FOV Software Architecture

High-Level Architecture Overview

Traditional OBS Studio collapses all active layers, captures, and audio inputs down into a single flattened canvas and one composite audio master track. The FOV variant alters this behavior by maintaining complete structural isolation of visual and auditory inputs from capture all the way to presentation in the Web GUI.

• Video Capture Track Isolation: Every distinct active video source inside the current OBS scene is dynamically allocated its own private frame view context and an isolated hardware encoder instance.
• Audio Source Tracking (1:1 Isolation): Unlike standard OBS which mixes multiple inputs together onto shared master tracks, FOVSystem intercepts audio at the source layer. Each active audio source is automatically hijacked and assigned its own exclusive mixer_id track mask. This prevents audio bleed and locks out human error by disabling custom user mixer modifications in the OBS UI.
• Unified Container Transport: All parallel encoded video and audio streams are packaged sequentially into a singular MPEG-TS (MPEG Transport Stream) before being pushed over the network.

Technical Pipeline Breakdown

Video Architecture

When a video source transitions to an active state, FOVSystem provisions a dedicated VideoTrack node:

1. View Context Allocation: An internal obs_view_t is mapped to the source, ensuring raw frame allocation occurs independently of the primary OBS program canvas.
2. Resolution Alignment: Frame dimensions are processed through an aspect alignment macro: OUT_ALIGN(dimension, 16). This guarantees compatibility with rigorous hardware macroblock requirements.
3. Muxer Binding: Encoders are appended to an obs_encoder_group_t tracking node and indexed continuously using obs_output_set_video_encoder2.

Audio Architecture

To ensure zero configuration effort and bulletproof stream isolation, audio tracks bypass global mixer matrices entirely:

1. Automated Mixer Hijacking: When an audio source is registered, FOVSystem overrides its bitmask using obs_source_set_audio_mixers(source, 1 << mixer_id). This forces the source onto a single exclusive pipeline and clears it from all other tracks.
2. Hardware UI Enforcement: By programmatically assigning and managing these bits at runtime, the platform ignores any changes made in the OBS “Advanced Audio Properties” layout, preserving track isolation.
3. Index Sequencing: Within the MPEG-TS multiplexer loop, audio track indexing begins exactly where video track array iteration terminates, ensuring flawless track allocation inside the transport stream:

$$\text{Transport Stream Track ID} = \text{Video Track Count} + \text{Source Track Index}$$

Core Class Reference: FOVSystem

The engine relies on FOVSystem to arbitrate scene modifications, monitor input states, and reconstruct the active encoder layout on the fly.

• void initSystem(obs_output_t *muxerOutput, obs_data_t *vSettings, obs_data_t *aSettings) Configures and initializes the system state. Binds the structural output pipeline to the ffmpeg-mpegts muxer reference and locks in current initialization profiles.

• void syncSources() The runtime management loop. Executed automatically to track context mutations:

• Enumerates all global sources utilizing obs_enum_sources.
• Checks filtering flags (OBS_SOURCE_VIDEO and OBS_SOURCE_AUDIO).
• Automatically registers active audio sources, binds them programmatically to a unique available mixer_id, and provisions an encoder for that isolated slot.
• Discards elements that have transitioned to inactive statuses, releases stale encoders, and appends new tracks to the multiplex group.

• void updateEncoderGroup() Performs critical cleanup of the output container mappings. It detaches all existing audio and video paths from the active muxer reference, registers the modified tracks under a clean unified obs_encoder_group_t, and remaps container tracks dynamically to clear stream PIDs.