Concepts

Your video is a series of frames, usually 60 or more per second. Interpolation estimates the motion between two frames, which makes it possible to add more frames in between.

For example, this clip was recorded at 240 fps. At 960 fps, three interpolated frames come between each pair of recorded frames.

The more frames you record, the less the picture changes from one frame to the next, and the more accurate the frames in between are.

With that many frames, you can blend them. Blending averages nearby frames into one, which gives the motion blur that a camera makes with a long shutter.

At the defaults, each 60 fps output frame is the average of 17 interpolated frames, and Intensity sets how many frames blend.

The estimate is not always right. When an object moves fast or changes shape between two frames, interpolation can tear it apart or leave holes in it. These errors are called artifacts.

Here, a 60 fps copy of a clip was interpolated to 240 fps, so you can compare the result with the frame that was recorded at the same moment. SVP tears the sword apart, and RIFE leaves small holes in it.

The first thing to try is recording at a higher frame rate, since more recorded frames make the frames in between more accurate. The rife engine is usually more accurate than svp too, but it takes longer to render. With svp, Speed and pre-interpolation also change how many artifacts you see.

interpolini also has three masking methods that protect fixed parts of the picture, such as the HUD, from these artifacts: peel, still and image.