Overloud Gem Comp76 FET Compressor and Limiter Plug-In - Manual (Page 53)

OVERLOUD GEMS
PREFERENCES
Each Gem has its own set of preferences to be set to customize its speci fi c behaviour. The avail-
able preference settings can vary upon the speci fi c Gem.
Ask con fi rmation when losing unsaved changes
Ask con fi rmation when overwriting another preset
Show scribbles
These are respectively related to: asking for a con fi rmation when you are about to lose changes
that you haven’t saved yet (i.e. if you load a preset after changing the current one), asking a con-
fi rmation when you are about to replace a preset by saving another one over it. And to show/
hide the scribbles.
Oversampling mode
Oversampling is a technique used to reduce aliasing and to increase processing accuracy. Every
time a digital sound processor introduces an harmonic distortion, a certain amount of aliasing is
generated as a side e
ff ect. In common working situations, where the level of generated harmon-
ics is generally low, there is no need for countermeasures. Hence, the default option, O ff , is al-
most always good, because aliasing audio artefacts are inaudible.
When, on the other hand, high values of gain are used, the aliasing e
ff ect level could raise. For
these cases, selecting Standard the processor performs an oversampling which e ff ectively re-
duces aliasing.
Last setting, Ultra , is for extreme situations, where multiple high gain settings do overlap driving
to a much distorted output. It’s very likely that this situation never occurs in real cases.
The trade-o
ff for oversampling are: CPU load and possible phase rotation on high frequencies.
The higher the oversampling the greater the load on the CPU. And for plugins like the Gems,
planned to be instantiated on a number of audio tracks, this really is something to take care of.
You may see that Oversampling is generally discouraged, especially when the label in the Prefer-
ences panel explicitly discourages its, because the plugin uses DSP techniques to increase the
fi lter accuracy and reduce the aliasing, without the unwanted e ff ects of the oversampling. Some
51