s3g-mc

3OAFX Guides

These guides match the Package Browser’s 3OAFX group. They cover ambisonic offline rendering, ambisonic convolution, spatial grains, and ambisonic spectral profile tools.

Source Format Convention

3OAFX processes use ACN/SN3D for ambisonic media and rendered ambisonic output. When a process accepts both ambisonic and non-ambisonic material, Auto by channel count reads 4ch as 1OA, 10ch as 2OA using the first 9 channels, and 16ch as 3OA. A true 9ch WAV may also be accepted as 2OA. Other channel counts are treated as non-ambisonic source objects. Each input channel is placed onto the selected 3OAFX directional layer, then encoded into the selected ACN/SN3D output order. This is a directional-layer interpretation rather than a decode of a standard speaker format such as 5.1 or hexagonal ring. Use the source-format override when the selected item needs to be interpreted differently.

References and related writings are listed separately in the documentation. The guide pages focus on how to use each process.

3OAFX AED Granulator

Use this to turn mono, stereo, or multichannel source material into a 3OA object-grain field. The selected media item is treated as ordinary source material, not as an ambisonic recording. Each grain reads one source channel, receives its own generated azimuth, elevation, and distance behavior, then is encoded directly to 3OA ACN/SN3D.

This differs from 3OAFX Spatial Grains, which expects ambisonic input and keeps every grain coherent across encoded HOA channels. It also differs from 3OAFX Particle Cloud, which can accept non-ambisonic sources but projects them into the direction layer before cloud processing. 3OAFX AED Granulator keeps the grain as the spatial object.

Method:

Trigger density and chance decide when grains happen. Voice rotation, source spray, pitch spread, reverse chance, source-position quantization, window morph, and drive controls shape the grain stream. AED trajectory modes decide where those grains appear in the ambisonic field.

Main controls:

3OAFX Send Return Controller

Use this controller for the live 3OAFX send/return workflow on a 72-channel track. The track holds a decoded 24-channel virtual speaker lane between an ambisonic decoder and encoder. The controller changes the included JSFX masks and mixer so an inserted 24-channel effect can be focused on a region of the virtual speaker layer.

Track layout:

  1. Ambisonic decoder before the 3OAFX send stage.
  2. JS: s3g 3OA Send.
  3. One 24-channel effect insert.
  4. JS: s3g 3OA Return Mask.
  5. JS: s3g 3OA Mixer.
  6. Ambisonic encoder after the mixer.

Recommended external plugins for this live workflow are SPARTA AmbiDEC before s3g 3OA Send and SPARTA AmbiENC after s3g 3OA Mixer. Set AmbiDEC to MMD and load the included 24-point coordinate JSON files for both the decoder and encoder. The JSON files are included in the package’s sparta_json folder.

SPARTA settings for this chain:

The 72-channel bus is divided into three 24-channel lanes:

The inserted effect processes only 1-24. After adding or moving an insert effect, click Pin inserts 1-24 in the controller so the dry copy and return mask lanes remain reserved for the included 3OAFX JSFX.

Controller setup:

  1. Select the 3OA FX track.
  2. Run 3OAFX Send Return Controller.
  3. Click Load/repair JSFX to add or repair the core chain.
  4. Add the desired 24-channel insert effect between s3g 3OA Send and s3g 3OA Return Mask.
  5. Click Pin inserts 1-24.

Main controls:

3OAFX Object Space

Use this when you want to transform a selected source into an ambisonic object/space relationship. The process accepts either ACN/SN3D ambisonic media or non-ambisonic media. In Auto by channel count, 4ch, 10ch, and 16ch are interpreted as 1OA, 2OA, and 3OA respectively; true 9ch WAVs are also accepted as 2OA. Other channel counts are treated as separate source objects placed onto the selected 3OAFX directional layer before ambisonic encoding.

Modes:

Initial setup:

  1. Select one WAV-backed media item.
  2. Leave Source format on Auto by channel count unless the item is ambiguous.
  3. Choose the target Output order.
  4. Use Object clarity around 0.5, Space amount below 1.0, and Peak normalize enabled while setting the relationship.

For non-ambisonic sources, Source object spread controls how broadly each input channel is encoded into the virtual direction layer before the object-space process. For ambisonic sources, the selected order is decoded to the virtual direction layer, transformed, and re-encoded to ACN/SN3D.

