Chapter 8: Failing test from earlier

Hey, congrats on getting shadows working! That's great.

As for the "Shading an intersection from the inside" test, it shouldn't be failing, and isShadowed should not be true for that test.

Here's why:

The default world has a single sphere of radius 1 at the origin, and a smaller sphere of radius 0.5 inside it, also at the origin. The default world also declares a point light source that is outside of those spheres.

The test in question should be replacing that point light with a new one that is inside the inner sphere (at (0, 0.25, 0)), The test also positions the viewer at the origin, looking along the z axis.

The test then creates an intersection with the "second object in w" (the inner sphere) at t=0.5. Because the light source and the viewer are both inside that sphere, there should be nothing casting a shadow at that point of intersection.

The culprit could be any of several things. Here are a few to check:

• Your isShadowed check may be returning true if there is any intersection with the shadow ray, even if the intersection lies behind the light source.
• The new light source in that test is not replacing the original one (which was external to the spheres). Or the new light source has been mispositioned.
• Because the intersection is on the inside of the sphere, you need to make sure you're using the correct normal vector (negated). But this is tested on page 95, in "The hit, when an intersection occurs on the inside", so you're probably okay.

Hopefully that gives you some ideas of where to look. If you'd like, I could take a peek too, if you've got your code online somewhere.

I'll admit that I'm only very superficially familiar with scala, so my read-thru may have missed any number of things. But it all looks good to me. The one bit that piqued my skeptical nature was at line 122 of WorldFeature.scala, where you clone the world and assign a new light. Have you verified that w.light is the expected value after that? I suspect it probably is, but like I said, I wondered.

If all of that is in order, the next step would be to walk through your code, step by step, either with a debugger or with copious print statements, to make sure that the values at each step are what you would expect. I'd look especially closely at the values you get in your isShadowed routine: what object is casting the shadow? At what point in world-space is the intersection occurring? Etc.

Sorry I can't be of more help! But I'm happy to answer any questions I can.

Hi Craig,

The is_shadowed should be false for this test. The issue is because you are calculating the over_point in the wrong place. You need to negate the normalv before calculating the over_point. If you move your if isInside... above the the over_point calculation you should be good.

Ian.