Technical specifications
Every function.
Every function.
Every setting.
The complete reference for PodcastMagic 360, the DaVinci Resolve Studio plugin that turns one 360° camera into a multicam podcast edit: what goes in, what comes out, what the plugin analyses and every option on every page of the panel.
Requirements
What you need
| Editor | DaVinci Resolve Studio 21.1 or later. The free DaVinci Resolve does not run workflow integration plugins; deflicker, noise reduction, AI UltraSharpen, Voice Isolation and the AI Speech Generator are Studio features. |
| System | Windows 10 / 11 (64-bit) or macOS on Apple Silicon. A GPU as Resolve recommends it – the camera reframing runs as a GPU tool in Fusion. |
| Resolve setting | Preferences › System › General › External scripting using: Local |
| Install | One line in PowerShell (admin) or Terminal, see the installation guide. Bundled: a minimal ffmpeg (LGPL) to split GoPro files without re-encoding and to compute the podcast sound. |
| Internet | For updates, the licence check (every 14 days, works 30 days offline) and translated subtitles only. Everything else runs offline on your computer. |
| Optional | An Anthropic API key for translated subtitles · Resolve's AI Speech Generator for dubbing (Help › Extras Download Manager) · Docker for the more accurate speaker analysis |
Input
Straight from the camera – no export, no stitching
Put the camera's own files in the folder. There is no GoPro Player or Insta360 Studio export, no stitching and no conversion to equirectangular first: the plugin reads the original lenses directly and reframes from them in full resolution.
| Source | File | What the plugin does |
|---|---|---|
| GoPro MAX / MAX 2 | .360 (also several chapters in order) | Copies the two video streams into two files (front, back) without re-encoding – seconds, not hours – and imports them. Reframes straight from GoPro's cube projection (EAC); the back half comes in through a Fusion loader. The camera's spatial audio (first-order ambisonics) is used to hear where each voice comes from. |
| Insta360 (X series) | .insv | Only a link named .insv.mp4 (Resolve takes .mp4) – no copy, no conversion. Reframes straight from both fisheye circles with the lens calibration stored in the file, blended softly at the seam. |
| Any 360° camera | .mp4 / .mov / .m4v, equirectangular 2:1 | Used as it is. |
| Podcast audio | a separate recording in the same folder (e.g. .wav) | Put on its own audio track and synced to the camera. The camera's own sound is kept, muted. |
| Resolution | any; 5.6K or more recommended | Close-ups are cut out of the sphere: the more pixels, the sharper they get. |
| Frame rate | any the camera records | The project takes the clip's frame rate before the audio comes in; playback is set to match. Under 50 Hz light record 25 fps (GoPro: anti-flicker 50 Hz), under 60 Hz 29.97 / 30 fps. |
| Proxies | Resolve's proxies | Used when they exist. |
| People | 1 to 9 | Each with a number key, a colour and its own framing. |
Output
A real Resolve timeline – finish anything afterwards
Nothing is baked. The plugin writes an ordinary timeline: every move is keyframes, every look a node, the sound a clip. Edit, Fusion, Color, Fairlight and Deliver stay fully yours.
| Timelines | 16:9 on the episode's timeline; every other format on its own copy of the timeline, framed anew (not cropped from 16:9). |
| The picture | A Fusion composition on the 360° clip with the DM360 Reframe tool: pan, tilt, field of view, roll and little planet as keyframes with real easing (Bezier handles), editable in the Fusion page's spline editor. |
| Two versions per clip | Final (every quality tool) and Cut (without the slow ones for smooth playback). Switch in the plugin's top bar – instantly, without writing again. |
| Finish picture | Pass 1 renders the cut once without the slow tools as ProRes 422 .mov at timeline resolution next to the camera file (<name>_dm360_hd.mov); pass 2 places it on track V2 with deflicker, noise reduction and sharpness – full quality with smooth playback. Recomputed only when the cut or the look changed. |
| Colour | The look as a CDL on the clip's first node on the Color page – keep grading with nodes and LUTs. |
| Sound | The podcast master as a WAV (48 kHz, 24-bit, −16 LUFS, −1.5 dBTP) next to the original, placed on the timeline; Voice Isolation and Dialogue Leveler as clip settings in Resolve. One click back to the original sound. |
| Subtitles | One .srt per language next to the video, timed from In – for YouTube or to drag into Resolve. |
| Dubbing | One audio track per language (Dub EN, Dub FR …) next to the original, muted until chosen. |
| Emotion display | PNG images next to the video (<name>_dm360_overlays/), merged in the composition and moved with the face. |
| The plugin's own data | The camera track (cuts, people, settings) as <video>.dm360.json, speech and reactions beside it – opened again with the clip. |
| Render | On the Deliver page as usual: any format, codec and resolution Resolve offers. |
Output formats
| Format | For | From a 1920×1080 timeline | From a 3840×2160 timeline |
|---|---|---|---|
| 16:9 | YouTube, TV | 1920×1080 | 3840×2160 |
| 2.39:1 | cinema widescreen | 1920×804 | 3840×1606 |
| 9:16 | Story, Reels, Shorts, TikTok | 1080×1920 | 2160×3840 |
| 4:5 | Instagram feed | 1080×1350 | 2160×2700 |
| 1:1 | square | 1080×1080 | 2160×2160 |
Analysis
What it measures
All analysis runs on your computer – in Resolve and in the plugin. Only translation sends the conversation's text to Anthropic, and only when you ask for it.
