Case Study — Capturing for Immersive

In brief

Immersion is decided at the render, not the microphone — you do not need an "immersive" rig.
A capture delivers either a coherent field (the space) or clean sources (to place later).
In an object engine, any microphone — even a mono one — is a spatial object.
Directivity is a distance choice; reaching spot mics must be time-aligned or they comb.

The question

Ask a room of engineers "how do I record in immersive?" and most will answer with hardware: a 3D microphone array, a tetrahedral Ambisonic mic, a Hamasaki square with height. Those are excellent tools, and the Recording & Capture part covers them in depth. But the premise deserves challenging, because it quietly confuses two different things — capturing and reproducing.

Immersion is decided at the render, not at the microphone. What a capture actually has to deliver is one of two things: a coherent field (the space itself, encoded in the inter-channel relationships between mics), or clean sources (individual sounds you will place later). Whether the result becomes stereo, 7.1.4 or a binaural stream is a rendering decision made downstream. In an object-based engine such as RIPL, the consequence is liberating: any microphone — even a single mono one — is a spatial object, a source with a position you assign, not a channel wired to a speaker. You do not need an "immersive" rig to work immersively; you need clean material and a renderer that treats space as something authored, not captured-and-frozen.

This chapter is about that reframing, and about the choices it opens up.

What a capture can deliver — and the two philosophies

Immersive capture splits along the same line as immersive delivery (see Capturing for Object/Scene Workflows and formats):

Capture the field. A coherent main array (Decca-tree-plus-height, ORTF-3D, a Hamasaki square) or an Ambisonic microphone samples the space — its width, depth, height and reverberation — as one consistent perspective. What the array captures is essentially what plays back. This is fidelity: the room, transported.
Capture the sources. Spot, close and accent microphones isolate individual sounds, each later authored as an object with a position (and elevation) you choose. The space is then reconstructed at the render rather than captured. This is freedom: the perspective is decided in the mix, not on the day.

Most real sessions do both — a coherent array for the bed and envelopment, spot mics for the sources — precisely so you can decide later how much of the space is captured and how much is authored (the guide's capture-once-decide-later principle). The object paradigm simply pushes that balance as far as it will go: capture clean sources, build the space downstream.

Different uses of capture

"Recording for immersive" is not one activity. At least four distinct uses coexist, often in the same session:

Capture a space. You want the place itself — a hall, a church, a forest, a room tone. A coherent array or an Ambisonic mic is ideal, but so is a good stereo pair, or even a single well-placed omni in a room with character: as an object with a captured room around it, it can carry a convincing sense of place. The point is the space, and modest means can capture it.
Go and get a source at the stage. You want a specific sound out of a crowded stage — one singer, one instrument, one effect — cleanly enough to place and move it. This is the world of spot, close and accent mics, and of directional reach: a more directional microphone effectively lets you sit farther from a source for the same direct-to-reverberant ratio (a cardioid pair reaches roughly √3 ≈ 1.7× farther than omnis — see the two-microphone reach of directivity). Reaching into the stage is how you get objects worth placing.
Capture an ensemble on a stage. A whole band or ensemble at once: an array of spot mics plus overheads/room, each close mic an object, the room capture the bed. The rock case study is exactly this.
Capture for movement and re-spacing. When the intent is to move a sound or re-place it per venue, you deliberately capture it isolated — because only an isolated source can become a freely-positioned object. Here the mic technique is chosen for separation, not for a captured image.

Arrays and reaching the stage — technical and artistic consequences

"Going to get the source at the stage" with directional mics and spot arrays has consequences that run in both directions.

Technical consequences.

Directivity is also a distance choice. Choosing a more directional pattern to reach farther simultaneously sets the direct-to-reverberant ratio and therefore the captured distance — the two cannot be decoupled (principles of spatial capture). And you cannot fix distance later with a fader: the D/R ratio is baked in at the moment of capture by where the mic sits.
Bleed and time-incoherence. Spot mics on a shared stage are at different distances from every source, so they are time-incoherent with each other and with the main array. Summed naively they comb-filter and smear the image — the first cancellation notch sits at f = c ⁄ (2·Δd) for a path difference Δd (mono compatibility). The fix is to time-align each spot to the array (delay by Δt = distance ⁄ c) and use it sparingly.
Bleed as glue, not only as enemy. On an ensemble, leakage between mics is often the natural cohesion of the room; managed rather than eliminated, it helps the stage read as one place.

Artistic consequences.

How you mic decides how free you are later. The more you capture isolated sources, the more you can place, move and re-space them — you are choosing construction. The more you capture the field, the more fidelity you get but the more the perspective is fixed at capture. Microphone technique is therefore already a compositional decision about how much spatial freedom you want downstream, not merely a technical one.
Reaching changes intimacy. A close, reaching mic gives a present, high-D/R source with little room — intimate and placeable — where a distant array gives the source in its space. Choosing between them is choosing the character of the sound, not just its cleanliness.