3OAFX Object / Field Split

Use this when you want to separate an ambisonic recording into a foreground object stream and a field-like spatial bed. Select one WAV-backed ACN/SN3D ambisonic item. The renderer decodes it to the 3OAFX directional layer, estimates object-like material from transient energy, directional concentration, and local spectral contrast, then re-encodes the object and field outputs as new ambisonic WAVs.

This version is automatic and does not need a reference file. The process estimates foreground and field behavior from the selected source itself.

Initial setup:

  1. Select one ambisonic WAV-backed media item.
  2. Choose the source order: 1OA / 4ch, 2OA / 9ch, or 3OA / 16ch.
  3. Leave Output on Both object and field.
  4. Use Object bias around 0.55, Transient weight and Directional coherence around 0.45, and Peak normalize enabled.

Raise Transient weight to place attacks in the object output. Raise Directional coherence to keep focused directional energy in the object stream. Raise Field smoothing to broaden the bed and reduce edge-like separation. Object / field crossfade blends the two outputs.

3OAFX Particle Cloud

Use this when you want one or more source files to be reassembled as an ambisonic particle cloud. Sources can be ACN/SN3D ambisonic media or ordinary mono, stereo, and multichannel WAV-backed media. Auto source format treats 4, 9, 10, and 16 channel sources as ambisonic; other channel counts are interpreted as directional objects and encoded through the 3OAFX direction layer.

Method:

The renderer emits grains from the selected sources. Ambisonic sources keep the same source-time position, duration, envelope, playback rate, and yaw transform across every encoded channel. Non-ambisonic sources are first spread across the direction layer, then encoded to the selected ambisonic order before grain rendering.

The diagram shows the selected media feeding a grain cloud and then an ambisonic output field. It updates with the selected order, source count, grain rate, grain duration, intermittency, asynchronicity, and yaw values.

Main controls:

3OAFX Pulsar Field

Use this to synthesize a new ACN/SN3D ambisonic item from pulsar streams. The process does not need a source item. It creates sound from short pulsarets whose repetition rate can move between rhythm, flutter, and tone.

Method:

Each stream emits a train of pulsars. The fundamental curve controls the emission period, the formant curve controls pulsaret width, and the pulse mask decides which pulsars are heard or left silent. Each stream is encoded along an AED trajectory before the result is written as 1OA, 2OA, or 3OA.

The diagram shows pulse trains entering a pulsaret/mask stage and then a moving ambisonic field. It updates with stream count, selected order, and azimuth range.

Main controls:

3OAFX Scene Navigator

Use this when you want several ambisonic recordings to behave like soundfield nodes on a navigable scene surface. Select two or more WAV-backed ACN/SN3D ambisonic media items. Each item becomes a draggable node in the map. Node size is editable: larger nodes keep influence over a wider area, while smaller nodes make tighter zones. The listener path is drawn through those nodes with editable XYZ position and normalized time breakpoints. By default, the listener head faces the direction of travel. A manual AED orientation mode is available when yaw, pitch, and roll are controlled independently from the trajectory. The editor includes a visual preview transport so the trajectory and head direction can be checked before rendering. The renderer writes a new ambisonic WAV representing that traversal.

This is a scene-interpolation and perspective-traversal process. It is not literal physical six-degrees-of-freedom translation inside a single ambisonic recording. Instead, it composes a path through multiple encoded soundfields by decoding each selected file to the same 3OAFX direction layer, weighting nearby nodes, rotating the virtual field from the listener perspective, and re-encoding the result.

Initial setup:

  1. Select two or more same-order ambisonic WAV-backed items.
  2. Leave Source order and Output order at 3OA / 16ch for third-order material.
  3. Drag the blue node spheres to arrange the selected files on the scene map.
  4. Drag the purple/red path points to define the listener trajectory. Leave Head orientation on Face trajectory unless independent head movement is needed.
  5. Use Blend field, Global node radius around 1.25, Perspective rotation around 0.8, and Peak normalize enabled.

The scene viewer has 3/4, Top, and Side camera presets plus camera azimuth/elevation controls. Top places nodes and path points across X/Y. Side edits X/Z for height changes. 3/4 shows the node field and listener trajectory together.

If the render duration is longer than one or more selected media items, those sources are looped under the hood with Source loop crossfade ms. This keeps the listener trajectory independent from file length: each selected soundfield can remain available as a node and becomes audible when the path enters its area.