| Signal | How | Used for |
|---|---|---|
| Who speaks when | Resolve's speech analysis on the podcast audio (GPU, fast); optionally a more accurate speaker analysis in a Docker container (pyannote, slow) | the basic cut, slightly ahead of the first word |
| Faces all around | face detection in 8 views around the sphere (45° apart, 70° wide, overlapping) | finding the people and their seats; framing |
| Voice to face | direction of the voice in the camera's spatial audio (GoPro) and mouth movement while each voice talks alone; best overall assignment | which person to show for which voice |
| Facial expressions | MediaPipe Face Landmarker: 52 face shape values and the head's rotation, twice a second | smile, surprise, doubtful, sad, angry, disgust, yawn, eyes closed, nod, head shake |
| Sounds | YAMNet audio classifier on the podcast audio | laugh, gasp, sigh, applause, crying, shouting, singing, whistling |
| Face position | stored with every measurement | the emotion display follows the head |
| Skin tone and exposure | the middle of each face; grey areas of the room | matching faces; white balance and exposure of the first grade |
| Horizon | set by hand (drag, , .) | levels the whole sphere for every shot |
All settings
Every page, every option
The panel works like Resolve: seven pages in a bar at the bottom, tabs on each page, the main button on top. Explanations sit behind ⓘ. Every setting is saved with the episode; profiles keep them across episodes.
Top bar and settings
| Setting | Options | What it does |
|---|---|---|
| Format | 16:9 · 2.39:1 cinema · 9:16 story/reels · 4:5 feed · 1:1 | applies to preview, calibrating, cutting and writing |
| Version | Cut · Final | which of the two compositions Resolve plays and renders |
| Record (R) | on / off | live overdub with the keyboard |
| Language | English · Deutsch | the whole panel |
| Licence | Activate · Buy a licence · Free this computer | one key per computer and major version |
| System check | – | Resolve Studio version, camera tool, ffmpeg, detection models, licence, API key, AI Speech Generator, updates – each with what to do |
| Anthropic API key | Save · Remove key | stored encrypted for your user account, never shown again |
1 · Media
| Setting | Options | What it does |
|---|---|---|
| Everything automatic | button (again: another variant) | set up, listen, look, match voices, reactions, direct, polish – from the folder to the first cut |
| Studio type | Table · Lounge · Interview (two facing) · Round / panel (4+) · Standing / stage | starting points for portrait size, two-shots, rounds and style |
| Target format | as in the top bar | every format except 16:9 gets its own timeline |
| Style preset | Calm · Dynamic · Talk show · Punchy (style packs by editors and colorists later) | a whole way of cutting in one choice: pace, moves, ease, camera operator, two-shots – each still changeable under Direct |
| Variation | Hold the style · Vary a little · Vary freely | the same as under Direct |
| Picture quality | Automatic · measured (default) · Fast · Best | Automatic measures the recording: brightness pulsing (mains flicker), noise, blown highlights, how much the close-ups are enlarged – and sets deflicker, noise reduction, sharpness, contrast and vignette only where needed, with the reason. Best finishes the picture right after writing |
| Episode | Only set up folder … · Load clip from Resolve · Add camera file … | a folder, the clip under the playhead (V1) or a single .360 / .insv |
| Podcast audio | Speech analysis with Resolve (fast, GPU) · more accurate analysis in the container (slow) | finds the separate recording and its offset |
| Start over | Cuts · People · Horizon · View · All | Ctrl+Z brings everything back |
| Settings profiles | Load · Save · delete | all settings under one name (your own moves included), without people and cuts – put the same style on any film, with new people in another room |
| Studio template | Apply · Save | same room next week: seats, names, extra lines, framing, horizon, voices, face matching and style – found by itself in the folder, the one above or next door |
2 · Calibrate
| Setting | Options | What it does |
|---|---|---|
| Horizon | drag with the right mouse button · , . (Shift: finer) · Reset | levels a tilted camera for the whole sphere |
