Streaming Engine section

RAM buffer

This value, in sample frames, is the size of the portion of each sound held in RAM to enable low latency playback, circumventing the inherent latency involved with disk seek times. The portion held in RAM plays while the streaming engine cues up the rest of the data from the hard disk. A larger value gives the hard drive longer to deliver the data, but is more demanding on RAM. If the setting is too low for your system, you are likely to suffer from dropouts and other audio artifacts.

This value determines the length of each sound played when the Preview RAM audio only option is enabled.

Settings of 4096, 8192, 16384, 32768 and 65536 samples are possible, selectable via a drop-down menu. A setting of 16384 or 32768 is fine for most systems, while fast newer machines may be able to run with settings of 4096 or 8192.


Stream buffer

This is the size, in sample frames, of the buffers of data being streamed into RAM from the hard disk for each voice. Generally, hard drives are more efficient at reading fewer large chunks of data than many smaller chunks. However, a larger Stream buffer uses more RAM, and may be inefficient if not all the data is used, such as when notes are choked.

Settings of 4096, 8192, 16384, 32768 and 65536 samples are possible, selectable via a drop-down menu. Settings of 8192, 16384 or 32768 are fine for most systems.


Load all to RAM

With this setting activated, the entire kit is loaded to RAM and no data is streamed from the disk during playback.

Please ensure you have enough RAM before using this option, and that you are running BFD3 within a suitable environment with the other engine preferences set to suitable values. Unless your system is fully 64-bit (64-bit plugin within a 64-bit host in a 64-bit OS) with a high amount of RAM, it is unlikely that you will be able to load full detail kits with Load all to RAM activated. In 32-bit environments, 16-bit mode should be activated and the Max velocity layers setting reduced to a much smaller value.


16 bit mode

In this mode, sample data is loaded into RAM and streamed from disk at 16 bit instead of 24 bit. Because 24 bit data is stored in memory as 32 bit floats, using 16 bit mode effectively halves the memory footprint required by BFD3.


Max voices

The Max voices setting dictates the maximum number of voices the BFD3 engine can play simultaneously. If the voice limit is exceeded, an intelligent voice-stealing system is applied, based on the oldest voice which is still playing.

The number of voices required for a performance can be much larger than anticipated. For example, decaying cymbals and toms can raise polyphony requirements very significantly. 64 is a safe number to use for this setting - voices do not consume a large amount of RAM when unused.


Max velocity layers

This setting limits the number of velocity layers used for each Drum articulation, thereby reducing the strain on the hard disk and RAM but at the expense of detail.

If a Drum is loaded that has more layers than the number specified with this setting, only selected velocity layers at proportional intervals over the range are loaded. The Drum's natural timbral variations are still heard but with less 'resolution' over the velocity range.

Smaller values can be useful as an efficient preview mode while composing, especially with other RAM-intensive plugins and instruments in your host/DAW project.

Simply increase the value and restart the engine before performing a final mixdown with full quality.

To change the setting, click the field and enter a new value between 1 and 256.

Note that the Detail section in the Engine Preferences allows further Detail settings for each Drum class -  these operate as proportional reductions of the Max velocity layers setting.


Max cache buffers

This setting represents the maximum number of disk streaming buffers cached in RAM. If the same velocity layer is triggered repeatedly, it is possible to avoid reading from the disk by reusing the disk streaming buffers with the same data. This parameter determines how much RAM to devote to storing layers to play again.

With any kind of dynamics in a performance or when using the Anti-MachineGun functions, layers do not tend to be reused very often so this setting can be reduced if you wish. Each voice does still require 1 cache buffer.


Preview RAM audio only

This setting forces the engine to play only the initial part of the sounds which are held in RAM without streaming the remaining parts from disk.

This mode may be useful while composing in order to avoid disk usage, and reactivated during mixdown. The actual amount of each sound stored in RAM is determined by the RAM Buffer setting.


Restart engine

If any of the following settings are changed, it is necessary to use the Restart engine button in order to make them active:

• Load all to RAM

• 16 bit mode

• RAM buffer

• Stream buffer

• Max cache buffers

• Max velocity layers

If any Drums are currently loaded when this function is used, they are automatically reloaded after the engine is reinitialized with the new settings.


