it just keeps bouncing between the 2. I can solve the infinite recursion by setting the maximum number of bounces to fall out, but what should the pixel colour be?
Third, in my current system, you dont see the light reflected on the objects, because i only compute the diifuse lighting. Am I right in thinking that to add this I have 2 options.
1. Use the phong type specular reflection model and restrict myself to point, spot and directional lights.
2. Use area lighting
Where when using area lighting I have the lighting for a particular point perform ray/surface intersection for each area light, doting the points normal and the light surface point normal. But then how do I compute the amount of light that surface is recieving? The previous compuation doesnt account for the fact that bigger area lights should have more of an effect, ie that the hemisphere above the point being lit sees more of the area light if it is bigger or closer and less if it is smaller and further away. Im not sure how using attenuation would work in this case or if its even valid to do so. Should I instead perform multiple ray samples and then times that by the surface area of the hemisphere? something like
accum = 0.0f;
for( number of sample )
{
ray = computeEquallySpacedRandomRayInHemisphere();
accum += DotRayAreaLight() * (HemispheresurfaceAreaOverNumOfSamples)
}Its likly that if the area light is small then the rays could very well all miss it, producing incorrect results
