media-os

Trim, cut, split, segment, and concatenate media with ffmpeg (stream copy when possible, re-encode across cut boundaries). Use when the user asks to trim a video, cut a clip, extract a segment by timestamps, remove a section, split into parts, join/merge videos, concatenate files, or build a segmented HLS-style playlist.

Install and configure OBS Studio programmatically: install via brew cask / winget / Flatpak / apt, author profiles (basic.ini, streamEncoder.json, recordEncoder.json), author scene collections (scenes JSON), manage global.ini, set defaults for encoder / output / audio / hotkeys, cross-platform config paths. Use when the user asks to install OBS, set up an OBS profile, create a scene collection from code, configure OBS defaults without the GUI, edit basic.ini, manage multiple OBS profiles, or script a fresh OBS install with known-good settings.

Deep media inspection + automated QC — ffprobe stream details, MediaInfo diagnostics, VMAF/PSNR/SSIM quality metrics, PySceneDetect scene cuts, crop/silence/black-frame/interlacing detection, ffplay scope debugging, NAL/SEI bitstream forensics, metadata audits, loudness compliance against Spotify/Apple/ATSC/EBU specs, and automated CI QC gates. Use when the user says "QC this file", "run VMAF", "compare encoders", "detect scene cuts", "check loudness compliance", "validate delivery spec", "automated QC pipeline", or any deep inspection / quality gating.

Routes media production requests to the right Media OS specialist subagent. ALWAYS use this skill when the user expresses ANY media production intent — "go live", "start streaming", "OBS broadcast", "wire up live rig", "NDI to stream", "PTZ camera setup", "DeckLink capture stream", "HLS deliver", "DASH package", "encode for streaming", "CDN upload", "make a HLS manifest", "multi-bitrate ladder", "CMAF package", "LL-HLS", "low-latency stream", "Widevine package", "DRM package", "package for streaming", "broadcast deliver", "MXF master", "IMF package", "ProRes master", "DPP deliver", "AS-11 deliver", "Netflix deliver", "broadcast spec deliver", "deliver for air", "create IMF", "Premiere to Resolve", "round-trip", "OTIO export", "editorial conform", "XML round-trip", "FCPXML", "EDL export", "AAF export", "convert timeline", "Avid to Premiere", "Resolve to Premiere", "upscale this", "interpolate frames", "denoise AI", "remove background", "rotoscope", "matte", "depth estimate", "AI upscale", "RIFE", "Real-ESRGAN"

Ingest from every source — web (yt-dlp), screen / webcam / mic capture, SDI (DeckLink), DSLR tether (gphoto2), NDI network sources, RTSP / IP cameras via MediaMTX, PTZ — then verify integrity (SHA-256 + full-decode pass), preserve metadata (EXIF / XMP / IPTC), normalize to MKV or FFV1 / J2K / ProRes archival containers, and push to cold-storage cloud (Glacier / B2 / Archive.org). Use when the user says "download YouTube playlist", "capture SDI for 24 hours", "archive IP cameras", "preserve VHS rips", "tether DSLR timelapse", "cold-storage upload", or any ingest-to-archive workflow.

Restore, upscale, and enhance existing footage using 2026 open-source AI models — Real-ESRGAN/SwinIR/HAT super-resolution, RIFE/FILM interpolation, DeepFilterNet/RNNoise audio denoise, rembg/BiRefNet/RVM matting, Depth-Anything v2 depth — with strict OSI-open commercial-safe license filter. Use when the user says "upscale old footage", "remaster", "enhance quality", "30 to 60fps", "AI denoise", "restore VHS", "remove background from video", or anything about AI-driven footage restoration.