| People | Calibrate people (find faces) · per person Show (into the picture) · − / + (zoom), ▲ / ▼ (height), × (remove) · Ctrl+number stores a view · click the name to rename · an extra line under it (e.g. “Host”) | up to 9 people, each framed from the measured face; name and extra line go into the lower third |
| Save the calibration | as a studio template | every change to a person is kept with the episode by itself; this keeps the calibration for the next episode in the same room |
| Portrait | tight (head fills the frame) · medium (head and shoulders) · wide (with upper body) | changes the framing at once, keeping your corrections |
| Voices | Start speech analysis · ▶ listen · assign to a person or “don't show” · Assign voices by room audio | which voice belongs to which face; remembered |
| Reactions | Detect reactions | measures faces and sounds over the episode (In to Out) |
3 · Direct
| Setting | Options | What it does |
|---|---|---|
| Compute suggestion · Another variant | buttons | the autopilot's cut; your own cuts are kept |
| Variation | Hold the style · Vary a little · Vary freely | moves, push-ins and look-arounds in both directions, how restless the camera is – who is shown when stays right |
| Cut | ||
| Style | calm (~6 changes/min) · dynamic (~7/min) · talk show (~11/min) · punchy (fast, varied) | pace and variety of the cut |
| Motion | cuts only · mixed · moves only | how the camera gets to the next person |
| Two-shot | off · on quick back-and-forth · also now and then in conversation | two people in one frame, also sitting far apart (up to 150° with wide-angle correction) |
| Lead | variable (0.25–1 s before) · 0 · 0.25 · 0.5 · 0.75 · 1 s before | on the speaker before the first word |
| Camera work | Slowly push in on long shots · Briefly show the listener during long monologues · Briefly show short interjections · Wide shot at the start and the end · Introductions · Farewells · When several talk at once (in turn instead of the wide shot) | each on / off |
| Moves | ||
| Move | Soft pan 1.2 s · Brisk 0.7 s · Whip pan 0.35 s · No easing 1 s · Slow move 3 s · With swing (slight overshoot) 1 s | the profile of every camera move |
| Cut into motion | off · now and then · often | some cuts land in a pan still settling |
| Camera operator | off (tripod) · light · medium (handheld) · strong (shoulder camera) | sway, breathing and a slight tremor; cuts stay hard |
| Ease-in | like the move · none · light · medium · strong | how gently a move sets off |
| Ease-out | like the move · none (hard stop) · light · medium · strong (glide to a stop) | how a move rolls out |
| Own move | curve editor: two handles for setting off and braking, below the line winds up first, above it overshoots and settles · a ball shows the speed · templates crane · steadicam · whip pan with wobble · dolly · snappy · wind-up · save · delete | your own course of a move, written one to one as Fusion's Bezier handles; the faint curves show how far the variation may stray from it |
| Time | fixed (seconds) · by distance (seconds per 90°, shortest and longest) | a long pan takes longer than a short one |
| Also for push-ins | on / off | the slow zooms follow the same curve |
| Move to the wide shot | like the move · any profile · an own move | its own course for moves into the wide shot |
| Widen on pans | off · light (+6°) · clear (+12°) | on pans from 30°, the view opens halfway and closes on the person |
| Motion blur on pans | off · light · strong (like film) | a directional blur along the camera's way, as long as half a frame's travel (180° shutter) |
| Reactions | ||
| Show reactions | on / off · at most rarely (every 30 s) · normal (12 s) · often (6 s) | cuts briefly to whoever reacts |
| Facial expression | smile · surprise · doubtful · sad · angry · disgust · yawn · eyes closed · nod · head shake – each off · less · normal · more | sensitivity per kind |
| Sound | laugh · gasp · sigh · applause · crying · shouting · singing · whistling – each off · less · normal · more | sensitivity per kind |
| Emotion display | show briefly now and then · at most every 3 min / 90 s / 45 s · in and out: fade softly / hard · which of the 18 kinds | fine corner marks, label, value and the last seconds as a curve, following the face; only shortly before strong reactions while the camera is on that person |
| Effects (sparingly, only on the wide shot) | ||
| Intro | on / off | fly from the little planet into the wide shot (adapts to a short opening) |
| Outro | on / off | out into the little planet at the end |
