By Balance IT · Updated
A clearer Studio: Sound, Look, Output
- Three primary steps lead from choosing sound to a visual world and a projector or music video. Look and Output are separate sections instead of nested inside Layers.
- Layers & text groups video layers, titles, soundwave and the world mixer behind named disclosures. Effects states whether the rack changes Program or a selected mapping surface.
- Worlds, Video library and Show open their actual tools. Projector names its destination; session import sits beside session save. A section selector stays available on compact screens and in Stage view.
- Changing sections preserves the actual output and transport, and keeps the selected section in the URL. Mobile controls use page scrolling instead of a second long scroll area.
Animate Program and surface effects with their own beat loops
- Each Kaleidoscope, Pixelation, RGB split and FX amount target can choose a beat loop independently from its audio-band or manual driver. Loops support bounded beat/bar lengths, phase offsets and Sine, Triangle, Ramp, Alternating and deterministic Step random shapes.
- Program and individual mapping surfaces share the visual timing grid. Track sync retains file-grid phase through seeks and trimmed films; the manual film grid starts at zero. Gentler motion makes loops four times slower, while Freeze and Blackout cover the final picture.
- Sessions, maps, captured compositions, surface acts and draft editors preserve their respective loops. Bypass retains routes; Manual, Clear links and Reset have distinct recovery behavior.
- The live guide owns the complete workflow, with readable Markdown, related film/music instructions and read-only MCP facts.
- Captured-act editors accept valid fractional look values such as 0.012 grain without an unrelated browser step mismatch blocking Save. Draft Save and Undo keep live Program and calibrated corners unchanged.
Keep the readable format guide tied to its canonical page
- The aspect-ratio guide’s Markdown response now sends its canonical HTML URL in an HTTP Link header, matching the other operational guides. This closes the live metadata check found during the Stage workspace release.
- The SEO publication gate now requires that response rule for the ratio guide. Stage workspace behavior and rendered output are unchanged.
Give your live picture room: Stage view and an adjustable workspace
- A taller desktop preview and Stage view keep the final picture in focus. Hide controls to reclaim the canvas area; transport, Freeze, Blackout, recording and Go live stay available. Return to Studio or press Escape to restore the full workspace.
- Resize the inspector with a draggable, keyboard-accessible divider. Width is bounded to preserve preview space; the preferred width returns when a window grows. Workspace preferences are local UI settings, separate from sessions, mapping, audio and film size.
- Use Workspace to reach worlds, the sequence editor and performance pads, or reset the layout. Explicit inspector navigation reveals hidden controls; mapping handles follow the fitted image.
- The live guide explains preparation, Stage view, mapping, recovery and output checks, with a first-party layout comparison and readable public documentation.
Drive each effect from its own audio band on Program and mapped surfaces
- Audio links independently connect Bass, Mid, High, Energy or the detected Transient to Kaleidoscope, Pixelation, RGB split and FX amount. Each connection has quiet/peak ranges, gain, a noise floor and Linear/Smooth/Punch response, including reversed ranges.
- Program and selected-surface editors show the actual resolved values and transfer curves. Manual, Clear links, Bypass and Reset have distinct behavior. Mapping Undo, sessions, maps, captured compositions and act drafts retain the respective settings.
- Live output, source/act previews, A/B composition, mapped transitions and causal file-render frames share the transfer rule. Legacy energy-only settings remain compatible; linked peaks are included during effect preparation.
- The live-performance guide owns the complete setup/recovery workflow; a Journal exercise, related film guides, readable Markdown and public MCP facts link to it.
Compose an audio spectrum, ring or waveform over your music visuals
- Soundwave opens a complete Program layer with responsive frequency bars, a radial spectrum and a time-domain waveform. Style, position, color, size, response, opacity and a dark background blend can be changed without replacing audio or starting sound.
- The same layer participates in composed Studio output, projector, Freeze/Blackout, real-time recording and causally analyzed file-render reviews and films at each actual output size. File Review freezes its settings with titles and copied artwork.
- Saved local/settings-only sessions retain the spectrum layer; legacy sessions default to disabled. Shared procedural-look links and independent mapped world sources remain separate.
- A dedicated spectrum-to-branded-video guide supplies first-party output examples, actual instructions, readable Markdown, discovery/MCP facts and links from relevant existing pages.
Place the track beat grid and keep its phase in every rendered passage
- Align file beat grid adds a local, silent 2–32 second waveform editor: beat-one placement, optional nearby-peak snap, exact seconds, close-up windows, 1/10 ms nudges, bar-reference movement and later-section drift inspection.
- Deliberate Apply sets BPM, exact file alignment and Track sync. Guarded Undo restores the previous tempo, phase and source without starting sound or moving the cue grid, transport, mapping or pulse delay. Find tempo hands its proposed BPM into this editor.
- Track sync now shares the live pulse envelope and absolute soundtrack alignment with first-frame review, causal storyboards and encoded file renders, including trimmed starts. A different file cannot borrow the loaded track’s grid.
- The music-file guide owns the full waveform-to-film instructions, diagnostic table and first-party evidence, with readable Markdown, task finder, MCP capability and related guide links.
Keep deployed security policies within Cloudflare header limits
- Each HTML document now receives CSP hashes for its own inline scripts instead of one growing site-wide hash list. Studio retains only its existing local Room connector destination; external framing remains limited to the embed player.
- The build rejects any _headers line over 2000 characters or more than 100 rules, preventing an oversized policy from silently disappearing on Cloudflare.
Make the BPM finder discoverable from the homepage
- The generated homepage learning navigation now links directly to the free MP3/WAV BPM finder. Its source template owns the link so subsequent builds retain it.
- Repeated site builds replace the current MCP tempo section instead of accumulating duplicate capability copy.
Find file tempo locally, compare pulse readings and prepare Studio deliberately
- A free standalone MP3/WAV BPM finder decodes a chosen 8–60 second section locally and silently, ranks periodic pulse candidates, shows transient strength and offers Club/Wide/Slow/Fast ranges plus explicit half/double and typed corrections. No audio upload, account or new dependency.
- Find tempo from loaded track in Studio reuses the current file. Apply tempo changes only BPM on an idle performance; Undo restores the previous tempo while its file/settings remain current. Beat-one alignment, pulse source, audio speed and monitoring remain separate.
- An explicit same-origin handoff brings the chosen track to a new Studio tab paused with monitoring off; the proposed tempo opens for review and is never applied by the link alone. The public tool, guide entries, readable Markdown and discovery copy explain the method, measurement and limitations.
Find the next practical step in guides, Studio and MCP
- Find a task searches 28 curated workflows by job, control, source or output. The existing guide hub keeps every ordered instruction, boundary and canonical owner link readable in HTML without JavaScript; its new Markdown companion contains the same workflows. Filtering runs locally without sending or storing typed searches.
- Help with this workspace opens the same task finder inside Studio, with Mapping, Audio and Effects context. Public help links open an exact workflow review without replacing settings, starting audio or requesting permissions. A separate action shows the relevant controls; opening File render retains its existing pause-audio behavior. Keyboard shortcuts remain available.
- search_docs and get_doc are two new read-only MCP tools for the same curated instructions, stable IDs, exact control locations and public guide links. Unknown or private/remote-control arguments are rejected. These tools do not inspect projects or operate Studio.
- Two fixed aggregate Studio milestones distinguish help opening and a deliberate jump to controls, without searches, topic IDs, private files or visitor identifiers. Known local/qa=1 tests are excluded.
Share a tuned procedural look, review it off air and take it deliberately
- Share procedural look captures the current Program world, all ten numeric source controls and canvas dimensions in a public review link. The bounded moving preview shows the clean source without starting sound. The visible link remains copyable when clipboard access is unavailable.
- Opening a shared link preserves the current Studio. Review, Apply look to Program and optional canvas adoption are separate choices. Undo restores the prior source/frame until newer edits, retaining current Gentler motion, Freeze and Blackout. Audio, media, calibrated mapping, titles, master FX, cue contents and MIDI bindings stay separate.
- The music-visualizer guide and scene catalog offer three authored starting directions, the exact sharing workflow and limitations in HTML and Markdown. create_look_link is the eleventh read-only MCP tool and returns the same validated public review recipe without private Studio or source access.
Place each act on the soundtrack, then render the complete film
- Soundtrack entry times gives every queued act an editable absolute music-file entrance. Place here reads the separate PCM preview playhead; each incoming transition begins on the nearest chosen film frame. The cue ruler shows the resulting arrangement without rewriting live holds.