Generate media from scratch with 2026 open-source AI — TTS voiceover (Kokoro / OpenVoice / Piper), image gen (FLUX-schnell / Kolors / Sana / ComfyUI), video gen (LTX-Video / CogVideoX / Mochi / Wan), music (Riffusion / YuE), lipsync talking heads (LivePortrait / LatentSync), OCR (PaddleOCR / Tesseract 5 / TrOCR), zero-shot tagging (CLIP / SigLIP / BLIP-2 / LLaVA). Strict commercial-safe license filter. Use when the user says "generate a video", "TTS voiceover", "AI explainer video", "clone my voice", "generate music", "AI image", "digital human", or anything about from-scratch AI media.

Full-stack audio routing across macOS / Linux / Windows, MIDI + OSC control surfaces, DAW integration over JACK, DSP (EQ / compression / loudness / spatial / binaural HRTF), AI denoise (DeepFilterNet / RNNoise / Resemble Enhance), stem separation (Demucs), TTS, Whisper transcription, and EBU R128 / Spotify / Apple / ATSC loudness certification. Use when the user says "set up DAW", "route audio through OBS", "MIDI control surface", "binaural ASMR", "stem separation", "loudness to −16 LUFS", "JACK transport", or anything audio-stack-related.

Produce broadcast-grade MXF OP1a or Netflix IMF masters with HDR dynamic metadata (Dolby Vision / HDR10+), SDI capture/playout, ACES color management, CEA-608/708 captions, and timecode-accurate delivery. Use when the user says "deliver to broadcast", "make an IMF for Netflix", "MXF OP1a", "SDI ingest", "Dolby Vision profile 8.1", "convert DV profile 7 to 8.1", "broadcast master", or anything about professional broadcast/OTT mastering.

Move timelines between NLEs (Premiere, Final Cut Pro X, DaVinci Resolve, Avid, Lightworks) using OTIO as the pivot format, with cuts, transitions, media links, and timecode preserved. Use when the user says "move my Premiere timeline to Resolve", "export from FCP to Avid", "round-trip AAF", "conform multi-rate edit", "generate proxies and relink", "FCPXML to AAF", or anything about editorial round-trip between pro NLEs.

End-to-end HDR authoring — HDR10 (static MDC + MaxCLL/MaxFALL), HDR10+ dynamic metadata JSON, Dolby Vision profiles 5/7/8.1/8.4, HLG, PQ↔HLG conversion, SDR↔HDR tone-mapping (hable/mobius/reinhard/bt2390/aces), dovi_tool + hdr10plus_tool orchestration, and multi-format dual-delivery. Use when the user says "HDR10+", "Dolby Vision", "PQ to HLG", "tone-map to SDR", "ACES RRT", "HLG broadcast", "DV profile 7 to 8.1", "YouTube HDR", or anything HDR-color-pipeline.

End-to-end live production — OBS as the mixer with obs-websocket control, MIDI/OSC cue triggers, DMX lighting sync, PTZ camera moves, system audio routing, NDI/SDI I/O, and multi-protocol egress via MediaMTX. Use when the user says "go live", "run a live show", "live stream", "wire up OBS", "scene switch on MIDI", "OBS + lighting + PTZ", "broadcast webinar", or anything involving real-time production with multiple control surfaces firing together.

Raw podcast recordings → polished, loudness-compliant (EBU R128 / Spotify / Apple Podcasts / ACX), caption-accurate (Whisper), chapter-tagged deliverables across MP3 / M4A / video-podcast. Covers multi-mic capture, AI denoise, sidechain ducking, auto-chaptering, multi-language subtitles, ID3/MP4 metadata, and batch release. Use when the user says "publish a podcast episode", "normalize loudness to −16 LUFS", "auto-chapter", "podcast subtitles", "batch process episodes", "interview diarization", or anything podcast-production-related.

Deliver live or VOD content over HLS / DASH / RTMP / SRT / WHIP / RIST with adaptive bitrate ladders, DRM (Widevine / PlayReady / FairPlay cbcs), multi-protocol fanout via MediaMTX, and CDN upload (Cloudflare Stream / Mux / Bunny / S3). Use when the user says "stream to HLS", "build an ABR ladder", "add Widevine", "multi-bitrate DASH", "WHIP ingest", "low-latency HLS", "SRT contribution", "stream to multiple platforms", "SFU for WebRTC", or anything involving large-scale distribution.

