Frame interpolation is a technique that creates new video frames between existing ones, making motion appear smoother. If a video was originally recorded at 24 frames per second, frame interpolation can generate additional frames to bring it up to 60, 120, or even higher. The technology shows up everywhere from living room TVs to VR headsets to PC gaming, though it comes with tradeoffs that make it controversial in some contexts.
How Frame Interpolation Works
Every video is a sequence of still images played in rapid succession. Frame interpolation analyzes two consecutive frames, figures out how objects moved between them, and then generates a new “in-between” frame that shows objects at their estimated midpoint position. The result is a video with more frames per second than the original, which the eye perceives as smoother motion.
The traditional approach breaks this into two steps: motion estimation and motion compensation. During motion estimation, the system divides each frame into small blocks of pixels and tracks where those blocks moved from one frame to the next. These movement paths are called motion vectors. During motion compensation, the system uses those vectors to build the new intermediate frame, shifting pixel blocks to their predicted positions halfway between the two real frames.
This method works well for simple, predictable movement. But it struggles with complex scenarios: objects passing in front of each other, sudden lighting changes, or very fast motion. In these cases, the algorithm is essentially guessing, and it can guess wrong.
Traditional Methods vs. AI-Based Approaches
Older frame interpolation systems rely on optical flow, a mathematical technique that calculates the direction and speed of every pixel’s movement between frames. This works as a two-step pipeline: estimate the motion, then synthesize the new pixels. The limitation is that optical flow is notoriously difficult to calculate accurately when scenes get complicated, particularly with large movements, occlusions (one object blocking another), or abrupt brightness changes.
Newer AI-based methods use deep neural networks trained on millions of video frames. Instead of rigidly calculating motion vectors and then building a frame in separate steps, these networks can estimate the output frame directly in a single pass. Some still use optical flow as an internal step but refine it with learned corrections. Others bypass flow estimation entirely, using a single operation to predict where each output pixel should come from. The result is generally fewer visual errors in difficult scenes, though AI interpolation requires significantly more processing power.
Common Visual Artifacts
Because frame interpolation is always predicting what a frame should look like rather than capturing a real moment, errors are inevitable. The most recognizable artifacts include:
- Ghosting: A faint, transparent trail behind a moving object, caused by the algorithm blending two positions together instead of choosing one.
- Haloing: A bright, shimmering aura around objects, especially visible along high-contrast edges during fast motion.
- Blockiness: Visible rectangular edges around moving objects, a telltale sign that the block-based motion estimation couldn’t accurately track the boundary.
These artifacts are most noticeable in complex scenes with lots of overlapping motion or fine detail. A slow pan across a landscape will interpolate cleanly. A fast-moving action sequence with debris, camera shake, and lighting effects will push any interpolation system to its limits.
The Soap Opera Effect
Most movies and many TV dramas are shot at 24 frames per second. That relatively low frame rate is a big part of what gives cinema its characteristic look. When a TV’s motion interpolation bumps that 24 fps content up to 60 or 120 fps, the result often looks oddly smooth and hyper-real, more like a live broadcast or a daytime soap opera than a feature film. This is called the soap opera effect.
The issue isn’t that the interpolated frames are wrong, exactly. It’s that the added smoothness strips away a visual texture audiences associate with high-quality filmmaking. A movie that cost $200 million to produce can end up looking like it was shot on a camcorder. The effect is jarring enough that in 2019, the UHD Alliance worked with TV manufacturers to create Filmmaker Mode, a standardized setting that disables motion smoothing, sharpening, noise reduction, and other post-processing so content displays as the director intended.
If you’ve ever turned on a new TV and thought the picture looked “too smooth” or strangely cheap, motion interpolation was almost certainly enabled by default. Every major TV brand ships with it turned on under a different proprietary name: LG calls it TruMotion, Vizio calls it Smooth Motion Effect, and Panasonic calls it Intelligent Frame Creation. Samsung and Sony have their own branded versions as well. You can typically find the setting in the picture or motion settings menu.
Frame Interpolation in Gaming
In gaming, frame interpolation solves a different problem: getting higher frame rates without the GPU doing the full work of rendering each one. AMD’s Fluid Motion Frames technology, for example, generates intermediate frames between the ones your graphics card actually renders. If your GPU can produce 60 fps natively, frame interpolation can bring the perceived output closer to 120 fps.
AMD supports this feature on its Radeon RX 6000 series and newer GPUs, with the RX 7000 series getting additional capabilities like borderless fullscreen support. A FreeSync-compatible monitor is recommended to keep the interpolated output looking smooth. NVIDIA offers a similar feature called DLSS Frame Generation on its RTX 40 series and newer cards.
The tradeoff in gaming is input latency. Each interpolated frame is generated after the fact, based on frames the GPU already produced. That means the image you see is always slightly behind your actual controller or mouse input. For casual gaming or visually rich single-player titles, the smoother motion is usually worth it. For competitive multiplayer games where reaction time matters at the millisecond level, many players prefer to keep interpolation off.
Frame Interpolation in Virtual Reality
VR headsets have some of the strictest frame rate requirements of any display technology. Dropping below the headset’s target refresh rate doesn’t just look bad; it can cause motion sickness. Meta’s Rift headsets, for instance, target 90 fps. If a game can’t sustain that rate, the system uses a technique called Asynchronous SpaceWarp to fill the gap.
ASW drops the application down to 45 fps and generates every other frame through interpolation, using head tracking data, camera movement, and animation detection from previous frames to predict what the next frame should look like. This nearly halves the processing power required while keeping the visual output close to the full 90 fps experience. It also means VR applications can run on lower-performance hardware than they’d otherwise need, which has been important for making standalone headsets viable.
Where It Helps and Where It Doesn’t
Frame interpolation genuinely improves the viewing experience for certain types of content. Sports broadcasts, nature documentaries, and video games all benefit from smoother motion, especially on high-refresh-rate displays. Fast camera pans and tracking shots look cleaner with less judder. For live content that was already shot at higher frame rates, interpolation up to a display’s native refresh rate introduces minimal artifacts.
Where it falls apart is cinematic content and scenes with complex, unpredictable motion. If you primarily watch movies and scripted TV shows, turning off your TV’s motion smoothing (or enabling Filmmaker Mode) will give you a picture that looks closer to what the creators intended. If you watch a lot of sports or play games, it may be worth experimenting with the setting at different intensity levels, since most TVs let you adjust interpolation strength rather than simply toggling it on or off.