- Seek preview prepares the passage around an act, and a movable preview window allows later parts of a long file to be inspected in bounded sixty-second PCM windows. Monitoring stays separate and off by default. Placing an entry while previewing keeps that transport running and invalidates old film Review.
- Unordered entries, bounds, transition overlap and holds shorter than 0.25 seconds block Review. Changing file, passage, FPS or show retains the old draft visibly as stale until Reset entries to fit all acts. The reviewed schedule uses the existing causal storyboard and serial landscape, portrait and square renderer.
- The film guide, eight-act article, HTML/Markdown discovery and read-only MCP explain the same cue-sheet workflow and boundaries.
An eight-act techno set, from an original Journal recipe to your own soundtrack film
- A complete eight-act techno recipe curates eight existing worlds with distinct movement, energy, detail, glow and transitions. Manual holds leave musical changes under the operator’s control; the shared starting grid is 150 BPM.
- Journal show exercises now offer the same explicit rehearsal path as a planned show: choose a local MP3/WAV paused with monitoring off, prepare, rehearse and open the complete queue with that File in film planning. Load and Undo remain deliberate.
- The original Journal article explains each visual role, timing choices, one-projector use and the complete film journey. Exact public settings, HTML, Markdown, internal discovery and read-only MCP reference the same recipe with accurate individual publication dates.
Keep read-only AI capabilities in step with the soundtrack storyboard
- The existing get_capabilities response now describes film-only act fitting, coverage approval, causal later-frame Review and their limits, with the same canonical instructions used by HTML and Markdown. A consistency check prevents the API catalog from omitting this workflow.
Fit every act to your soundtrack, review later pictures, render the complete film
- A loaded planned show can open File render with the same local MP3/WAV and full sequence. The explicit action stops rehearsal playback without starting a film or replacing live cues.
- Fit all acts to selected passage reserves the existing transitions and creates editable film-only holds. Keep live cue timings remains available; missing acts, clipped holds/transitions and an extended final act require explicit coverage approval before Review.
- Sequence Review causally replays every frame of the first selected output at film FPS, then shows sampled hold, transition and final pictures with soundtrack times. The frozen reviewed timeline feeds the existing serial 16:9, 9:16 and 1:1 delivery set, including Cancel/Continue and protected completed films.
- The existing YouTube and live-show guides explain this full workflow in HTML and Markdown; public AI/MCP capabilities describe the same controls and boundaries.
One planned show, the same local soundtrack, a guided rehearsal
- The homepage live-show builder can pass its selected MP3/WAV directly into Studio while preserving the complete public act recipe. Receipt is explicit, paused and with speaker monitoring off; Load planned show and its settings-only Undo remain separate actions.
- A rehearsal guide beside the planned-show review leads through Prepare show, deliberate soundtrack start, sequence playback and a short real-time recording for inspection. It reuses the existing scene engine, cue transport and recorder; edits to acts, calibrated geometry, frame or media attachment require fresh preparation.
- The live-visuals guide explains this complete workflow, its private file boundary, manual file fallback and actual saved-take check in HTML and Markdown.
Keep the prepared soundtrack waveform during first-frame Review
- Preparing target pictures now stops the separate soundtrack preview without clearing an already prepared waveform. Review still pauses Studio and holds Program; starting the actual encoder retains its existing memory-release behavior. A stateful panel regression check fails against the earlier GPU lock and passes with this correction.
Keep the delivery evidence as a referenced resource, with noindex headers
- The multi-format delivery report now receives the same JSON content type and noindex/follow header as the earlier soundtrack and Journal reports. The guide remains the indexable workflow owner. Local SEO and deployed HTTP checks now require those headers for JSON evidence resources.
One soundtrack, a reviewed set of landscape, portrait and square films
- File render now prepares up to three delivery targets from one local MP3/WAV passage and reviewed composition. Keep Current Studio frame or choose 16:9, 9:16 and 1:1 with a 540, 720 or 1080-pixel short edge. Each target checks its actual encoder settings before rendering; Program dimensions stay in place.
- Review renders a real first-frame still separately at every selected target size, with the captured title and independent artwork. The same compositor supplies clips, Fit/Fill, master effects, mapped sources and act transitions in the film. A still is explicitly not a motion preview. Relevant changes require another review.
- The delivery queue runs one GPU job at a time and retains each completed film. Cancel stops the unfinished target; Continue missing targets skips completed results. Each output has its own download, inspection, saved confirmation and protected discard within one shared 750 MB retained-result budget. The existing YouTube guide owns the complete HTML/Markdown workflow, with matching ratio guidance, product facts and read-only MCP boundaries.
Read the method, load its exact recipe: four practical Journal workshops
- The four existing Journal articles now include a shared public recipe, direct exercise review, three practical steps and criteria for evaluating the result. Contrast carries Space, Pressure, Release and Return into four named manual acts; the one-projector study assigns a focal sculpture and two distinct supporting sources; the audio-selection study prepares its actual A/B candidates; portrait composition opens the same world on three actual canvas shapes.
- Opening a workshop does not replace Studio settings or start sound. Load exercise applies the reviewed settings with current-tab Undo exercise. The catalog separately reviews and explicitly loads its two comparison slots, with Undo A/B exercise; its current local track, selected passage and speaker preference stay separate. Stage and framing exercises are genuine single setups with no fabricated second act.
- Each workshop can carry its full known public recipe and chosen frame into the existing custom-work brief, review, saved draft and reopen workflow. No private media, passages or calibrated venue settings enter the reference. Article HTML, Markdown, publication/modification dates, discovery facts and the existing ten read-only MCP tools describe the same exercises; existing titles and canonical URLs remain.
Choose a soundtrack passage once, preview its decoded audio and review the same cut for your film
- The catalog can explicitly propose its selected passage when transferring the same local MP3/WAV to Studio. The option starts off and sends no passage in the public URL. File render shows Use proposed fragment and Keep current range; nothing changes the film trim until the proposal is accepted and the timeline is reviewed.
- File render adds a waveform, seekable soundtrack preview, Stop and Whole track. Preview uses up to the first sixty seconds of the selected range without shortening the film. Hear soundtrack preview starts off; choosing a file, changing the cut, closing/hiding the panel and preparing a render stop the separate preview. Existing Program, clips, mappings and completed output remain in place.
- Soundtrack preview and export share one streaming content timeline. Verified, recognized MP3 gapless metadata removes encoder priming and padding; unknown or inconsistent metadata retains an explicitly labeled encoded timeline. This resolves the evidenced tagged-MP3 cut discrepancy without applying a fixed millisecond correction to every track. The existing YouTube workflow owns the HTML/Markdown instructions, with aligned catalog guidance, privacy and public-only MCP facts.
Audition the world catalog on your own music, then take your choice into Studio
- The 108-world catalog now searches and filters the same results in its rail and cards. Choose one local MP3/WAV up to 200 MB, validate it silently and use the shared single-canvas preview with explicit playback, seek, frame and motion controls. Original test WAV uses the same twelve-second source as File render; speaker monitoring defaults off.
- Assign two worlds to A/B and explicitly preview each with the same selected 0.25–60-second soundtrack passage. Each audition stops at the first callback at or after its end. Existing Favorites carry candidates into Studio without media or replacement of saved banks; failed storage remains an explicitly temporary shortlist.
- Continue with the chosen world, canvas, motion and same in-memory File through the existing one-use same-origin Studio receipt. The soundtrack arrives paused and File render still requires its own segment/timeline review. Catalog HTML/Markdown owns the complete audition workflow, with matching music-guide cross-link, public facts, privacy and read-only MCP boundaries. Fixed QA-excluded counters distinguish local load, an actually rendered comparison and a saved favorite.
Your own track on the homepage: local preview to a paused Studio soundtrack
- The central homepage composer now accepts one local MP3/WAV up to the existing 200 MB file-render limit. Validate the actual audio before graphics, play/pause/seek explicitly and compare worlds or landscape/portrait/square framing without selecting the track again. Preview monitoring starts off; Hear preview audio is an independent deliberate choice. The music-video test source now previews the same original twelve-second WAV offered by File render.
- Continue with my track in Studio opens a same-origin tab. Receive selected track explicitly requests a one-use local File transfer from the original page. Studio verifies actual loading before acknowledgement, keeps audio paused with monitoring off, and opens the existing film review/render path. Media, filenames and transfer nonces never enter public setup links, session files or measurement fields. Popup, timeout, cancel, rejected-file and page-lifecycle failures retain a manual-file recovery route.