Global node radius, the selected node’s Node radius, Distance falloff, and Blend sharpness determine how quickly the renderer moves from one soundfield node to another. When the listener trajectory moves outside a node, that node rolls off by distance rather than switching off abruptly. Near-field blur softens transitions close to nodes. Height sensitivity controls how strongly Z-distance affects node weighting. Perspective rotation controls how strongly the listener’s derived or manual head orientation rotates the decoded direction layer. The Time value on each listener breakpoint controls traversal pacing: breakpoints placed close together in time move quickly, while wider spacing slows that part of the path.

3OAFX Spatial Occupation Montage

Use this to create an ambisonic montage from one or more selected WAV-backed media items. The process fragments the selected sources into overlapping events, distributes those events through a virtual direction layer, and re-encodes the result as ACN/SN3D ambisonic output.

The source-format convention is the same as 3OAFX Object Space: 4ch, 10ch, and 16ch are treated as 1OA, 2OA, and 3OA in Auto by channel count, with true 9ch WAVs also accepted as 2OA. Other channel counts are treated as non-ambisonic source objects placed onto the selected 3OAFX directional layer.

The optional stereo expansion is a practical source-expansion step: stereo material contributes left/right object cues plus mid/side-derived front, rear, and side occupation cues before ambisonic encoding. It is not intended as a strict decoder for a historical matrix format.

Initial setup:

  1. Select one or more WAV-backed source items.
  2. Leave Source format on Auto by channel count unless the items need an override.
  3. Choose the target Output order.
  4. Use Events around 180, Spatial occupation around 0.7, and Peak normalize enabled.

Event density controls how many requested events are admitted. Min segment ms and Max segment ms set the fragment size range. Spatial occupation spreads each event through the virtual direction layer, while Spatial motion rotates or offsets event material over short blocks. Dense settings and long segments can build level quickly, so normalization is recommended while exploring.

Stereo Expand to Ambisonic Bed

Use this when you want mono or stereo source material to become an ambisonic bed for later 3OAFX processing. The renderer writes a new ACN/SN3D 1OA, 2OA, or 3OA WAV. Mono sources are treated as a center object. Stereo sources are split into left/right plus mid/side cues, then distributed as front, side, rear, and optional height material before ambisonic encoding.

This is a package-specific expansion tool rather than a decoder for a named matrix format.

Initial setup:

  1. Select one WAV-backed mono or stereo media item.
  2. Choose the target Output order.
  3. Use Balanced bed, Stereo width near 1.0, and Peak normalize enabled.
  4. Raise Rear amount, Side amount, or Height amount to make a broader bed.

Decorrelation adds diffuse support to the derived field. Source spread controls how tightly the derived components land on the virtual direction layer. Bass mono below Hz centers low frequencies before expansion when the output will later be folded down or decoded to compact speaker layouts.

3OAFX Ambisonic Kernel Collage

Use this when the second set of files are not impulse responses, but ambisonic recordings you want to impose on another ambisonic source. The earliest selected WAV is the source. Every later selected WAV becomes a kernel recording. Any number of kernels can be selected. Files use ACN/SN3D; kernels may be 1OA, 2OA, or 3OA when mixed-order adaptation is enabled.

The process decodes the source to a small virtual direction layer, convolves those source feeds with the kernel recordings, and sums the result back to an encoded ambisonic WAV. It is closer to spatial cross-convolution than room simulation: transients in the source can excite the spectral and spatial body of the kernel recordings, while sustained sources can become smeared or clouded by them.

Initial setup:

  1. Select a 1OA, 2OA, or 3OA source item first on the timeline.
  2. Select one or more same-order ambisonic recordings to use as kernels.
  3. Run 3OAFX Ambisonic Kernel Collage.
  4. Use Cycle kernels across directions, Max kernel window sec around 2-4 seconds, and Wet pre-gain dB around -18 for an initial render.

Direction layer chooses the virtual directional structure. Auto by order uses four tetrahedral directions for 1OA and eight practical directions for 2OA/3OA. Sparse 4-direction tetrahedral can also be used with higher-order sources when you want a simpler four-region behavior. Practical 8-direction keeps the eight-region layout regardless of order.

