We often specify different matrices for different objects in our scene. When we use a ray tracer, we may like to use the absolute value of \( t \) as distance if our ray was formulated as \( \vec{r} = \vec{r}_0 + t\vec{r}_d \). In this case, should we normalize \( \vec{r}_d \)？ If so, where should we do the normalization?(Note that \( M \) is the model transformation for a specific object)
A. Do not normalize.
B. Normalize after applying \( M^{-1} \).
C. Normalize after we find the world-space direction, and before we apply \( M^{-1} \).
D. Normalize after we find the world-space direction, and again after we apply \( M^{-1} \).
查看答案
In OpenGL, we could not see objects in front of our \( z_{near} \) plane, and objects behind our \( z_{far} \) plane. What objects can we see, with respect to the virtual screen in ray tracing?
A. We can only see objects behind our virtual screen.
B. We can see objects on either side of our virtual screen.
C. We can only see objects in front of our virtual screen.
Can we define objects in a ray tracer without vertices?
A. Can we define objects in a ray tracer without vertices?
B. No, it does not make any sense to define an object without vertices.
C. No, if we cannot render the object using OpenGL, we cannot render the object in a ray tracer.
D. Yes, but we must implicitly evaluate a function and generate vertices.
Can we use our recursive shadow algorithm in a rasterizer, such as OpenGL?
A. Yes, but it will be as slow as a ray tracing framework.
B. No, this algorithm cannot be parallelized.
C. No, it is impossible to write a similar algorithm using shaders.
D. Yes, and at a faster rate than ray tracing
How do we choose which object to show, given a specific pixel?
A. Z-buffer, as in OpenGL.
B. We sort the objects by their Z value.
C. We shoot rays into the scene, and average the contribution of each.
D. We use the closest intersection of the ray shot into the scene.
微信支付 
支付宝支付