- The existing music guide owns the complete local-track workflow in HTML/Markdown, with a short YouTube cross-link, aligned privacy/facts and public read-only MCP instructions. Four fixed, QA-excluded action counters distinguish local load, actual local preview, successful Studio receipt and failed receipt; they remain independent actions rather than unique visitors, linked conversions or completed films.
Room setup to daily operation: a complete one-projector commissioning workspace
- Room control now separates Operate room from Set up room. Daily mode/brightness/Freeze/Blackout controls stay accessible, while setup prepares exact Node launch, Loxone HTTP ON/OFF, status and Program 16 wiring from validated private IPs and a safe absolute key-file path. These inputs stay only in the open panel; no scanner, installer, automatic audio or configuration replacement is introduced.
- Start new check records eleven actual acknowledged commands after the check begins: three modes, brightness 0/50/100, Freeze and Blackout on/off, and Stop. Accepted queue requests, local button clicks, stale/duplicate receipts and rejected commands do not fabricate checks. Browser history, present connection and manual equipment declarations remain separate. Copy handover excludes addresses, paths, keys, private names, calibration and receipt IDs.
- The existing live guide supplies the complete workflow in HTML/Markdown and links an updated eleven-page Polish 1.53 setup and operation PDF. A public-only one-projector reference opens the existing installation enquiry without exporting private setup data. Fixed anonymous Studio counters track requested setup actions and real connection/Arm/ACK milestones; native QA is excluded.
Loxone room control: saved venue modes, a local connector and a Program function block
- Room control adds captured Ambient, Party and Techno procedural looks, picture brightness, Freeze, Blackout, projector/mapping handoffs and a selected connection-loss policy. Modes keep the current calibrated geometry, masks, crops, opacity, blend, visibility, surface order and output dimensions; changed surface IDs reject a recall. Save session retains room modes and settings without keys or connection state.
- A dedicated Node 22 connector links Loxone Virtual Output HTTP commands to one explicitly paired and armed Studio. Separate loopback/LAN listeners, Miniserver IP restriction and independent Studio/control/status keys keep the channel local. Persistent operator-owned keys survive connector restarts. Bounded queues, expiring commands and browser acknowledgements distinguish acceptance from actual application; reconnect never replays stale commands or arms itself.
- Supplied Program 16 PicoC logic sends mode pulses, bounded brightness, Blackout/Freeze levels and priority Stop through configured Virtual Outputs. Ready is derived from read-only Basic HTTP status. The existing live guide supplies exact fields, wiring and commissioning checks in HTML/Markdown, with discoverable product facts and privacy details.
Correct the Nexus homepage catalog link
- The homepage guide link now displays the actual number of Nexus worlds from the shared catalog. The generator evaluates its count at every build, correcting a literal template expression that production verification caught. Surface Look Studio and its accepted behavior remain as released in 1.51.0.
Surface Look Studio: rehearse a world and its FX before taking it to one mapped surface
- Mapping now opens the complete 108-world library for a named surface. Browse & tune world copies its source look and surface FX into Surface Look Studio. Search, collections, favorites and banks help find a world while a bounded muted source preview shows its motion and effects off air.
- Tune palette, energy, motion, detail, glow and scale alongside kaleidoscope, pixelation, RGB split, FX amount, audio linking and bypass. Reset draft restores the complete opening look; Cancel and Escape discard edits. Take to the named surface deliberately replaces only its world, look and FX, preserving alignment, crop, masks, opacity, blend, order and every other surface. One Mapping Undo restores the previous take.
- Take checks the current destination and performance state again. Removed or changed targets, active recording/rendering, cue preparation/playback, transitions and A/B mixing block application. A failed preview retains a clearly labelled still and leaves output running. Closing releases preview contexts; recording stops its GPU work. Its lower frame rate retains the intended Motion tempo.
- The existing live guide, one-projector Journal article, product facts, Markdown, llms and read-only MCP explain this same workflow and its limits. Applied looks remain in existing map/session/cue persistence and mapped recording; no new public route or MCP control tool is introduced.
A larger creative instrument: 36 Nexus originals, a complete world browser and the Studio Journal
- Browse worlds opens a full visual library with readable scene descriptions, suggested uses, search, families, curated collections, favorites, set banks and recent choices. Inspecting a world leaves the current picture and audio alone; Use world deliberately applies it while retaining your output frame. The compact performance strip remains available.
- Thirty-six original Nexus worlds expand the collection from 72 to 108, spanning spatial topology, optical deformation and kinetic light. Their stable IDs work across Studio, mapped surfaces, share links, show cues and the read-only MCP catalog. Existing worlds and their saved IDs remain intact.
- The new Journal contains four first-party articles on set contrast, designing for one projector, choosing audio-reactive scenes and composing portrait visuals. Each includes practical exercises, labelled visual references, direct Studio links and related technical guides. HTML, Markdown, canonical metadata and the sitemap share one source.
- The homepage, visual catalog and guides connect the Journal to the free Studio. Newest-world filters and collection counts now follow the current catalog. No offline application download or installation is introduced.
From a selected look to a complete first film, with a recoverable four-act example
- Music-video entries from the homepage, scene catalogue, comparison tool and guides open the frame-by-frame renderer with their selected world, frame and output settings. Existing Studio-file Make a video carries the same local File into the renderer silently; selecting a soundtrack again is unnecessary. Live and performed realtime workflows stay separate. A public selected-world entry starts without stale detached video layers; saved library entries remain in Assets for deliberate recovery.
- Load four-act example prepares the complete first-party session, both original muted loops and a twelve-second WAV before applying anything. It replaces worlds, mapping, acts, Clip Deck, title and tempo explicitly, retaining the receiving output frame, size and motion setting. Nothing loads, plays or renders merely by following the example link.
- Undo restores the prior setup and loaded clip URLs/decoders paused, without choosing the clips again. Missing downloads and Cancel retain the previous setup, soundtrack and finished exports. Use original twelve-second test WAV keeps the current composition. Review is required before rendering.
- The existing YouTube guide owns the complete first-film instructions in HTML and Markdown. Related discovery paths, comparison facts and read-only MCP point to the same practical workflow, with current limits and source evidence. No new thin indexed pages or private media in public links.
Render a complete composed film from local MP3 or WAV, one frame at a time
- File render advances a separate film clock, decodes a local MP3/WAV silently, renders each planned picture and waits for the encoder. A slow computer changes processing time, rather than skipping film frames. The current composed look or a reviewed act sequence includes both trimmed video layers, rate/loop/playback intent, blends/order, Program FX, current mapping geometry, independent surface looks/FX and cut/dissolve/wipe/dip transitions.
- Review the selected soundtrack segment, Studio output dimensions, 24/30/60 FPS, 6/12/24 Mbps, browser-tested MP4/WebM and full act timeline. Manual holds require explicit seconds. Bars round their hold end to the next manual-BPM bar boundary after the incoming transition; the first act starts with a cut and the last holds to the selected audio end.
- Progress distinguishes film time from processing time. Cancel, missing media, stale review and unsupported resources retain Program and prior exports. Successful replacement requires explicit confirmation for an unsaved prior file render. The soundtrack is not monitored; Studio playback remains paused until resumed deliberately. Download and reopen the actual file through the existing local recording inspector.
- A synthetic two-surface, four-act session, separate original sources, a real rendered output and measured decode evidence accompany the YouTube guide in HTML/Markdown. Public facts, privacy and read-only MCP expose the same bounded workflow. Fixed anonymous milestones exclude sound files, names, geometry, frames and private settings.
Transfer complete acts between sets without replacing Program
- Act Packs exports selected complete cues in queue order to local JSON, with a visible save link, generated JSON and clipboard fallback. World controls, clip assignments/playback intent/trims/blends/order, Program FX, captured surface looks/crops/FX and cue timing move together. Media bytes and venue geometry remain separate.
- Incoming packs are reviewed before append. Compatible media entries can be reused or admitted as detached descriptors with fresh IDs. Surface looks require deliberate binding to current surfaces or explicit exclusion. Capacity checks and stale-review protection preserve existing acts, loaded files, Program, audio, calibration, BPM, MIDI and output options.
- Undo last pack import removes its unchanged acts and unused detached entries while retaining loaded or still-used media. Imported acts can be inspected in the complete off-air editor, recovered through Assets, prepared and fired deliberately.