VFX-grade I/O and handoff — read/write USD (Pixar Universal Scene Description), deep and multi-part OpenEXR, color-managed OpenImageIO with ACES/OCIO configs, dailies, cryptomatte extraction, plate generation for compositors, and delivery back to editorial or IMF. Use when the user says "ACES pipeline", "EXR sequence", "USD stage", "VFX plates", "cryptomatte", "ARRI LogC to ACEScg", "deep compositing", or anything VFX-I/O related.

Traditional VOD finishing — transcode to mezzanine + proxies, cut/trim/concat, color grade (LUT / OCIO / manual), stabilize handheld, denoise, retime (slow-mo / reverse / speed ramps), chromakey, lower thirds, burn subtitles, loudness normalize, thumbnails/sprite sheets, chapter metadata, and final delivery H.264/AV1 for YouTube/social/archive. Use when the user says "finish my YouTube video", "post-production", "color grade and stabilize", "convert to vertical TikTok", "batch wedding videos", "sprite sheet for scrubbing", or standard VOD finishing.

Search and fetch docs for system-level audio routing subsystems (lives BELOW ffmpeg): PipeWire (docs.pipewire.org — pw-cli, pw-dump, pw-link, pw-cat, pw-top, pw-loopback, WirePlumber session manager, PA+JACK compat), JACK Audio Connection Kit (jackaudio.org, jackd/jack_lsp/jack_connect, JACK1 vs JACK2, JackTrip for network audio), Apple Core Audio (developer.apple.com archive — HAL plugins, Aggregate/Multi-Output devices, BlackHole virtual driver), Windows WASAPI (learn.microsoft.com — exclusive vs shared mode, MMDevice, Audio Engine, WDM/KS stack, VB-Cable, VoiceMeeter). Use when the user asks to look up a PipeWire/JACK/CoreAudio/WASAPI API, verify a system audio routing command, check virtual audio driver setup, or find subsystem docs.

Look up Blackmagic DeckLink SDK docs at blackmagicdesign.com/developer: DeckLink C++ / COM-style API (DeckLinkAPI.h, DeckLinkAPI.idl), Desktop Video SDK (driver + runtime), DeckLink Arduino SDK. Core interfaces: IDeckLink, IDeckLinkInput, IDeckLinkOutput, IDeckLinkDisplayMode, IDeckLinkConfiguration, IDeckLinkNotification, IDeckLinkProfileAttributes. BMDPixelFormat (8BitYUV/10BitYUV/ARGB/BGRA/10BitRGB/r210/v210), BMDDisplayMode (SD/HD/2K/UHD/4K/8K). Relationship to ffmpeg decklink demuxer/muxer. Use when the user asks to look up a DeckLink API, check supported pixel formats or display modes, verify an IDeckLinkInput or IDeckLinkOutput method, or find Blackmagic SDK documentation.

Search and fetch official FFmpeg documentation from ffmpeg.org: ffmpeg, ffmpeg-all, ffmpeg-filters, ffmpeg-codecs, ffmpeg-formats, ffmpeg-protocols, ffmpeg-devices, ffmpeg-bitstream-filters, ffmpeg-utils, ffmpeg-scaler, ffmpeg-resampler, ffprobe-all, ffplay-all, general, faq, platform, developer. Use when the user asks to look up an ffmpeg option, find filter parameters, check a muxer or demuxer option, search ffmpeg docs, get the official description of a codec flag, find out what an ffmpeg flag does, or verify ffmpeg syntax against the real documentation.