Questioning the received paradigms

Seen from an object-based renderer, the two classic immersive-capture paradigms each carry assumptions worth questioning — not because they are wrong, but because they bind the result to decisions made at capture time.

The main-array / channel-bed paradigm (Theile & Wittek; Williams; Hamasaki) maps microphones directly to a loudspeaker layout. Its strengths are real — physically-correct inter-channel relationships, one coherent perspective — but the guide's own account of it is candid about the cost: you cannot rebalance one instrument without moving the whole image, the direct-to-reverberant ratio is fixed by mic placement, and the array fixes the perspective at one listening point (immersive & 3D recording). It also tends to require many microphones in precise geometry, and ties the capture to a specific delivery format.
The scene-based / Ambisonic paradigm (Gerzon; Zotter & Frank, 2019) captures the whole field at one point, which is elegant and rotatable — but assumes a single observation point and, at practical orders, offers limited spatial resolution and a real sweet spot. The perspective is again fixed at capture.

The common thread: both decide the space at the microphone. An object-based approach inverts the order — capture material, decide the space at the render — which is exactly what makes "any mic is an object" more than a slogan.

Parametric and object approaches decouple capture from reproduction

A substantial body of work formalises this inversion, analysing a capture into sound plus spatial parameters — or objects plus metadata — that are re-rendered to whatever system is present, rather than fixing the image at capture. A review tutorial makes the split explicit in its very title, treating scene acquisition, modification and reproduction as three decoupled stages (Kowalczyk et al., 2015).

Directional Audio Coding (DirAC) estimates, at a single point and per time–frequency tile, the direction of arrival and the diffuseness of the field, splits the signal into non-diffuse and diffuse parts, and re-synthesises for whatever loudspeaker layout is present — analysis fixed, rendering adapted to the system (Pulkki, 2007).
Object and scene coding — MPEG Spatial Audio Object Coding (SAOC) — transmits several objects at roughly the bit-rate of a two-channel signal and lets the decoder render the scene interactively, onto a target loudspeaker configuration or a portable device (Herre et al., 2012). The scene is built at the far end, not fixed at capture.
Virtual Microphone Control (ViMiC) makes the microphone itself a render parameter: it synthesises the signals of virtual microphones with chosen positions and directivities, so "the perspective" is chosen at reproduction rather than set by a physical placement (Braasch, Peters & Valente, 2008; also in amplitude panning).

Two results bear directly on the number and placement of microphones. Parametric methods have been shown to record perceptually from a small microphone array and reproduce to arbitrary loudspeaker configurations (Politis, Vilkamo & Pulkki, 2015) — the clearest published support for "fewer microphones, less critical positions, layout decided downstream." The field-encoding idea itself runs back to Gerzon's founding Ambisonic proposition (1973), that a sound field can be encoded independently of the speakers that reproduce it, and to wave-field synthesis on the reproduction side (Berkhout, de Vries & Vogel, 1993). What object-based capture adds is to push that independence down to the individual source: capture clean material, and let the render decide the space.

An honest boundary

These references establish two sourced claims: that parametric and object approaches decouple capture from reproduction, and that small arrays can feed arbitrary layouts. They do not all support a blanket "use fewer mics." ViMiC concerns virtual microphones synthesised at the render; Gerzon and wave-field synthesis are about field encoding and synthesis, not economical capture. The strong, sourced statement is the decoupling; the specific "fewer mics, looser positions" benefit is demonstrated by parametric small-array methods (Politis et al., 2015) and is the design position RIPL takes, below.

The thesis: any microphone is an object

In RIPL's unified source model, the format of what you captured stops being a routing problem and becomes a source type:

a mono spot or room mic → an object with a position (and elevation) you assign;
a stereo pair → a decoded field (HSR) whose primaries and ambience are separated before placement, not two mono channels to scatter;
an Ambisonic mic → a rotatable scene;
a native multichannel stem → a field carried as itself.

All four are first-class sources in one scene, rendered together to any layout. Immersive capture, in this light, is not a gear tier you must buy into; it is a way of treating whatever you recorded.

Fewer microphones, looser positions. Because RIPL reconstructs the space at the render rather than reading it off a fixed channel map, a microphone's position informs an object instead of hard-wiring a channel. RIPL's array approach leans on exactly this: you can go further with fewer microphones and less strict positioning than a channel-bed array demands, because the exact geometry no longer has to be the playback format — it feeds a model that is re-rendered to whatever system is present. A pragmatic capture of clean sources plus a sense of the room can become a full immersive scene, where the classical channel-bed approach would have required a larger, more precisely-placed array to capture the same result directly.

(This is RIPL's design position, and the general entailment of object-based capture — capture material, author space downstream. The underlying principle has published support in parametric spatial-audio research, where small microphone arrays are shown to feed arbitrary loudspeaker layouts (Politis, Vilkamo & Pulkki, 2015); the specific RIPL claims are offered as the tool's approach, distinct from that sourced work.)