- Original four-act packs, a real generated export and matching live-guide HTML/Markdown demonstrate this local workflow. Public product facts, privacy and read-only MCP describe the same bounded capability; fixed anonymous milestones exclude pack contents, names, media and mappings.
Edit a complete saved act off air, then perform it deliberately
- Edit cue opens an isolated complete-act workspace for the Program world and controls, two muted video layers, trim, playback intent, blend, fit, flip, speed, order, Program effects and captured surface looks, crops and effects. Current calibrated corners, masks and feather stay intact.
- Preview draft explicitly prepares borrowed media and a separate low-resolution moving composition. It shares existing audio analysis without starting music or requesting an input. Editing releases its preview resources; Cancel discards the draft. Save updates the stored cue without taking it onto Program.
- Undo saved edit restores one complete saved revision in this tab while its cue still matches. Save is blocked during recording, file loading, pending launches, transitions and a running sequence. Missing dependencies remain visible; save can retain them, while preview and preparation require usable original media.
- Legacy cues keep their original live clip/FX behavior. Explicit Enable clips & FX starts an empty complete composition; Copy current Program into draft requires a later Save. Session recovery retains edited settings and dependencies without media bytes.
- The existing live guide, HTML/Markdown product facts, privacy and read-only MCP describe the same workflow. First-party before/after sessions and a checked short performance capture make the saved revision replayable. Fixed action counts contain no private drafts, clip names, calibration or settings.
Prepare, recover and perform a complete local media set
- Assets now supplies a dedicated video workspace: one muted off-air preview with seek and loop, searchable display names and bins, attached/missing/used-by-acts filters, sorting and default trim. Save clip details changes future Take actions while current layers and captured acts keep their own trim.
- Batch import reports added, reattached, already-loaded and rejected files. Cancel retains admitted clips. Explicit original-file reattachment validates the selected missing descriptor. Detach releases media while retaining assignments and act dependencies; Remove is blocked for captured dependencies. Pending launches and transitions lock editing.
- Save complete session and local Media checklist retain organizational metadata and dependencies without media bytes. Reopen restores missing placeholders and stopped transports; reattach paused, Prepare show, then open performance pads. A public first-party recovery kit and matching HTML/Markdown instructions cover the complete path.
- Speakers on/off independently disables this Studio monitor output for files and demo audio while keeping analysis and recorded-audio level separate. The preference stays in this browser and is not changed by session import. Microphone and shared-tab inputs remain unmonitored to avoid feedback or duplicate playback.
- Public capabilities, read-only MCP, product facts, comparison and privacy describe the same bounded workflow. Three fixed action counts contain no media names, bins, trim, private session contents or visitor identifiers.
Perform complete acts from pads on a deliberate beat grid
- A dedicated performance workspace supplies eight large cue pads per bank for all 32 captured acts, current and preparing/queued states, final mapped Program monitoring, Run/Next/Stop, Cancel, Freeze and Blackout. Audio Play/Pause and record/speaker level remain reachable inside the workspace, with a visible muted-recording state. Numeric 1–8 and T/R/H/B shortcuts avoid typed input; the final monitor copies one output at up to 10 fps, without another renderer.
- Now retains the immediate workflow. Next beat, Next bar and Every 4 bars prepare complete captured cues off air against the manual 4/4 BPM/Resync grid. Ready clips wait paused; a later request replaces an earlier one. Late preparation or a callback more than 100 ms late waits for a future boundary, with no catch-up burst. Cancel preserves the current picture; Freeze and hidden-tab return re-arm rather than firing stale deadlines.
- Cue/session restoration retains launch mode, Program clips, playback intent, master FX and surface looks while keeping calibrated geometry. Media bytes remain separate and restored transports stay stopped. Request and completed-draw diagnostics expose their separate delay and missed-boundary count; automatic sequence holds keep their prior transition-then-hold behavior.
- The existing live guide documents the complete workflow in HTML and Markdown and includes an original downloaded/reopened pad rehearsal, actual session and timing report, plus a click-to-load Studio path. Public capabilities, MCP, discovery information and privacy explain the same bounded behavior. Separate fixed-label counts never transmit cue, media or timing details.
Turn a visual direction into a complete custom-project scope
- The custom-work page now takes a release, show or installation from a real product direction into a reviewed enquiry: deliverables, duration, timing, budget/currency, up to eight physical output rows, pixel dimensions, playback/audio and visual requirements. It identifies unresolved scope and hardware questions before any quote or booking.
- Homepage projects carry their selected world or complete 2–8-act procedural show into the brief, including per-surface FX, holds, transitions, tempo and A/B banks. Studio carries a starting world and actual canvas, including custom sizes; the comparison chooser carries requirements as feasibility questions. These validated public references use link fragments and exclude private media, titles and calibrated venue geometry.
- Download an editable local JSON draft and reopen it without losing the reference or entered fields. Invalid or oversized imports leave current details intact. Copy or download the complete reviewed TXT; short briefs can open a matching email draft, while long briefs retain copy/download without a truncating mailto link. The page sends no enquiry automatically.
- Original recordings and world references lead into all three project paths. The canonical HTML, matching Markdown, Service data, discovery directory and privacy explanation describe the same scoped custom work, with public Studio tools remaining free. Separate fixed-label action counts include no draft content or visitor identifiers.
Record on one clock, inspect the actual file, and check the saved show
- New timestamped capture uses the existing audio graph clock for video frames and stereo audio samples, with a same-origin AudioWorklet and browser-native WebCodecs. WebM VP9/VP8 with Opus and MP4 H.264/AAC are probed at the chosen size; browser capture remains available for compatibility. Frame-rate limiting and bounded encoder queues preserve actual gaps instead of squeezing a delayed picture into an earlier timestamp.
- Capture now inspects its own finalized MP4/WebM locally. Reopen a saved video, read codecs, dimensions, channels, sample rate, packet counts, track starts, available ends and the largest picture gap. Missing final packet duration is labeled as an unknown tail, not a measured end difference. Play the start, middle and end, and download a timing report before confirming that the video is saved. Invalid or unsupported files remain an explicit failed inspection.
- Recording preparation can be canceled; audio and video finalization has a watchdog, old download URLs survive until replacement, and file writing handles positional container updates using Blob chunks. The third-party dependency is unmodified, with its exact source archive and MPL 2.0 notice available independently of the private application repository.
- The existing YouTube, live and troubleshooting guides explain the complete take → local inspection → playback → download → reopen → saved confirmation flow in HTML and Markdown. Measurements distinguish preview draws, encoded packet timelines and content-based A/V checks.
Build a complete live show from the homepage
- A live set now builds a complete 2–8-act show in the central homepage composer: worlds, per-surface FX, palettes, Manual/Seconds/Bars holds, Cut/Dissolve/Dip/Wipe transitions, manual BPM and distinct A/B banks. Choose Screen, Triptych, Corner or Booth for one projector, Program or independent sources, demo/file/live input, four frame shapes and motion.
- Start from Warehouse pressure, Living metal or Optical drift; download their native Session JSON or customize the whole set. A fourth library entry opens the original Layers in motion clip kit with its real recorded rehearsal. Selecting any of the 72 catalog worlds can carry it into the Drop act of a complete editable show.
- Copy and reopen the exact settings plan or download a full native Session JSON. Studio keeps the current setup until deliberate Load planned show, with its existing Undo and Prepare show controls. The read-only MCP now exposes ten tools: list_show_starters discovers all four entries, and plan_mapped_show returns a matching homepage review link. HTML, Markdown and public capability descriptions explain the complete journey.
- Separate aggregate labels count homepage show preparation, requested session downloads, recipe copies, Studio handoffs and post-load audio activation. Only fixed event/page labels, release and UTC day are stored; no recipes, files or visitor identifiers.
Capture and replay complete VJ acts with local clips and effects
- Add current look and Recapture now store both Program video layers, trim, rate, loop, opacity, blend, framing, flip, order, requested playback and the Program FX rack alongside procedural and surface looks. Incoming playing clips restart at In; paused clips stay paused. Legacy cues retain their existing live-layer behavior until recaptured.
- Incoming clip decoders prepare off air before a whole cue switches. Prepare show checks every cue's files and warms its worlds before rehearsal. Missing or unusable files stop recall and preserve the current output. Dissolve, Dip and Wipe blend complete compositions; incoming decoders transfer without rewinding. A bounded GPU pass combines both clip stacks, racks and the transition for an unchanged Program screen or unmapped output. Other maps retain the general compositor. Freeze and Blackout remain global.
