Lecture 9: Intro to Ray-Tracing & Accelerating Ray-Scene Intersection (17)
JerryIshihara
In summary, ray tracing with triangle mesh is a technique that involves shooting a ray of light from the camera into the scene and calculating how it interacts with the objects it comes into contact with. To render a triangle mesh, the ray is compared to each triangle in the mesh to see if it intersects with it. If an intersection is found, the triangle's color and lighting are calculated and added to the final image.
I can tell that ray tracing with triangle mesh is a computationally intensive process that necessitates a significant amount of processing power in order to render complex scenes. It can, however, produce highly realistic images with precise lighting and reflections, making it a popular technique in the film and gaming industries.
Staffethanweber
Nice summary! A slight note at the beginning where you say "shooting a ray of light from the camera into the scene", I guess it's not quite a ray of light shooting out but rather the reverse, where we are trying to calculate the light going into that pixel. :)
ld184
Is it somehow possible to express this problem in terms of linear algebra? it seems like we have a surface (span of the triangles), and we have a line, which can be described as a null space of some matrix. This seems like an optimization problem somehow...
jacklishufan
How do we handle edge cases of Tangent Intersection (Smooth surface)? In particular, if a light ray intersects with a sphere, there would be one and only one intersection. So all points on the tangent line with distance further than the tangent intersection will be indistinguishible from points inside the object since they both have 1 intersections?
JustinHuey1
For ray tracing, does rendering the triangles color and lighting basically use barycentric coordinates like we learned or is there a different method that does this?
jonathanlu31
@jacklishufan my guess would be that if the point really was on the inside of the mesh, then there would technically be two/even # of intersections, one for positive time/direction and one for negative time. If the ray intersected at a tangent, then there would only be one/odd # of intersections for all t (-inf, +inf).
In summary, ray tracing with triangle mesh is a technique that involves shooting a ray of light from the camera into the scene and calculating how it interacts with the objects it comes into contact with. To render a triangle mesh, the ray is compared to each triangle in the mesh to see if it intersects with it. If an intersection is found, the triangle's color and lighting are calculated and added to the final image.
I can tell that ray tracing with triangle mesh is a computationally intensive process that necessitates a significant amount of processing power in order to render complex scenes. It can, however, produce highly realistic images with precise lighting and reflections, making it a popular technique in the film and gaming industries.
Nice summary! A slight note at the beginning where you say "shooting a ray of light from the camera into the scene", I guess it's not quite a ray of light shooting out but rather the reverse, where we are trying to calculate the light going into that pixel. :)
Is it somehow possible to express this problem in terms of linear algebra? it seems like we have a surface (span of the triangles), and we have a line, which can be described as a null space of some matrix. This seems like an optimization problem somehow...
How do we handle edge cases of Tangent Intersection (Smooth surface)? In particular, if a light ray intersects with a sphere, there would be one and only one intersection. So all points on the tangent line with distance further than the tangent intersection will be indistinguishible from points inside the object since they both have 1 intersections?
For ray tracing, does rendering the triangles color and lighting basically use barycentric coordinates like we learned or is there a different method that does this?
@jacklishufan my guess would be that if the point really was on the inside of the mesh, then there would technically be two/even # of intersections, one for positive time/direction and one for negative time. If the ray intersected at a tangent, then there would only be one/odd # of intersections for all t (-inf, +inf).