Dimmer Director Filter
This document describes how to use the dimmer director filter.
How it works
The filter takes n groups of X/Y/Z dimmer channels as input. The channels of a
group are addressed by spatial coordinates: a group might, for example, hold
3×2×6 channels, with a single channel addressed as groupname_X_Y_Z. The output
is not a scalar but a three-dimensional vector.
For each group — configured globally, but applied per group — the filter accepts a configurable number of dimmers, which will be mixed as well as offset. A global mixin or offset is applied to all channels; a local one is specific to a single channel, and if local ones are configured, they exist for each input channel independently.
Signal processing
The filter operates on floats; the signal flow is shown in the figure. The group inputs — the values put into the filter — are normalized to [0, 1] first, then mixed (multiplied). Next, offsets are applied (added) and the results are clamped to [0, 1]. In a final stage, the effect engine is applied, creating the outputs from these values.
Effect engine
For each group, the effect engine generates the output by multiplying the channel value with the selected effect curve (a float keyframe between -1 and 1). The effect is applied to each channel atomically, with the time argument shifted by that channel’s X/Y/Z phase. Effect curves can be configured to be repeated or to be started based on a value change above a configurable threshold (default: 10%); the threshold is measured against a window of previous values. A currently running triggered effect curve is not restarted until the input value has stopped changing — the curve finishes first and then holds its last value (it always uses the current input value, of course). The result is clamped to [0, 1]. Finally, the dimmer curve function (for example Lin, Sig, or a custom Bézier) is applied for each output.