- Load an original four-act rehearsal kit directly in Studio after one explicit click: two paused source loops, full session JSON, separate test WAV and a real 960 × 540 VP9/Opus recording. The existing live and recording guides explain capture, preparation, replay, session restore and reattachment; public capability descriptions and MCP guidance match the workflow.
Refresh guide styles when their content changes
- Guide pages now reference editorial.css with a content revision. Returning visitors receive updated show-builder proportions immediately rather than retaining the previous cached stylesheet for up to four hours. Unchanged styles keep the same revision. The complete show workflow introduced in 1.39.0 is unchanged.
Keep show-builder scene references in their original proportions
- The complete show builder now displays its three scene-reference thumbnails at their natural 16:9 proportion. HTML image dimensions no longer leave 540-pixel-tall images in narrow columns, so surface FX and timing controls remain closer to the selected act. The complete 2–8-act show, native sessions, MCP planning, prepared worlds and scheduled-transition clock from 1.39.0 remain available.
Build your own complete mapped show, from planner or MCP
- The existing live-guide planner now builds 2–8-act shows for Screen, Triptych, Corner or Booth. Choose Program or independent worlds, individual surface FX, act palettes, Manual/Seconds/Bars holds, Cut/Dissolve/Dip/Wipe transitions, manual BPM and two different A/B act banks. Canvas and motion travel with the plan. Map-only starters remain available.
- Download full Session JSON or copy a public recipe link that restores the complete builder. Studio preserves current settings until Load planned show; Review planned acts lists the choices. Undo planned show restores replaced settings in this tab until reload. Save session preserves later calibration and edits; media files stay separate.
- The read-only plan_mapped_show MCP tool uses the same validated model and returns full settings-only Session JSON plus review and Studio links. It rejects private session/media fields and does not apply settings, calibrate a projector, start audio or operate Studio remotely.
- The homepage, live guide, MCP HTML/Markdown and product facts explain the complete workflow. Six fixed-label aggregate milestones cover requested session downloads, recipe copies, Studio handoffs, arrival, explicit loading and audio activation without sending configuration or personal names.
- Scheduled surface cue transitions now start on the same frame-clock sample used by the renderer, preventing automatic Seconds/Bars cues from stopping the visual engine with a backward-clock error. Manual cue launch keeps its existing clock behavior.
- Load planned show prepares the chosen worlds before performance, moving first-use shader compilation out of timed cue transitions. The review collapses to a shorter ready-state panel after loading. Recording and requested sizes remain device-dependent.
Perform moving surface transitions and mix complete mapped shows
- Surface cues now blend two moving complete mapped outputs with Dissolve, Dip through black or Wipe over 0.25–4 seconds, using the same audio and preserving current calibrated corners and masks. Cut and existing Program-only dissolves remain available. Freeze pauses the transition; Blackout hides its output. Interrupting a transition cuts to the newest cue.
- The existing A/B visual mixer can capture every surface look into two banks, animate both banks and blend their final pictures with its manual fader and MIDI controls. End live mix keeps the nearer bank. Surface-look and list editing pause while it runs; calibration geometry remains editable. Both banks share one bounded surface renderer rather than allocating a context per world.
- The existing live guide includes a reproducible three-surface techno show kit: a real Studio recording with original 160 BPM test audio, four cue acts, A/B surface banks and full session downloads in 16:9, 4:3, 9:16 and 1:1. A Studio link keeps your setup until Load example session and offers Undo example session. Audio requires its own click; downloaded sessions do not contain media.
- Full session import/export retains both mapped banks and the inclusion switch, but no active mix, playback or permission. Missing surface IDs reject a start before changing output. Public kit JSON uses the same validated model as the Studio launcher and remains an asset rather than a duplicate indexed page.
- A short desktop browser test and decoded export verify this workflow; achieved frame rate varies with worlds, surface count, output size and hardware. Physical projector alignment and hours-long performance remain unverified.
Build and perform a complete one-projector show
- The existing live guide now includes a complete projector planner: One screen, Triptych, Room corner and DJ booth + wings, with Program crops or independently selected procedural worlds, output shape/size/motion and demo, local-file or live starting audio paths. Its schematic explains one output; it is not a measured room or a live projection preview.
- Eight static map downloads work without JavaScript. Customised map JSON is generated locally using the same validated model as Studio and the new read-only MCP plan_projection_setup tool. Program triptychs use thirds; Program corners use halves; independent surfaces receive complete worlds with editable starter looks.
- A copied setup link reopens the complete plan. Studio opens Mapping and shows it without changing the current map, canvas or audio. Apply layout & canvas is explicit, blocked during recording, preserves Freeze/Blackout and adds the previous map to Mapping Undo. Undo restores the map, not earlier canvas dimensions or motion. Calibration starts off; physical corner alignment remains manual.
- Up to 32 cues. With Mapping enabled, Add cue captures each surface visual, look, FX, crop, visibility, opacity and blend. Recall preserves current calibrated corners, masks, canvas and Mapping enable; missing surface IDs stop the sequence without changing output. Surface cues use cuts, while Program-only cues can use 0.25–4 second dissolves. Manual, seconds and manual-BPM bar holds remain available. Mapping can append Intro/Build/Drop/Breakdown as four Manual-hold cues. Recapture updates a saved cue from the current output. Full sessions save map and cues; map files and starter links do not save a cue show. Cues do not switch video clips or recall global Program rack effects. Import never starts the sequence.
- The eighth MCP tool returns the review link, Studio link, exact dimensions and valid map JSON with public starter names. It does not inspect a room, operate a projector, receive media, access private code or execute Studio. HTML, readable canonical Markdown and agent facts describe these boundaries.
- Independent aggregate counts cover a requested map download, copied setup, Studio click, marked arrival, first audio activation and an applied map. Events do not contain recipe choices, worlds, geometry, filenames or visitor IDs. Public choices remain readable in setup URLs and hosting requests. Counts are not a joined user journey or proof of physical projection. No ranking, citation, traffic or revenue improvement is claimed.
Build, preview and share your visual project
- The homepage now has a complete three-step project builder: choose music video, live set or exploration, find any of the 72 actual worlds through name/style search and family filtering, select the canvas, output size and gentler/full motion, choose a compatible audio starting path, review the complete setup and launch it in Studio. This replaces decorative step labels and fixed per-intent links.
- An explicit Preview click lazily loads one real procedural renderer and public demo audio. Changing the world, canvas or motion updates that preview. Pause, hidden-page and navigation cleanup prevent background playback and rendering; cancellation owns its audio source. No renderer, permission prompt or media request starts just from loading the homepage. Preview resolution is intentionally lower than the planned Studio output.
- Copy setup link shares only validated public configuration. Reopening the homepage link restores the same complete project for review. Studio receives the chosen scene preset, exact aspect ratio, output dimensions and motion, with audio starting separately after a click. The new read-only MCP plan_visual_project tool uses the same model and returns both review and Studio links. Existing MCP planners remain available; HTML and Markdown connection documentation explain all seven tools.
- The existing MP3/WAV guide now has a complete file launcher: open your own track or run the fixed public MP3 and WAV tests with Interference and a preselected 16:9, 9:16 or 1:1 frame. Test frames use a 540-pixel short edge: 960 × 540, 540 × 960 or 540 × 540. Static default links also work without JavaScript; the page does not start a visual renderer or preload the test audio.
- Studio provides an explicit file-source handoff: click to load or resume sound, switch between the two controlled 24-second 160 BPM test inputs, replace them with your own MP3/WAV while retaining the current look and canvas, then optionally continue into the existing video flow. Tests and the public PIP — TARSAN demo are identified separately; ordinary Demo still plays PIP — TARSAN. Public test tempo is known metadata, not automatic beat detection.
- Public sample fetches use a fixed same-origin allowlist and exact byte bounds. Provenance changes only after a successful load; a failed sample request cannot relabel a private track as a public test. Loading and recording protections prevent overlapping source changes. A 540 quick-preview size is also represented in output controls and saved sessions.
- HTML, canonical Markdown and agent discovery descriptions describe the same executable file path, original-input hashes, privacy and downstream recording limits. The existing page owns file playback; the YouTube guide still owns export formats. No extra keyword pages or visual-world rebuild were added.