Search and fetch official GStreamer documentation from gstreamer.freedesktop.org: application-development, tutorials, plugin-development, tools (gst-launch-1.0, gst-inspect-1.0, gst-device-monitor-1.0, gst-discoverer-1.0, gst-play-1.0, gst-typefind-1.0), libraries (core gstreamer, base, video, audio, rtp, webrtc, sdp, pbutils, app, coreelements, coretracers, libav). Use when the user asks to look up a GStreamer element, find an element's properties or pads, check caps negotiation rules, search GStreamer docs, verify a gst-launch pipeline, look up the gstreamer webrtcbin or hlssink API, or verify a GStreamer call against the real docs.

Search and fetch docs for HDR dynamic-metadata CLIs: dovi_tool (Dolby Vision RPU authoring — profiles 4, 5, 7, 8.1, 8.4 — github.com/quietvoid/dovi_tool) and hdr10plus_tool (HDR10+ SEI authoring — github.com/quietvoid/hdr10plus_tool). Canonical subcommand lists (info, generate, editor, export, plot, convert, demux, mux, extract-rpu, inject-rpu for dovi_tool; extract, inject, remove, plot, editor for hdr10plus_tool), HEVC Annex-B + Matroska I/O, integration with ffmpeg bitstream filters + mkvmerge + MP4Box. Use when the user asks to look up a dovi_tool or hdr10plus_tool subcommand, check Dolby Vision profile support, verify HDR dynamic-metadata tool usage, or read the canonical GitHub README.

Search and fetch official MediaMTX documentation from mediamtx.org: kickoff (introduction, install, upgrade), features (architecture, publish, read, record, playback, authentication, hooks, metrics, forward, proxy, performance, rtsp-specific-features, rtmp-specific-features, webrtc-specific-features, srt-specific-features, absolute-timestamps, always-available, on-demand-publishing, decrease-packet-loss, extract-snapshots, remuxing-reencoding-compression, start-on-boot, embed-streams-in-a-website, expose-the-server-in-a-subfolder, logging, configuration), publish + read howtos (ffmpeg, gstreamer, obs-studio, python-opencv, golang, unity, raspberry-pi-cameras, rtsp, rtmp, hls, srt, webrtc, rtp, mpeg-ts), references (configuration-file, control-api). Use when the user asks to look up a MediaMTX config option, find a control-api endpoint, verify supported protocols, check auth modes (internal/HTTP/JWT), or read mediamtx howtos.

Search and fetch official NDI documentation from docs.ndi.video (Vizrt, formerly NewTek NDI): SDK landing, release notes, port numbers (mDNS 5353, messaging 5960, streams 5961+), HDR spec (PQ and HLG), SDK vs Advanced SDK differences, NDI 6 features (native HDR, HX3, Agilex 7, NDI Bridge), plus DistroAV (the OBS plugin formerly obs-ndi) README fallback. Use when the user asks to look up an NDI API, NDIlib_send/recv/find functions, check NDI port numbers, verify NDI wire protocol details (SpeedHQ codec, NDI|HX vs NDI full, NDI 6 vs NDI 5 differences), read NDI SDK docs, check SDK-vs-Advanced-SDK feature matrix, integrate NDI with OBS via DistroAV, or verify an NDI call against the real documentation.

Search and fetch official OBS Studio documentation: docs.obsproject.com (backend-design, plugins, frontends, graphics, scripting, reference-core, reference-modules, reference-core-objects, reference-libobs-util, reference-libobs-callback, reference-libobs-graphics, reference-libobs-media-io, reference-frontend-api), install wiki, obs-plugintemplate README, obs-websocket protocol reference. Use when the user asks to look up an OBS API, find an obs_ function signature, check libobs callback signatures, read OBS scripting API, search obs-websocket request/event/opcode details, learn OBS plugin authoring, look up frontend API calls, or verify an OBS SDK call against the real documentation.

