What is real-time rendering?
Real-time rendering is the process of generating images from a 3D scene quickly enough that frames are produced and displayed as the viewer interacts with the scene. Frame generation typically targets rates such as 30, 60, 90, or 120 frames per second, with each frame produced within a few milliseconds so motion appears continuous and input feels responsive.
Unlike offline rendering, where a single frame may take minutes or hours to compute, real-time rendering operates under a strict per-frame time budget. This constraint shapes the techniques used, generally favoring rasterization, approximations of lighting and shadows, and increasingly hybrid approaches that combine rasterization with selective ray tracing.
How real-time rendering works
Most real-time rendering follows a pipeline that transforms 3D scene data into a 2D image on screen. The stages below describe a typical rasterization pipeline as implemented on modern GPUs.
- Scene setup. The application submits geometry, materials, textures, lights, and camera parameters to the graphics API. Visibility culling and level-of-detail selection often happen here to reduce the work passed to the GPU.
- Vertex processing. Vertex shaders transform 3D vertex positions into screen space, applying model, view, and projection matrices. Per-vertex attributes such as normals and texture coordinates are prepared for interpolation.
- Primitive assembly and rasterization. Transformed vertices are grouped into primitives (usually triangles) and converted into fragments, one per covered pixel. Depth and coverage information is computed for later visibility tests.
- Fragment shading. Fragment (or pixel) shaders compute color for each fragment, sampling textures and evaluating lighting models. Shadow maps, environment maps, and other precomputed inputs may be referenced here.
- Output merging. Depth testing, stencil testing, and blending combine fragment outputs into the framebuffer. Post-processing passes such as tone mapping, anti-aliasing, and bloom may then refine the image.
- Presentation. The completed frame is handed to the display system and shown on screen, often synchronized with the display refresh to reduce tearing.
Common techniques and trade-offs
Real-time rendering relies on a mix of techniques chosen to fit the available time budget and hardware. The items below describe approaches that tend to appear together in modern engines.
- Rasterization. Converts triangles directly into pixels and remains the dominant approach for interactive workloads due to its predictable performance characteristics.
- Ray tracing and hybrid pipelines. Selective ray tracing can be used for reflections, shadows, or global illumination while rasterization handles primary visibility, balancing image quality against frame time.
- Shadow mapping and screen-space effects. Approximations such as shadow maps, screen-space ambient occlusion, and screen-space reflections trade some accuracy for speed.
- Upscaling and temporal techniques. Methods like temporal anti-aliasing and learned upscalers render at a lower internal resolution and reconstruct a higher-resolution image, freeing time for other work.
- Level of detail and culling. Distant or occluded geometry is simplified or skipped so the renderer focuses effort on what the viewer is likely to see.
Where real-time rendering is used
Real-time rendering underpins interactive 3D experiences across many domains. Video games are the most visible example, but the same techniques drive simulators, virtual and augmented reality systems, architectural and product visualization, digital twins, virtual production, and live broadcast graphics. In immersive video workflows, real-time rendering may also be used to play back volumetric or light field content, where the rendered view is reconstructed on the fly from captured data rather than authored 3D assets.
See also
Rasterization — the dominant primitive-to-pixel technique used in real-time pipelines.
Ray tracing — a physically motivated rendering approach increasingly used alongside rasterization in hybrid pipelines.
Graphics pipeline — the sequence of stages that transforms scene data into displayed pixels.
Shader — programs that run on the GPU to compute vertex transforms, fragment colors, and other per-element results.
Volumetric video — captured 3D content that can be played back through a real-time renderer for interactive viewing.
Additional resources
- Real-Time Rendering — companion site to the standard reference textbook on the field, with curated links and resource lists.
- NVIDIA RTX Ray Tracing — developer documentation describing hardware-accelerated ray tracing in real-time pipelines.
- Unreal Engine rendering features — engine-level documentation covering lighting, materials, and post-processing in a production real-time renderer.
- Vulkan Specification — a low-level graphics API used to build real-time renderers across platforms.