| Look around | on / off · smooth in and out · cut in · now this, now that | now and then once around the room, the camera rising to each head; smooth sets off and brakes slowly, cut in cuts into the turn and out of it while it still turns |
4 · Cut
| Setting | Options | What it does |
|---|---|---|
| Start recording | R | overdub live with the keys; several passes add up |
| Follow track | F on / off | the preview shows the cut as Resolve will |
| Recording over replaces the suggestion | on / off | where you cut, your choice wins |
| Move with Shift+number | the six move profiles | profile for moves you trigger |
| Duration | as profile · 0.3 · 0.6 · 1 · 1.5 · 2 · 3 · 5 s | length of those moves |
| Effects U I O P | slow · normal · fast · lightning fast | speed of the effects you trigger |
| Camera track | range: whole clip · 10 min · 2 min · 30 s | click selects a change, drag its edges; per change: target, transition (hard cut or a move profile) and its length 0.3–5 s |
| Preview | ♪ Podcast / ♪ Camera · 360° overview (M) · track (T) · sidebar (B) · emotion track (G) · speed 1× · 1.5× · 2× · 3× · 4× · In / Out | listen to either sound, check the whole sphere, set the episode; the emotion track shows per person who speaks and which reaction, 15 s back and ahead |
5 · Picture
| Setting | Options | What it does |
|---|---|---|
| Look | Natural · Warm · Bright & airy · Dark & classy (muted) · Bold & colorful · Black & white | measured white balance and exposure plus the look, as a CDL on the first node |
| Contrast | soft · medium · hard | part of the look |
| Vignette | off · light · strong | an ellipse in the camera composition |
| Apply coloring · Neutral | buttons | puts the grade on or takes it off |
| Light flicker | off · fast · thorough (very slow) | ResolveFX Deflicker for 50/60 Hz lamps |
| Noise reduction | off · light (fast) · strong (dark rooms, slow) | ResolveFX Noise Reduction, temporal and spatial |
| Sharpness | off · natural (skin soft, details sharp) · strong (AI UltraSharpen) | on the finished frame |
| Match faces | on / off | evens out brightness and colour cast between people |
| Correct distortion in wide shots | on / off | straight lines in wide shots and two-shots |
| Short fade at In/Out | on / off | from and to black |
| Titles | ||
| Lower thirds | Introduce people when they first appear · 3 · 4 · 6 s | name and extra line in each person's first shot alone that is long enough, half a second after the camera arrived, apart from the other titles |
| Title and credits | title · second line · 3 · 4 · 6 · 8 s · credits, one line per line | the title from In, the credits up to Out (over the fade) |
| In Resolve | track “DM360 Titel” above the picture | each title a Fusion composition with Text+: text, font, colour and position changeable on the Fusion page, nothing to render again; off while the picture is finished, so it is not burnt in |
6 · Sound & Language
| Setting | Options | What it does |
|---|---|---|
| Podcast sound · Original sound | buttons | a processed copy replaces the original in Resolve; one click back |
| Remove reverb | automatic · measured (default) · normal 70 % · strong 90 % · maximum 100 % | Resolve's Voice Isolation on the clip; automatic measures the room's reverb time, mains hum (then filtered out) and the speakers' levels |
| Subtitles | English · German · French · Italian · Spanish · Portuguese · Dutch · Russian · Chinese (simplified) · Japanese · more languages by code (e.g. ko, pt-BR) | translated by Claude (Anthropic) from the whole conversation; Chinese, Japanese, Korean and Thai with short lines broken after punctuation |
| Dubbing | per person: built-in Female 1 / 2 · Male 1 / 2 · a voice model you trained in Resolve | Resolve's AI Speech Generator speaks the translation, one track per language |
| Audio version in Resolve | Original · Dub XX | which track plays |
7 · Output
| Setting | Options | What it does |
|---|---|---|
| Write to timeline | Replace · As new timeline · then finish the picture | writes the cut, both versions and every format chosen; with the slow tools on, finishes the picture right after |
| Overview | – | before writing: timeline, format, range, camera, reactions, fade, flicker, vignette, quality |
| Cut version: leave out | Noise reduction · Light flicker · Sharpness | what the light version skips |
| Finish picture | button, with progress and time left | the two passes described above |