Kernel assignment controls how the kernel recordings are distributed across those directions. Cycle is predictable and can use any number of kernels. Random one per direction changes the mapping with the seed. Kernel index equals direction treats the selected kernels as explicit direction slots, leaving missing slots silent and ignoring extra kernels. Region smear gives every selected kernel an implied position and blends nearby kernels into each virtual direction. Dense all kernels per direction can produce large, saturated spatial masses because every direction is convolved with every kernel.

Adapt mixed-order kernels lets 1OA, 2OA, and 3OA kernel recordings be used together. The selected output order is set by the source/order menu. Higher-order kernels are reduced to that order. Lower-order kernels keep their available channels, and missing higher-order channels are inferred from the lower-order directional energy. This makes mixed-order collage possible, but it is not the same as having native measured material at every order.

The kernel window, fade, wet pre-gain, soft limit, and peak normalize controls shape level and duration when a whole recording is used as a convolution kernel. Shorter kernel windows keep more of the source articulation. Longer windows extend the kernel imprint over time.

3OAFX Image Sonogram Field

Use this to render a PNG image as a spectral-spatial score. The horizontal axis becomes time, the vertical axis becomes frequency, and color data determines AED positioning: hue to azimuth, lightness to elevation, and chroma/saturation to distance. The preview draws a reading graph over the image so the X/time and Y/frequency interpretation is visible before rendering. Use the read-orientation button to transpose the interpretation so vertical image position becomes time and horizontal image position becomes frequency.

Amplitude is separate from lightness so elevation and loudness do not have to be tied to the same visual feature. Choose an amplitude source:

The amplitude preview panel uses the same grayscale convention for each mask: white means higher amplitude and black means little or no amplitude after threshold and amplitude-curve shaping.

When Luminance or Inverse luminance is used as the amplitude source, the elevation source can be changed from image lightness to a uniform elevation or a frequency-spread elevation. This keeps brightness available as amplitude without requiring the same brightness values to determine height.

Composing images for this process:

Synthesis modes:

Granular controls:

Output can be 3OA ACN/SN3D or a multichannel ring. Sphere elevation maps lightness across -90..90; Hemisphere maps lightness across 0..90. Invert distance changes whether chroma moves material nearer or farther away.

Use smaller Time columns, Frequency rows, and Max active rows per column while testing. Higher values admit more image detail and can increase render time.

3OAFX Offline Ambisonic Convolve

Use this for offline convolution of ambisonic source material with ambisonic impulse responses. It belongs with the 3OAFX offline family because it can either convolve one same-order ambisonic file with one same-order ambisonic IR, or use an intermediate directional layer before returning to ambisonic format. In directional-bank mode, the source is decoded or transformed to directional feeds, those feeds are convolved with corresponding ambisonic IRs, and the wet result is summed back into ambisonic format.

Selection:

  1. Select the ambisonic source WAV as the earliest selected item on the timeline.
  2. Select either one same-order ambisonic IR WAV, or a direction-accurate IR bank.
  3. Run 3OAFX Offline Ambisonic Convolve.

The source and IR items use the same ambisonic convention: ACN/SN3D. Choose the source order to match the item: 1OA / 4ch, 2OA / 9ch, or 3OA / 16ch.

Convolution methods:

In directional-bank mode, the IRs are encoded ambisonic WAVs, not P-format files. The source is transformed into the intermediate direction layer, then each directional feed is convolved with an encoded ambisonic IR. The summed output is a new encoded ambisonic WAV in the selected order.

Directional-bank mode does not reuse or wrap IRs. Select either one correctly stacked bank or the exact number of separate IR files required by the method: 4 files for first order, 8 files for 2OA or 3OA. A first-order four-direction bank is not an accurate substitute for an eight-direction 3OA bank. Render time increases with source duration, IR length, ambisonic order, and number of virtual directions.

Adapt lower-order IRs to output order allows a lower-order directional bank to be used in a higher-order render, for example eight 1OA IRs or one 32-channel stacked 1OA bank in a 3OA render. The adaptation estimates direction and signed energy from the lower-order IR, then re-encodes that response to the selected output order. This preserves the lower-order directional information as an inferred higher-order response rather than a measured 3OA IR.

Allow sparse 4-direction FOA bank lets four 1OA directional IRs, or one 16-channel stacked FOA bank, drive a 2OA or 3OA render. This uses the P-format / tetrahedral directions as the source-feed layer. The additional 8-direction higher-order measurement positions are not filled or guessed.