Search and fetch official OpenTimelineIO (OTIO) docs from opentimelineio.readthedocs.io and github.com/AcademySoftwareFoundation/OpenTimelineIO: timeline primitives (RationalTime, TimeRange, TimeTransform), containers (Timeline/Stack/Track), content (Clip/Gap/Transition), MediaReferences, tutorials (quickstart, timeline-structure, time-ranges, write-a-schemadef-plugin, adapters), Python API (schema/core/opentime/algorithms/media_linker/hooks), adapter catalog (otio_json, otiod, otioz native; cmx_3600 EDL, fcp_xml, fcpx_xml, aaf_adapter, ale, burnins, maya_sequencer, xges, svg via OpenTimelineIO-Plugins). Use when the user asks to look up an OTIO class, find an adapter for a NLE format, verify Python bindings, or read OpenTimelineIO docs.

Search and fetch docs for PTZ camera control protocols: Sony VISCA (Command List v2.00, BRC-H900, SRG-300H, PTZOptics VISCA over IP) — byte-level serial or UDP:52381 with payload-wrapped VISCA, and ONVIF (onvif.org profiles — S streaming deprecated, T successor with H.264/H.265/metadata, G on-device recording, Q deprecated, M metadata/analytics, A/C/D access control). WS-Discovery multicast 239.255.255.250:3702, SOAP/XML with WS-Security UsernameToken digest, Device/Media/PTZ/Imaging/Events/Recording services. Use when the user asks to look up a VISCA command, find an ONVIF service, check WS-Discovery packet structure, verify PTZ protocol syntax, or read the canonical PTZ specs.

OBS plugin authoring and ecosystem: scaffold a new C++ plugin from obs-plugintemplate (CMake, cross-platform GitHub Actions CI), register obs_source_info / obs_output_info / obs_encoder_info / obs_service_info, implement obs_module_load / obs_module_unload with OBS_DECLARE_MODULE, plus install / configure common community plugins (StreamFX, Advanced Scene Switcher, Move, NDI, Source Record, Backtrack, obs-websocket). Use when the user asks to write an OBS plugin, build a custom source / output / encoder, use obs-plugintemplate, install a specific OBS plugin, or set up the OBS plugin build environment.

Write Python / Lua scripts that run inside OBS Studio: script lifecycle (script_load, script_unload, script_description, script_properties, script_defaults, script_update, script_save, script_tick), registering frontend event callbacks, hooking signals, accessing sources + scenes + filters from a script, using obspython / obslua bindings, building script UIs with obs_properties_t. Use when the user asks to write an OBS Python script, write an OBS Lua script, automate something inside OBS without a compiled plugin, add a script-side hotkey, build a tools-menu script, or hook OBS signals from Python.

Remote-control a running OBS Studio instance via obs-websocket (WebSocket API, default port 4455): authenticate with SHA-256 challenge+salt, Identify handshake, start/stop streaming and recording, switch scenes, toggle sources and filters, set audio mixer levels, trigger replay buffer + virtual camera, subscribe to live events, send request batches. Use when the user asks to control OBS from a script, switch OBS scenes remotely, start or stop recording programmatically, automate a stream, send an obs-websocket request, subscribe to CurrentProgramSceneChanged events, connect to ws://host:4455, or build a browser/CLI OBS remote.

Use when building a skill, creating a SKILL.md, packaging a workflow, making a slash command, or asked "how do I make a skill". Scaffolds the folder, generates SKILL.md from a template, validates against spec. Produces a complete ready-to-deploy skill folder: scripts, references, assets. Also use to review or improve an existing skill.

Manipulate macOS audio at the Core Audio HAL layer: afinfo (audio file metadata), afplay (play), afconvert (convert format/rate/bit-depth/channel layout), say (TTS), SwitchAudioSource (change default input/output/system device — brew install switchaudio-osx), audiodevice CLIs (npm/Go variants), AppleScript via osascript to drive System Preferences Sound. HAL plugins at /Library/Audio/Plug-Ins/HAL/ load virtual drivers (BlackHole 2ch/16ch/64ch, Loopback, Background Music). Aggregate Device (drift-corrected multi-in) and Multi-Output Device via Audio MIDI Setup.app or AudioHardwareCreateAggregateDevice. Use when the user asks to change default macOS audio device, route audio between apps on Mac, install BlackHole, build an Aggregate Device, convert audio files with afconvert, or script Mac audio from the terminal.