Solutions — with RIPL

Treat every input as a source, not a channel. Mono spots, stereo pairs, Ambisonic mics and multichannel stems all enter the one model and are placed, decoded or carried as appropriate — then rendered to Atmos, ambisonics, WFS or binaural from one scene.
Alignment is placement. Because RIPL renders with gain and delay, time-aligning a reaching spot to the room capture — the cure for comb filtering — is part of positioning it, not a separate patch.
Decide the space in the mix. With sources captured clean, perspective, distance and width are authored downstream, and re-authored per venue, instead of being frozen on the day.
Fewer mics, re-rendered anywhere. The captured geometry feeds a model, so a lean capture becomes a scene that renders to whatever system the piece meets next.

Pitfalls and checklist

Checklist

Are you buying "immersive" hardware when clean sources plus a renderer would do?
Did you decide, consciously, how much to capture the field (fidelity, fixed perspective) versus capture sources (freedom to re-space)?
Are reaching spot mics time-aligned to the array, or are they combing the sum?
Did you get the array distance right on the day (D/R is baked in — no fixing it with faders)?
If you intend to move a sound, did you capture it isolated enough to become a real object?
Does your capture survive being re-rendered to a different layout, or is it wedded to one channel bed?

Bibliography

Theile, Günther, and Helmut Wittek. "Principles in Surround Recordings with Height." AES 130th Convention, London, 2011, preprint 8403.
Wittek, Helmut. "Development and Application of a Stereophonic Multichannel Recording Technique for 3D Audio and VR (ORTF-3D)." AES 143rd Convention, New York, 2017.
Hamasaki, Kimio, Koichiro Hiyama, and Reiko Okumura. "The 22.2 Multichannel Sound System and Its Application." AES 118th Convention, 2005, paper 6406.
Rumsey, Francis. Spatial Audio. Oxford: Focal Press, 2001.
Zotter, Franz, and Matthias Frank. Ambisonics: A Practical 3D Audio Theory for Recording, Studio Production, Sound Reinforcement, and Virtual Reality. Springer Topics in Signal Processing 19. Cham: Springer, 2019. Open access (CC BY 4.0). DOI: 10.1007/978-3-030-17207-7.
Gerzon, Michael A. "Periphony: With-Height Sound Reproduction." Journal of the Audio Engineering Society 21, no. 1 (1973): 2–10.
Berkhout, Augustinus J., Diemer de Vries, and Peter Vogel. "Acoustic Control by Wave Field Synthesis." Journal of the Acoustical Society of America 93, no. 5 (1993): 2764–2778. DOI: 10.1121/1.405852.
Pulkki, Ville. "Spatial Sound Reproduction with Directional Audio Coding." Journal of the Audio Engineering Society 55, no. 6 (2007): 503–516.
Braasch, Jonas, Nils Peters, and Daniel L. Valente. "A Loudspeaker-Based Projection Technique for Spatial Music Applications Using Virtual Microphone Control." Computer Music Journal 32, no. 3 (2008): 55–71. DOI: 10.1162/comj.2008.32.3.55.
Herre, Jürgen, et al. "MPEG Spatial Audio Object Coding — The ISO/MPEG Standard for Efficient Coding of Interactive Audio Scenes." Journal of the Audio Engineering Society 60, no. 9 (2012): 655–673. (Standard: ISO/IEC 23003-2.)
Kowalczyk, Konrad, Oliver Thiergart, Maja Taseska, Giovanni Del Galdo, Ville Pulkki, and Emanuël A. P. Habets. "Parametric Spatial Sound Processing: A Flexible and Efficient Solution to Sound Scene Acquisition, Modification, and Reproduction." IEEE Signal Processing Magazine 32, no. 2 (2015): 31–42. DOI: 10.1109/MSP.2014.2369531.
Politis, Archontis, Juha Vilkamo, and Ville Pulkki. "Sector-Based Parametric Sound Field Reproduction in the Spherical Harmonic Domain." IEEE Journal of Selected Topics in Signal Processing 9, no. 5 (2015): 852–866. DOI: 10.1109/JSTSP.2015.2415762.

A note on sourcing: the AES-convention entries elsewhere in this chapter are convention papers/preprints; the parametric and object-coding references in this section are peer-reviewed journal articles with verified metadata. The exact edition year of the ISO/IEC 23003-2 standard is worth a final check against iso.org if cited by date. The claims attributed to RIPL are the tool's design position — the decoupling of capture from reproduction is sourced; the "fewer mics, looser positions" benefit is supported specifically by parametric small-array work (Politis et al., 2015) and otherwise offered as RIPL's approach, flagged as such and kept distinct from the sourced material above.

The question​

What a capture can deliver — and the two philosophies​

Different uses of capture​

Arrays and reaching the stage — technical and artistic consequences​

Questioning the received paradigms​

Parametric and object approaches decouple capture from reproduction​

The thesis: any microphone is an object​

Solutions — with RIPL​

Pitfalls and checklist​

Bibliography​

See also​