- Fixed aggregate events separately count homepage setup, first rendered audio preview, copied setup and Studio click, then marked Studio arrival and first successful activation. They also separate public tests from own-file playback and count guide milestones independently. No search text, choices, media, filenames, URLs or visitor IDs enter these counters; public settings remain readable in configuration URLs and normal hosting requests. These are independent counts, not joined individual journeys. Before this release, public WAV tests were included in studio_file_started, so historical own-file counts are not directly comparable. No ranking, traffic or revenue improvement is claimed.
Check your output before a set or recording
- Studio now has a complete output check under Output, guided video Capture and recording settings. A 1.5-second warm-up and eight-second sample compare actual preview draws, display callbacks, frame intervals and CPU frame work with a selected 30 or 60 fps target. Hidden tabs, stopped audio, Freeze, Blackout, engine interruption, recording or changed settings invalidate a sample instead of producing a ready result.
- Explicit Performance detail and 720 output controls let you adjust and retest. Performance preserves output dimensions; 720 preserves the aspect ratio, including custom frames, while reducing output size. A ready sample leads to the existing ten-second recorder, preview and protected download. The report can be downloaded locally without media, media filenames, driver information or visitor identifiers; it is never automatically transmitted.
- On browsers with canvas requestFrame support, the recorder now requests a frame immediately after the complete output is drawn, aligning capture with the existing rendering schedule. Browsers without that API retain automatic frame capture. Each capture owns its video tracks and cloned audio; capture interruption stops and recovers the take without marking the visual engine as lost. This does not bypass background throttling or promise an encoded frame rate.
- Finished recordings also report the preview cadence observed during that take and warn about long gaps, a hidden tab or low cadence. Neither the setup check nor the capture draw count is claimed as encoded-video FPS, physical projector refresh, measured input latency or long-set certification. No browser flags, device settings or shaders are changed.
- The existing troubleshooting guide in HTML and canonical Markdown now owns setup measurement, interpretation, browser visibility, one-change-at-a-time comparisons and recording validation. The existing YouTube guide links to that workflow. Titles, canonical owners and the seventeen-page sitemap architecture are preserved.
From the demo to your first checked video
- The guided video path now accepts both your local MP3/WAV and the public PIP — TARSAN example. A persistent track strip names the source and lets you replace it with your own audio without resetting the scene, frame or output size. A visible video entry also appears after playing a track outside the guide. Demo audio still loads only after a user click.
- A ten-second test capture stops automatically and opens the existing recording preview with sound. Capture length is explicit in the recorder; full recording remains available. Test deadlines belong to each recording session, with the existing track-end, size, encoder cleanup and source ownership protections. Manual stop remains available. Audio playback must be ready before the test clock and encoder start. Preparing a capture blocks duplicate starts and source changes; a playback-start failure leaves the previous take available.
- The finished take shows the captured track name and dimensions. Preview, download, user-confirmed saving and explicit replacement keep the existing unsaved-video protection. Switching songs does not silently replace an earlier take. Local audio and recorded video bytes stay on the device.
- The existing YouTube guide in HTML and canonical Markdown explains the complete demo-to-local-track-to-test-to-download path. Three fixed aggregate labels distinguish test start, a ready test and user-confirmed saving; they send no track name, source type, format, duration, media or visitor identifier. Independent event counts do not prove a joined conversion rate or traffic uplift.
Choose a visual tool and enter the right workflow
- A complete /compare/ hub combines a five-tool source-linked comparison with a keyboard-accessible requirements chooser. Its HTML and canonical Markdown describe PHASE//VOID, Resolume Avenue/Arena, Synesthesia and MadMapper, including publisher disclosure, review date, supported workflows and edition/hardware limits. Existing guide titles and responsibilities remain unchanged.
- The chooser sends music-video, live-set and single-projector tasks to exact Studio entry points in 16:9, 9:16 or 1:1. Native video exchange, OSC, shader coding, multiple outputs, mesh, 3D calibration, timecode and DMX are checked against documented capability sets. Timed lyrics, cloud rendering and combinations without a complete documented match show limitations and partial alternatives rather than an unconditional winner. No audio, permission request or WebGL renderer starts on the comparison page.
- Homepage and guide-directory links connect the comparison to existing setup guides, actual saved recordings and separate custom-project briefs. The public demo description in discovery and FAQ now correctly identifies the owner-supplied PIP — TARSAN MP3 rather than synthesized audio.
- Fixed completion, Studio-click, arrival and first-activation labels use the existing minimal aggregate counter. Chooser answers, tool names, media, URL queries and visitor identifiers are not stored. Counts remain independent and do not establish a joined conversion rate, ranking or citation uplift.
Deeper worlds, sharper light and evolving structures
- Rebuilt the latest 36 worlds while preserving the original 36, all 72 stable IDs, existing favorites, banks, cues and projection assignments. The library keeps its three new identities: articulated metal, optical surfaces and spatial light choreography.
- Metal sculptures have wider silhouettes, changing openings, concave surfaces, separate tiers and continuous limbs. Finite reflected light panels, dark intervals, brushed normals and localized engraving replace the repeated broad white bands. Bound-start tracing, resolution-aware hit tolerances and integrated fine detail improve surface continuity. Bass and mid change geometry; treble reveals selected details.
- Optical worlds now include perspective cloth intersections, layered refractive planes and cellular membranes, displaced relief, deep finite diaphragms and overlapping lace. Fold, aperture and layer relationships evolve. Fine rulings fade to their pixel coverage instead of shimmering at unresolved scales.
- Light worlds gain converging and crossing perspective banks, a dark warehouse with moving rays and floor reflections, beveled changing apertures, a perforated charged membrane, curved nerves, folding prismatic planes, particle currents and depth-distributed meteor trails. Ribbons and organic fins now retain foreground occlusion; the ceiling and triangular resonators reveal stronger depth with finer localized signals. These are screen visuals, not hardware laser control.
- New-world thumbnails are recaptured from the integrated renderer. Verification compares the previous release and the candidate under the same controlled signal, then checks motion, isolated audio bands, 16:9, 9:16, 1:1, Freeze, Blackout and actual Studio recordings. Short checks and GPU samples on one desktop device do not establish physical latency, projector alignment, integrated-GPU performance or long-set reliability.
- Demo in Studio, scene previews and embedded players now streams the complete owner-supplied PIP — TARSAN (prod. TAR$AN) MP3 from PHASE//VOID. Its title, 142 BPM and G minor attribution are visible; tempo comes from the supplied track information, not automatic tempo detection. Nothing loads before the user starts it. The demo supports pause, seek, restart, loop, analyzed audio response and recording through the existing audio graph. The 3:52 stereo file is published as an example; user-selected files still remain local. Historical controlled-test recordings retain their explicitly documented original source.
Wrapped stage controls keep their own row
- The small-screen toolbar height rule now follows broader responsive rules, giving its wrapped controls enough vertical space instead of overlapping the canvas. Verification checks both horizontal containment and the actual bounds of both control rows with Boost FPS visible. Renderer geometry, audio, favorites, banks, mapping and recording are unchanged.
Small-screen performance controls stay reachable
- On narrow screens the stage toolbar now wraps when the low-frame-rate Boost FPS control appears, keeping Fit, fullscreen and scene navigation inside the viewport. The browser layout check explicitly includes this conditional control. New-world rendering, audio, bank data, mapping and recording remain unchanged.
Stable 72-world share metadata
- Repeated SEO builds now replace the complete world-count token in the homepage share description, keeping the actual count of 72. The previous update could preserve leading digits and append another digit on each build. Added exact-count and repeated-build checks for both source and published descriptions. The 72-world library, new geometry, favorites, bank behavior, mapping and recording are unchanged.
72 worlds, favorites and performance banks
- Thirty-six new original procedural worlds double the library to 72: metallic and organic sculptures, psychedelic optical surfaces, and light/particle choreography. Each has its own geometry and animation, with audio response. Existing world IDs and order stay stable. The additions work in Studio, A/B, off-air Preview, saved cues, independent mapping surfaces, deep links, public previews and the read-only MCP catalog.
- Favorite any world with its star and filter the library to Favorites. Create up to twelve named banks, add or remove the selected world, and combine banks with new-world/family/search filters. Curation leaves Program unchanged. Banks and favorites persist on this browser and transfer through a separate JSON export/import. They contain stable world IDs, not audio, media, parameter looks or cue order. Invalid imports preserve the existing library. Concurrent edits in supported browser tabs are serialized against the latest stored library. Delayed imports cannot replace newer choices, and storage failures keep an exportable in-memory copy with a visible warning.
- The narrow Studio inspector now stacks A/B decks and wraps their Preview controls, so buttons remain separate and reachable. The expanded library and open bank controls fit desktop and small mobile widths.