Use Dry level when the IRs were captured with little direct sound and you want to add the original source back separately. Use Wet pre-gain dB to lower the convolution bank before normalization if the wet result builds up.

For designed impulse responses, run 3OAFX Synthetic Ambisonic IR Bank to create encoded ambisonic IRs for the same direction layer. It uses room dimensions, material absorption, scattering, source distance, early reflections, and late diffuse taps to define a synthetic acoustic response.

The designer can write separate ambisonic WAVs, one per virtual direction, or one stacked multichannel bank where each direction occupies a block of ambisonic channels. The convolver detects either format. The practical 2OA and 3OA stacked banks are designed to fit REAPER’s 128-channel track limit. The designer writes a direction-map CSV next to the generated IRs, and the convolver prints the same azimuth/elevation map in the console so measured banks can be checked against the expected order.

3OAFX Synthetic Ambisonic IR Bank

Use this to create encoded ambisonic impulse-response banks for 3OAFX Offline Ambisonic Convolve. The generated files use ACN/SN3D and match the direction layers expected by the convolver.

Output formats:

Direction formats:

The stacked 2OA bank is 72ch. The stacked 3OA bank is 128ch, which fits REAPER’s maximum track channel count. The script also writes a direction-map CSV next to the generated IRs so the direction order can be checked later.

The browser-based Imprint Sketch can preview the same bank groups, response positions, and stacked channel ranges before rendering. Export a project JSON sketch from the browser tool, then use Load Imprint Sketch JSON in this renderer. Imported sketches can set dimensions, material values, source distance, duration, pre-delay, early-reflection count, and ambisonic order. Polygon geometry, field offset, connected regions, open boundaries, and Bank Map positions are also available to the renderer. The offline renderer reinterprets natural and abstract Imprint Sketch spaces, while .s3gimprint export preserves their complete directional event model for s3g Ambi Imprint 64.

Main controls:

3OAFX Offline Renderer

Use this when you want the 3OAFX idea rendered offline directly from an ambisonic media item. Unlike 3OAFX Send Return Controller, this process does not host an external 24-channel insert effect. It uses built-in NumPy effects and bakes the result into a new ambisonic media item.

Selection:

Method:

The process decodes the ambisonic item to a virtual speaker layer, applies one of the included effects over a moving AED focus region, mixes dry and wet behavior using the focus mask, then re-encodes to a new ambisonic item.

Included effects:

Important controls:

For a clear moving-spotlight test, use a strong effect amount, low dry remaining at focus, and an azimuth breakpoint moving from -180 to 180.

3OAFX Spectral Profile Subtract

Use this when you want to reduce or extract spectral material from an ambisonic recording without treating each HOA channel as an unrelated signal.

Selection:

  1. Select a WAV-backed ambisonic source item.
  2. Select a WAV-backed ambisonic profile item that contains the material to reduce or extract.
  3. Run 3OAFX Spectral Profile Subtract.

The source and profile use the same ambisonic order and channel format. The renderer decodes both to the same 3OAFX directional layer, builds a spectral profile per direction, applies subtraction, and re-encodes the result to ACN/SN3D.

Settings:

3OAFX Spectral Profile Tools

These actions share the same source/profile workflow as 3OAFX Spectral Profile Subtract: select a WAV-backed ambisonic source item first, then select a WAV-backed ambisonic profile or reference item. Both items are decoded to the same directional layer before processing and re-encoding.

Use the variants for different intentions:

Useful first controls:

3OAFX Spatial Freeze Trace

Use this to turn a moment or short region of an ambisonic recording into a longer spatial-spectral object. Select one WAV-backed ACN/SN3D ambisonic media item, choose the output order, and render a new ambisonic item.

The process works directly on the encoded channel set. The same STFT frame or trace path is applied to every encoded channel, so the ambisonic channels remain coherent while the spectral content is held, stretched, or ghosted.

Modes:

Main controls:

3OAFX Spatial Grains

3OAFX Spatial Grains applies the same grain micro-control to every encoded component channel. Grain position, duration, envelope, playback rate, overlap, and navigation mode are shared across the 1OA, 2OA, or 3OA channels, so the renderer can work directly on the encoded ambisonic file.

Use Navigation mode to decide how source time is used as a spatial index:

Room memory increases minimum grain length and overlap to help retain reverberant or time-based spatial cues. Yaw controls add optional HOA-domain rotation; order weighting can soften or emphasize higher-order spatial detail.