Ember App Amplify (Beta)

Amplify is an image processing technique in Ember App v2.1 that enhances or amplifies small movements in a video for easier visual recognition. This can be used, for example, to highlight and troubleshoot vibrations in a mechanical system.

Amplify works best when the source video has low noise and minimal background movement such that the region of interest where motion is occurring remains in the same place in the video. The frame rate should be approximately 5-20x higher than the highest motion frequency of interest. (In certain cases, frame rate aliasing can also be used to visualize even higher frequencies.)

Amplify requires at least 4GB of VRAM or Apple Shared Memory to operate at full resolution, though 16GB is recommended. Amplify is available on iPad and iPhone at up to 1/2 Resolution. 1TB iPad Pro with 16GB of RAM and compatible Macs are able to run at full resolution. Amplify exports at the selected resolution when added to the Queue.

The following subsections explain the adjustable parameters used by Amplify.

Scale

This determines which features sizes are considered for motion amplification. Small features are at the low end of the scale and large features at the high end. Good results can usually be obtained by leaving at the default settings of 0 - 8. For Height settings below 640, better results may be obtained by setting the high end of the scale to 7 or 6.

Gain

This is the amount of amplification applied to the motion. Settings of around 5 to 10 are usually quite good. Too high a gain can cause noise and unwanted artifacts to appear. A gain of 1 is equivalent to Amplify being turned off.

Frequency

This is the frequency range over which motion amplification is applied. The left slider is a high-pass filter and the right slider is a low-pass filter. The high-pass filter is effective for removing any unwanted bulk movement in the video, for instance caused by slow panning or operator jitter. The low-pass filter can be used to reduce noise and improve the visual appearance of the video.

It is usually best to start with the low-pass filter fully to the right (off) and tune the high-pass filter for best visualisation of the motion. Then the low-pass filter can be used to reduce noise if necessary. Alternatively, if the frequency of the movement under study is known then set the range accordingly to bring out that motion.