- An official-source Resolume Avenue/Arena audit guides this release and the next priorities: per-parameter FFT/BPM automation, performance macros and ordered FX racks. Native plugin/NDI/Spout/DMX/DJ-hardware capabilities are not claimed as implemented browser features. All public tools remain free and the private repository stays private.
- Acceptance includes actual WebGL motion, isolated audio response, landscape/portrait/square, Freeze/Blackout, real WAV playback, favorites and bank persistence/import recovery, A/B plus Preview, cues, independent surface assignments and reopened moving stereo recordings. Thumbnails come from this renderer. These bounded checks do not establish physical projector alignment, device latency, integrated-GPU performance, long-set reliability or search-growth effects.
Different visuals and effects on every projection surface
- Each of up to eight mapping surfaces now chooses its own source: Program, the complete shared composition, or one of the 36 procedural worlds. Independent worlds have their own palette, audio response, motion speed, detail, glow, zoom, rotation, symmetry, grain and chromatic light. Use current studio look copies the editable look into the selected surface. Changing a surface leaves its corners, crop and mask in place.
- Every surface has its own Kaleidoscope → Pixelation → RGB split rack, amount, audio-energy link and bypass. This rack processes the chosen source inside its crop before perspective, mask, opacity and blending. Program surfaces retain the main single-world/A-B composition, local footage, global effects, title and artwork; independent worlds exclude those Program layers.
- Worlds share the existing audio signal and selected pulse, with separate motion clocks. One additional reusable renderer updates audio envelopes and textures once for a complete surface batch; it does not allocate a renderer or audio analyser for every surface. Visible independent sources share a 4-megapixel raster budget with a 1920-pixel edge cap and obey render detail. More worlds can still reduce frame rate or incur first-use shader compilation.
- Assignments, looks and surface effects remain local and survive map/session downloads, validated imports, reload, duplicate and undo/redo. Existing maps default to Program and neutral surface effects. Calibration replaces all sources and bypasses their FX. Global Freeze, Blackout and Gentler motion apply to the complete output, which remains shared by Studio, projector, PNG and recording.
- The live-performance guide, Markdown and read-only MCP describe independent surface visuals and their limits. MCP planners still create procedural-world links and do not create or modify projection maps. Public tools remain free; no private repository or user media is published.
- Acceptance checks independent source pixels and movement, one-surface look/FX changes, audio sharing, stable geometry, backward-compatible persistence, calibration, resource bounds, mobile layout, identical projector output and reopened stereo recordings in 16:9, 9:16 and 1:1. These short browser checks do not establish physical projector alignment, live-device latency, integrated-GPU support, long-set reliability or search-growth effects.
Studio menu startup correction
- The production browser check found that renaming moved header controls queried them while their menu was still detached from the document, interrupting Studio initialization. Labels are now set before moving those controls. Mapping, the renderer and saved configuration are unchanged. The complete Studio workflow is rechecked after this correction, including menu downloads, projector output and recordings.
Performance console and single-projector mapping
- Studio adopts the requested navy performance-console layout: tool sidebar, large final stage, Layers/Mapping/Effects/Audio inspector, searchable 36-world shelf and compact show-cue rail. Audio, timing and MIDI live in Audio; composition decks and local footage live in Layers; final FX has its own inspector. Header Go live opens the existing single-projector window. Session, guided workflows and help remain in the header menu.
- One final output can contain up to eight independently named surfaces. Each uses a true four-corner projective transform, source crop, opacity, Normal/Screen/Add/Multiply/Difference blending, visibility and stacking order. Rectangle, ellipse and editable 3–16-point non-crossing polygon masks support inversion and feather. Drag corners or the surface, nudge by one output pixel with arrow keys or ten with Shift, enter numeric coordinates, duplicate, remove and undo/redo changes.
- Numbered grid, checkerboard, color/orientation and white test patterns help manual setup. Mapping is applied after the procedural/A-B base, muted local footage, final effects, titles and artwork. Stage, projector, PNG and recording use the same final canvas. Studio edit handles never enter that canvas. Freeze holds its complete pixels and Blackout masks every surface and calibration pattern.
- Map settings remain local and can be downloaded as a projection-map JSON or in a session. Imports validate geometry, surface count, masks, crops and supported properties before replacing the current configuration. Legacy sessions bypass mapping. Media bytes are never embedded, uploaded or accepted by the MCP server; restore does not restart audio, footage, MIDI, sequence or calibration patterns.
- The existing live-performance guide, Markdown, privacy notice, discovery facts and read-only MCP describe the actual one-projector workflow. MCP planners still preselect procedural worlds and do not generate, calibrate or apply maps. Multi-projector routing, 3D mesh mapping, projector edge blending and native media-server protocols are deferred.
- Acceptance covers actual GPU orientation and projective geometry, masks, crops, blends, bounded targets, resource-failure black output, editing and persistence, identical browser-projector pixels, complete Freeze/Blackout, desktop and mobile layouts and reopened recordings in 16:9, 9:16 and 1:1. Separate clip/FX regression checks retained the existing composition workflow. This bounded browser validation does not establish physical room alignment, live-device latency, controller compatibility, long-set reliability or search-growth effects.
Pressure pack: 36 worlds and focused shader compilation
- Twelve original compositions expand the free library from 24 to 36. Chrome knot, Spectral organ, Shutter drive, Sonic crystals and Alien membrane add interlocking tubes, curved cable walls, occluding iris leaves, faceted tips and merging perforated lobes. Vortex sheet, Laser switchyard, Op art axis, Neural web, Diffraction sea, Circuit run and Ribbon detonation add folded point fields, scanning laser walls, optical ribbons, travelling graph signals, iridescent caustics, perspective lanes and percussive ribbon fronts.
- Each world has its own geometry and movement with measured audio response. These are authored procedural scenes, not third-party Resolume footage, recoloured copies or a claim of native media-server parity. The New filter selects this twelve-world Pressure pack; all 36 remain searchable by name, description, use and family.
- The renderer specializes its program to the selected world and caches compiled programs for revisits, A/B decks and transitions. Opening the renderer no longer asks the graphics driver to compile one 36-way scene branch. A first visit to another world can still incur driver compilation; no universal startup-time or frame-rate guarantee is made.
- The public catalog, homepage, techno guide, their Markdown representations and read-only MCP expose the same library. MCP capabilities now describe local Clip Deck, fixed output FX, copied cues, off-air Preview and MIDI while clearly retaining the existing planners’ one- or two-procedural-world scope. No private repository or user media is published.
- Acceptance checks actual GPU motion in landscape, portrait and square, same-clock isolated audio response, Freeze/Blackout, a real first-party WAV, A/B plus off-air Preview, copied cues, mobile containment and reopened video/audio recordings. Published previews come from this renderer. These bounded checks do not establish live-device latency, physical MIDI, a long show, integrated-GPU compatibility or search-growth effects.
Clip Deck example fits every screen
- The new recorded three-layer example now uses the guide’s responsive media frame. Its 1280 × 720 source fits the article on desktop and at 390/320 px without horizontal overflow. This closes a production layout issue found during final acceptance of the 1.27 Clip Deck release; the layer engine and export workflow are unchanged.
Local Clip Deck: three layers and final output effects
- Compose your procedural world or A/B mix with two independent, muted local video layers. A library of up to twelve files supplies layer A and B; Take decodes a candidate before replacing usable media. Clip audio never enters the soundtrack. MP3/WAV or live audio remains the separate signal and recording source.
- Each video layer has play/pause, restart, loop, seek, 0.25–2× speed, trim In/Out, opacity, Normal/Screen/Add/Multiply/Difference, Fit/Fill and horizontal flip. Swap the two layers’ bottom-to-top order. Removing a source releases its URL and stops only layers using that file.
- A final GPU rack processes the complete layered image in a fixed Kaleidoscope → Pixelation → RGB split chain. Adjust the amount, link it to audio energy or bypass the rack. GPU allocation or context errors bypass final effects while the base and clips continue. Titles and artwork sit above the rack.
- Freeze holds complete final pixels and pauses both video transports. Blackout masks worlds, video layers, effects and overlays. Stage, projector and recording consume one final canvas, including landscape, portrait, square and custom dimensions.
- Session storage and downloaded sessions retain clip descriptors and settings, never media bytes, blob URLs or playback state. Reopen or import, select the original files again, then press Play. Legacy sessions start with an empty Clip Deck. MIDI Learn adds layer opacity, play/restart, FX amount and bypass; physical-controller compatibility remains unverified.
