Pixel smoothing algorithm for upscaling or image transformation — averages four neighboring pixels for intermediate values. Faster than cubic, fewer artifacts than nearest neighbor.
When upscaling footage or performing image transformations in compositing, you need an algorithm that can invent missing pixels. Bilinear interpolation works by considering the four immediate neighbors of a calculated pixel—top-left, top-right, bottom-left, and bottom-right. It weights their color values based on position and averages them to a new value. Visually, the result sits between the harsh aliasing of Nearest Neighbor and the more computationally intensive cubic interpolation.
In a practical workflow, you'll notice the difference immediately. If you need to upscale 2K material to 4K or transform a compositing element in 3D space, you'll usually set bilinear as the default in Nuke or After Effects. The reason: speed. While cubic methods consider 16 or more pixels and cost you time on large projects, bilinear calculates significantly faster and in most cases delivers a clean image without visible interpolation artifacts. For moderate transformations—scaling an element by 20-30%, slight rotations—you won't see a visual difference compared to cubic, but your timeline will remain responsive.
The catch lies in extreme upscales. If you upscale 720p to 2K or higher, you'll lose information sooner or later—bilinear will compensate for this with slightly soft edges. Here, cubic or even better methods (like Lanczos) would be beneficial if you can afford the render time. Another critical point: when working with motion blur or complex distortions, bilinear can lead to artifacts at the edges of fine details, which can be difficult to repair later.
On set itself, you don't need to worry about this—the decision is made in post-production. However, the VFX supervisor should know early on: will the material be radically scaled, or will it remain within the normal range? This determines whether bilinear is sufficient or if you need to plan for better interpolation methods (see also Cubic Interpolation, Sampling Methods in Compositing). In real-time engines and live compositing, bilinear becomes the standard choice because it maintains the optimal balance between quality and performance.