Enable BFD3 extended channels

When this setting is deactivated, only the OH, Room and Amb3 ambient channels are active - extra Mono and Comp channels are ignored, meaning that kits use much less RAM.

When the setting is activated, all channels possessed by a Drum are loaded.


Synth Engine section

Drummer perspective

This preference provides the default setting for the Audience/Drummer switch in the Kit display.


Disable SideStick tuning

With this setting activated, the SideStick articulation is not affected by any tuning changes for the Snare. This results in behaviour that more closely resembles how a real snare works.


Anti-MachineGun mode

This preference provides the default setting for the AMG button in the BFD3 Dashboard.


Retrigger threshold

This setting defines a period of time after each received MIDI note within which further received notes are ignored. This setting is useful for trigger systems such as electronic kits that are prone to double triggering.

The value is set in seconds. The default value is 0.050 (50 milliseconds).


Fades section

A choke fade, or fade, occurs if an articulation in a Drum is triggered before an older articulation from the same Drum has finished decaying.

For example, when playing a high tom twice rapidly in succession, the first is faded out when the second is triggered. The fade settings allow you to adjust the duration of these fades that occur when using choking.

Fade times for each Drum slot are available within the Choke Response section of the Drum Editor (on the Model page).

The fade settings in the Preferences represent default settings to use in an empty launch state. Hihat, Tom and Cymbal slots possess their own fade defaults in this section - the Default choke fade (base) and Default choke fade (range) settings apply to all other Drum types.

How fades work

There are two components to choke fade times: fade (base) is the minimum fade time, while fade (range) is the maximum additional time added to the (base) value, according to BFD3's 'dominant excitation preservation' algorithm. This algorithm allows longer fade times for louder events (higher velocities) when choked by quieter events (lower velocities) and reduces the fade time when a low velocity event is choked by a subsequent higher velocity articulation. A quiet event choked by a loud event has a fade time of fade (base), whereas a loud event choked by a quiet event has a fade time of fade (base) + fade (range). This approach achieves realistic results while reducing the disk streaming load when possible.

Please note that adjusting fade settings to extreme values can achieve results which may sound unrealistic. Subtle use is advised if realistic results are desired. Extreme settings are, however, allowed as an aid to experimentation.

Additional hihat fades

Hihats feature 2 additional fade times: Hihat closed tip choke fade and Hihat pedal choke fade.

The pedal is the only thing that chokes an open hihat when playing a drumkit. When playing BFD3's sounds from an electronic drumkit, the pedal note transmitted when the pedal is moved down fully is used to choke any playing open hihat articulation. It has also become conventional for sample-based drumkits to choke open hats with a closed hat note.

Both of these settings should be set to low values, with the other fade times for the Hihat (which cover all open positions and the Closed Shank articulation) set to higher values.

Additional snare fade

Snares feature an additional choke articulation and associated fade time for choking Drag articulations, allowing you to shorten drag roll articulations. The fade time for this type of choke is set with the Snare drag choke fade setting. This setting is also available in the Drum Editor via the Special setting in the Choke Response section.


Detail section

The Detail settings specify the level of velocity layer detail for each Drum Class: Kicks, Snares, Hihats, Toms, Cymbals and Percussion.

These settings operate relative to the Max velocity layers setting, with proportional layer selection throughout the maximum range occurring in the same way.

Four detail level options are available for each Drum class:


A single velocity layer is loaded for each articulation in the Drum class.


Approximately 25% of the number of layers specified by the Max velocity layers setting are loaded.


Approximately 50% of the number of layers specified by the Max velocity layers setting are loaded.


The maximum possible number of layers as specified by the Max velocity layers setting are loaded.


Simulation section

Kick to Tom Spill
Snare to Tom Spill
Kick to Tom Resonance
Snare to Tom Resonance

These settings provide overall gain scaling levels for the Drum Editor's Res Trim and Spill Trim settings for the Kick and Tom slots.