| Export now | YouTube, social · MP4 (H.264) · Archive · ProRes 422 | choose where – the panel writes first if needed, switches to Final, finishes the picture when the slow tools are on, renders with sound and shows the file |
| Also export the highlight clips | on / off | after the episode, each highlight clip rendered anew from the 360° original in 9:16, on its own timeline “… · Clip 1 · 9:16” (replaced next time), with lower thirds, the files beside the episode |
| Upload the episode | don't upload · YouTube · Vimeo | Resolve renders with its Quick Export for the platform and uploads through the account under Preferences › Internet Accounts |
| Upload the clips | don't upload · YouTube (Shorts) · TikTok | the same for every highlight clip; title, description and visibility follow Resolve's account settings |
| View a sample | 5 s from the playhead | rendered by Resolve in full quality and played in the panel |
| Finishing report | – | what “automatic” measured and set for picture and sound, and why |
| Highlight clips | Find highlights · View · Write 9:16 | the best 45-second stretches by reactions and lively talk, out to whole sentences, each on its own 9:16 timeline from the 360° original, with lower thirds |
| Thumbnail proposals | 3 pictures | the strongest moments of the emotion track, framed as in the episode, as JPG |
| Chapters for YouTube | Make chapters · Copy | by Claude from the transcript (your Anthropic key), from 0:00, ready for the description |
Processing chains
Exactly what happens to picture and sound
| Reframing | DM360 Reframe (Fusion fuse, GPU): equirectangular, GoPro EAC and Insta360 dual fisheye read directly; rectilinear view with wide-angle correction for wide shots and two-shots; little planet and roll as effects. |
| Picture order | reframe → match faces → deflicker → noise reduction → sharpness → vignette → fade → emotion display |
| Speed | Final with deflicker and noise reduction on the 360° source: about 0.5 fps. Finish picture runs them on the cut instead: about 12 fps (measured on our test computer; depends on your GPU). |
| Podcast sound | clipped peaks rebuilt (declip) · rumble filter 80 Hz · light broadband noise reduction · warmth +3 dB at 180 Hz · less boxiness −1.5 dB at 450 Hz · less harshness −2.5 dB at 3.5 kHz · air −3 dB above 8 kHz · de-esser · compressor 3:1 · loudness −16 LUFS, −1.5 dBTP · 48 kHz 24-bit WAV · then Voice Isolation and Dialogue Leveler in Resolve |
| Grade | white balance from grey areas only (not the whole sphere), exposure mostly by the measured faces, look and contrast as ASC CDL |
Keyboard
Every shortcut
| 1–9 | cut to a person |
| Shift+number | move to them |
| Alt+number | their medium shot |
| 0 | wide shot |
| U · I · O · P | upside down (again: back) · roll · 360° spin through the room · little planet (drag to turn it) |
| Q / W · E | zoom in / out (hold) · reset zoom |
| R | recording on / off |
| Space | play / pause |
| J · K · L | −10 s · stop (speed 1×) · play, L again: faster |
| Home · End | to In / Out, again: start / end |
| Alt+I / O / X | set In · set Out · clear In and Out |
| Page ↑ · Page ↓ | 1 minute back / forward |
| [ · ] | selected change ±1 frame (Shift ±10) |
| Del | delete the selected change |
| Ctrl+Z | undo – every change |
| M · T · B · G | 360° overview · camera track · sidebar · emotion track on / off |
| F | follow track on / off |
| Ctrl+number | store the current view on that key (Calibrate) |
| , · . | level the horizon (Shift: finer) |
Licence, updates, privacy
The fine print, in plain words
| Licence | $99 once. One key per computer, for all 1.x versions, for life. Free this computer to move it. |
| Free trial | everything works; writing to the timeline and finishing the picture cover the first two minutes of an episode |
| Licence check | through Polar every 14 days, 30 days offline; on activation it sends the key, a one-way hash of the computer's ID, the computer's name and the system (to tell your computers apart in your Polar account) and the plugin version |
| Updates | automatic within the major version: downloaded in the background, Ed25519 signature checked, used from the next start; the previous version stays as a fallback |
| Your data | recordings, faces, voices and analysis stay on your computer. Translation sends the conversation's text to Anthropic, only when you press Translate. The plugin has no tracking and needs no account. |
| Files | plugin, starter and camera tool – the paths are in the installation guide |
| Panel languages | English, German |