Route professional low-latency audio with JACK Audio Connection Kit (jackaudio.org): jackd server (ALSA/CoreAudio/PortAudio/ASIO backends), jack_control (D-Bus JACK2 config), jack_lsp (list ports/connections), jack_connect/jack_disconnect (manage links), jack_cpu_load, jack_samplerate, jack_bufsize (runtime period size), jack_transport, jack_rec (multi-channel WAV recording), jack_iodelay (round-trip latency), jack_wait, jack_midi_dump. JACK1 C maintenance vs JACK2 C++ multicore. Modern Linux swaps jackd for PipeWire's libjack shim — same client API. JackTrip for uncompressed multichannel UDP over the Internet. Use when the user asks to run a JACK server, connect audio clients, measure roundtrip latency, record from JACK ports, set JACK period size, or bridge JACK over the network with JackTrip.

Route and introspect Linux audio/video/MIDI graphs with PipeWire: pw-cli (interactive object/module control), pw-dump (full graph JSON with --monitor), pw-link (list/create/destroy port links), pw-cat / pw-play / pw-record (playback+record), pw-top (real-time DSP load + xrun viewer), pw-metadata (shared metadata store), pw-loopback (source/sink bridges), pw-midiplay/pw-midirecord, pw-jack (JACK client compat shim), pw-mon, pw-profiler. WirePlumber session manager handles routing/defaults/profiles. Replaces PulseAudio (via pipewire-pulse) and JACK (via libjack.so shim) simultaneously. Use when the user asks to route Linux audio, link two apps, list audio devices on Linux, create a loopback, record system audio, monitor the audio graph, or set up a virtual audio device on modern Linux.

Manipulate Windows audio via WASAPI (learn.microsoft.com/coreaudio): SoundVolumeView + svcl.exe (NirSoft — set default device, per-app volumes, save/load profiles), AudioDeviceCmdlets PowerShell module (Get-AudioDevice, Set-AudioDevice, Get-AudioDevicePlaybackMute), nircmd setdefaultsounddevice. Shared mode (Windows engine resamples all streams to endpoint fixed format) vs Exclusive mode (bit-perfect, audiophile, ~sub-3ms latency with IAudioClient3 + event-driven callbacks). WASAPI then MMDevice then Audio Engine (audiodg.exe) then WDM/KS. ASIO is Steinberg's parallel stack bypassing Windows audio. Virtual audio devices: VB-Audio Virtual Cable, VoiceMeeter/Banana/Potato. Use when the user asks to change Windows default audio device, route app audio, use VoiceMeeter, install VB-Cable, set a specific device from PowerShell, or enable exclusive-mode low-latency output.

Google MediaPipe Tasks API (current supported, NOT legacy mediapipe.solutions): Vision Tasks (Face Detector, Face Landmarker, Face Stylizer, Hand Landmarker, Gesture Recognizer, Pose Landmarker, Holistic Landmarker, Object Detector, Image Classifier, Image Segmenter, Interactive Segmenter, Image Embedder, Image Generator), Audio Tasks (Audio Classifier), Text Tasks (Text Classifier, Text Embedder, Language Detector), GenAI Tasks (LLM Inference — on-device Gemma/Phi-2/Falcon/StableLM). Three run modes: IMAGE, VIDEO, LIVE_STREAM (async). Python: mediapipe.tasks.python.vision/audio/text/genai. Docs at ai.google.dev/edge/mediapipe/solutions. Model Maker for transfer learning. Use when the user asks to detect pose/face/hand landmarks, classify an image, run a body-pose model, segment people, recognize gestures, do on-device LLM inference via MediaPipe, or build a mobile/embedded CV pipeline.

