Ray tracing

How Does Raytracing Work?

To understand how raytracing works, it is first necessary to understand the basics of intersecting geometry, bounding volume hierarchies, and rays. Then, we will learn about rays' recursive generation, the way they hit a pixel, and the colors they encounter on different surfaces. Raytracing is a popular technology used to create realistic 3-D images. Here are some examples of 3D graphics that use raytracing.

What is raytracing

If you have been interested in the technology behind 3D graphics, you have probably heard about ray tracing. It's been used in the cinematic industry for a while, and a recent example is the recent hit AAA game Control. The technology improves light modeling and is also used in engineering and architecture software. Listed below are some of the most popular uses of ray tracing. Read on to learn more about it!

For years, ray tracing was only available for high-end Nvidia graphics cards, such as the Geforce RTX 2080Ti. This made it very difficult for mainstream gamers to take advantage of it. Even today, however, more games are starting to support both versions of the technology. This means that even if you own a high-end gaming PC, you might still be unable to play games that utilize ray tracing.

The technology behind ray tracing has many uses, but most notable is its use in animated films and games. It can give viewer experiences that look more lifelike, with realistic reflections and shadows. Real-time ray tracing is an advanced form of computer graphics, it's increasingly being used in high-end games. This is especially useful for games with realistic lighting. As the technology continues to advance, more games will be able to utilize ray tracing in real time.

Intersecting geometry

Ray tracing is simply about starting rays from the camera's eye to each pixel on the camera's view plane within the viewport, then those rays are intersected with the scene geometry. The closest intersection is the surface visible from the camera's perspective, the other ihits are behind the first. Instead of projecting all geometry onto the camera's view plane and then ordering according to 'z / depth' (= distance from the eye) and choosing the closest.

Intersecting geometry can be computationally expensive as mathematically challenging. From triangles, the simplest and cheapest form, to curved surfaces like splines for example. Therefore the geometry is often organised into spatial hierarchies so that the number of intersection tests can be minimised with each ray.

Bounding volume hierarchies

The bounding volume hierarchy is a tree structure on a set of geometric objects, wrapped in different bounding volumes. A hierarchy is created by grouping small sets into larger sets, which are themselves enclosed within larger sets. Ultimately, the hierarchy is a tree, with the top-most bounding volume as the root. The tree structure can support several operations on geometric objects, including ray tracing and collision detection.

Shadow ray

To create a realistic shadow effect, a shadow ray is first traced toward each light in the scene starting from an object's intersection point with the camera ray. If the ray reaches a light then that light lits the object's intersection point, otherwise if blocked by another surface then not, and the object's intersection point is in shadow.

Reflection and refraction

Reflection is similar to the shadow ray problem, but the incident ray is mirrored to the normal vector and traced like that. The reflected ray and the incident ray are on the opposite sides of the normal in other words. If the reflected ray hits a surface then that surface's colour contributes to the camera ray intersected object's intersection point. Reflection rays can be bounced further from surface to surface into infinity. Recursive nature.

The recursive nature of ray tracing means that every intersection produces two new rays. Because of this, ray tracing algorithms can become very complex on GPUs. This is one of the primary reasons why GPUs are commonly used for real time ray tracing.

To simulate the reflection effect without the cost of ray tracing through the entire scene, a cubemap of the surrounding scene can be sampled to approximate the reflection. The resulting image is a photo-realistic simulation of light refraction.

Distribution raytracing models a surface by randomly scattering rays, or "distributed" ray traces. These models use a distribution of reflection rays that correspond to a surface's glossy or dullness. Because they are distributed, the model can simulate specular reflection. This method requires that the scene be defined, as well as defining the maximum number of recursions. Obviously, the more recursion points, the more accurate the result. However, this also increases rendering time.

Refraction lets the ray pass through a surface but bends that when entering or leaving. Refraction brings the problem of transparency into the equation, that greatly reduces early ray termination opportunities, and further slows the computation.

Resources (open source)

Back to Hub