Computer vision on images + video with OpenCV (docs.opencv.org/4.x/): core (Mat / ndarray, linalg), imgproc (filtering, morphology, geometric transforms, color conv, histograms, contours), imgcodecs (imread/imwrite — PNG, JPEG, TIFF, WebP, EXR), videoio (VideoCapture / VideoWriter — FFMPEG, GStreamer, V4L2, AVFoundation, MSMF backends), calib3d (camera calibration, stereo, solvePnP, homography), features2d (ORB, AKAZE, BRISK, KAZE, SIFT), objdetect (FaceDetectorYN, FaceRecognizerSF, HOG, ArUco, QRCode), dnn (ONNX/TF/Caffe/Darknet, CUDA + OpenVINO + Vulkan backends, model zoo at github.com/opencv/opencv_zoo), photo (inpainting, denoising, HDR Debevec/Robertson), stitching (panorama), video (optical flow Farneback/LK/DIS, BackgroundSubtractorMOG2/KNN, tracking KCF/CSRT/MIL/Nano/Vit/DaSiamRPN). Python-first via cv2. Use when the user asks to detect faces, track objects, run YOLO inference, capture + process webcam frames, calibrate a camera, stitch a panorama, or use OpenCV from Python/C++.

Capture and play back SDI/HDMI video via Blackmagic DeckLink hardware: ffmpeg decklink indev/outdev (-f decklink -list_devices 1, -f decklink -list_formats 1, -f decklink -i 'DeckLink Mini Recorder 4K', -f decklink 'DeckLink Mini Monitor'), SDK Samples (CapturePreview, LoopThroughPreview, SignalGenerator test-pattern output, StatusMonitor, DeviceList, TestPattern, 3DVideoFrames, StreamOperations, FrameServer, AudioMixer). Pixel formats (uyvy422 = bmdFormat8BitYUV, v210 = bmdFormat10BitYUV), four-CC display modes (Hp60 = 1080p60, 4k60 = UHD60). Requires Desktop Video driver + ffmpeg built --enable-decklink. Use when the user asks to capture from DeckLink, output to SDI, ingest HDMI via Blackmagic, list DeckLink devices/formats, generate test patterns out SDI, or install the DeckLink SDK.

360° / VR and stereoscopic 3D video with ffmpeg: v360 filter (equirectangular, cubemap 3x2/6x1/1x6, fisheye, flat, hequirect, dfisheye, cylindrical, perspective, barrel, sinusoidal, half_equirectangular, stereographic, mercator, ball, hammer, pannini, eac, c3x2/c6x1/c1x6, rfish), stereo3d filter (SBS ↔ TAB ↔ interlaced ↔ anaglyph), framepack, spatial audio placeholder. Use when the user asks to convert 360 video projections, equirectangular to cubemap, unwrap fisheye, convert side-by-side 3D to top-bottom, make anaglyph 3D, reproject 360 video, or handle VR180/VR360 content.

Apply audio filters with ffmpeg -af / -filter_complex: loudnorm (EBU R128), volume, dynaudnorm, compand, equalizer, bass/treble, aresample (sample rate), atempo (speed), apad/atrim, amix, amerge, pan (channel layout), silenceremove, afade, asetrate. Use when the user asks to normalize loudness, boost/reduce volume, EQ audio, resample, change speed/tempo, trim silence, mix audio tracks, merge channels, swap/remap channels, convert mono/stereo/5.1, or fade in/out.

Creative and restoration audio effects with ffmpeg: chorus, flanger, aphaser, tremolo, vibrato, aecho, crystalizer, adecorrelate, adeclick, adeclip, apsyclip, asoftclip, speechnorm, dialoguenhance, adynamicequalizer, adynamicsmooth, afreqshift, aphaseshift, asupercut, asuperpass, asuperstop, atilt, afftfilt, bandpass, bandreject, biquad, allpass, ladspa, lv2. Use when the user asks to add echo/chorus/flanger, create psychedelic audio effects, enhance dialogue, de-click vinyl rips, soft-clip audio, apply LADSPA or LV2 plugins, or shift frequency.