- The canonical live-performance guide, its Markdown alternate, product facts and AI-readable discovery text explain the entire import → layer → perform → record workflow and limits. Admission is limited to 200 MB per file, 512 MB per library and 4096 pixels on either decoded edge; codecs depend on the browser. DXV, native routing, plugins, quantized launch and projection mapping remain unavailable.
- Acceptance uses original local WebM loops, a separate first-party WAV, real hardware rendering, full-frame Freeze/Blackout checks, projector comparison, transport and session restoration, and reopened video/audio exports in 16:9, 9:16 and 1:1. Short Chrome checks do not establish integrated-GPU support, physical MIDI, live latency or long-set reliability. No search-growth or native media-server parity claim is made.
24 visual worlds: metal, wire, ink and live Preview
- The procedural library doubles from 12 to 24 independently rendered compositions. Ferro bloom, Kinetic blades, Void tendrils and Fractal forge add metallic and organic 3D sculpture; Wire loom, Helix run, Acid ink, Photon lattice, Plasma weave, Orbital glass, Rave fan and Moiré engine add mesh, flight, fluid, refractive, laser and optical motion. These are original generated visuals, not licensed third-party footage or recoloured copies.
- Studio has searchable world names, descriptions and uses, a New filter and five visual families. Every new world works with scene links, copied cue looks, saved sessions, both A/B decks, projector output and recording. The public catalog, embed selector and read-only MCP share the same 24-record source; HTML and Markdown describe the actual library.
- The A/B studio now renders a separate, low-resolution moving Preview of the current editable look. While A and B perform, changing the preview scene or controls does not alter the stage, projector or recording. Edit A/B in Preview copies a deck to the controls; Take Preview into A/B explicitly updates that deck. Preview falls back to a static scene reference if its WebGL surface cannot be created and slows its refresh during recording or low frame pacing.
- Picture cuts work with [ and ] while the mixer is active. MIDI Learn now maps an absolute 7-bit CC to the A/B picture crossfader and note/CC buttons to Cut A/B or Take Preview into A/B. These controls do not change audio routing; mappings stay local or in an exported session and MIDI still connects only after a user click.
- The canonical live-performance guide and Markdown alternate document the complete prepare-preview-take workflow with a first-party console capture and the existing playable, actual A/B recording. Product facts and AI-readable discovery text state the new capabilities and hardware limits without adding a thin comparison page or claiming search ranking gains.
- Browser checks cover independent preview and stage frames, keyboard cuts, synthetic MIDI fader and cut events, A/B output in 16:9, 9:16 and 1:1, and 320/390 px layouts. Synthetic MIDI is not a physical-controller compatibility test; a short browser run does not establish long-set, projector or integrated-GPU performance.
- A controlled hardware-WebGL run checks moving output, same-clock audio-band response, exact Freeze and Blackout for all twelve additions in landscape, portrait and square. Published previews are captures of this renderer. Detailed verification and its limits are recorded in the project evidence.
Share a complete A/B mix and watch it move
- Studio can copy a link containing both procedural deck looks, their control settings, the picture crossfader, preset or custom frame dimensions and a chosen audio starting path. A fresh browser restores the recipe without automatically starting sound, capture permissions or the performance. The link contains no media or artwork.
- The read-only MCP connection adds plan_live_mix, returning a validated two-world Studio link and manual setup steps. Its public documentation, product facts and AI-readable discovery text explain the behavior and limits without exposing the private repository.
- The existing live guide now has a one-click 16:9 Rave laser and Interference recipe, exact steps and a playable seven-second first-party A/B export. The saved 1920 × 1080 VP9/Opus WebM was reopened with three distinct decoded frames and 48 kHz stereo audio from the original 160 BPM WAV. The homepage, scene catalog and embed guide link to this canonical example; the old OBS setup note was removed.
- Local Chrome checks covered link restoration in a fresh browser, A/B performance and 16:9, 9:16 and 1:1 output. A short export does not establish sustained frame rate, physical stage compatibility, search indexing, AI citations or ranking gains.
Perform with two moving worlds and a picture crossfader
- The free Studio now has an A/B visual mixer: copy the current procedural look into either deck, keep both worlds animating against the same audio analysis, blend them with a manual picture crossfader or cut directly to A or B. Freeze and Blackout affect both worlds; fader endpoints render only one world.
- A/B mixing takes control from Auto scene and show sequences; a fired cue or newly enabled Auto scene takes control back. Ending the mix lands on its more visible deck. Deck looks and fader position survive in browser settings and downloaded session files without automatically restarting the performance.
- The existing live-performance guide, its canonical-linked Markdown alternate, AI-readable product facts and llms.txt explain the mixer, a first-party midpoint frame and limits. This remains procedural two-world mixing, not imported video clips, a multilayer grid, OSC or projection mapping.
- Integrated local Chrome checks confirmed distinct A/mid/B canvas frames under demo audio, lit output in 16:9, 9:16 and 1:1, global Freeze and Blackout, mode handoff and a 390 px mobile layout. This does not establish long-set or physical projector performance, search indexing, AI citation or ranking gains.
Take live audio from input to a checked projector output
- The free Studio now guides a live set through microphone or supported shared-tab audio, a moving signal check, scene and frame selection, manual BPM, audio or beat-grid pulse, timed Auto scene and cue setup, then a projector window or stage fullscreen. Freeze, Blackout, a short test recording and Disconnect are available in the same path. It does not request capture permission until the user chooses an input.
- Homepage live actions and the existing live and techno guides now lead into that path. The live guide also has a canonical-linked Markdown alternate with browser, input, recording and latency limits; no duplicate search landing page or unverified ranking claim was added.
- Fixed first-party counts record only a guided live-path opening and a successfully connected projector window. Audio, device names, scene choices and visitor identifiers are not sent with those counts.
- Local browser checks covered microphone signal, three output ratios, projector frames, Freeze and Blackout, a decoded WebM test with Opus audio, released capture tracks, denied and silent tab sharing, blocked pop-ups, fullscreen fallback and 390/320 px layouts. A separate v1.23.0 Chrome shared-tab run detected all 32 synthetic pulses, with 42.2 ms median and 49.8 ms 95th percentile from estimated source output to shader submission; this is not physical sound-to-screen or projector latency. Search and conversion effects remain unverified.
Make a music video from one guided studio path
- A guided path in the free Studio now takes a local MP3 or WAV through scene choice, 16:9, 9:16 or 1:1 output, browser recording, preview and download. The homepage music-video choice and its lower creator path lead directly into it; live and scene-exploration choices remain separate.
- The guide reuses the existing real-time recorder and unsaved-take protection. It shows browser-supported formats and output dimensions, recommends a short test, and keeps the selected audio on the device. It does not convert a full track instantly or claim that a download request proves the file was saved.
- The existing MP3/WAV and YouTube guides now explain the same real workflow and limits in HTML and canonical-linked Markdown alternates. Fixed first-party counts distinguish homepage guided-video clicks and Studio flow openings from file starts and video download requests, without media names or visitor identifiers.
- Local browser checks produced and decoded video with Opus audio at 1280 × 720, 720 × 1280 and 720 × 720. Search indexing, ranking, AI citations and conversion improvement remain unverified until external reports are available.
Embed a live audio-reactive scene on your own site
- The free embed builder lets a site owner choose any of twelve actual visual worlds, preview it, pick 16:9, 9:16 or 1:1 and copy a responsive iframe. HTML and WordPress setup instructions, a Markdown alternate and limitations are available at /embed/.
- The embeddable player starts its WebGL scene and original demo sound only after a visitor click, or reacts to a local MP3/WAV selected inside the frame. It pauses off-screen or when hidden, does not access the host page's audio and cannot request a microphone. The player alone permits cross-site framing; other pages retain their frame restriction.
- The public MCP connection now offers a validated, read-only get_embed_snippet tool. It returns HTML and a player URL, never private renderer source or visitor media. First-party counts distinguish copied snippets from started embedded players without host URL, file name or visitor ID fields.
- This adds a new distribution path and measurable usage signals. Search indexing, third-party embed adoption, AI citations and ranking gains remain unverified until external data arrives.
Explore all twelve visual worlds before opening Studio
- A public scene catalog now lets visitors compare all twelve real visual worlds in readable HTML and Markdown. A single click starts one audio-reactive WebGL preview driven by original browser-generated 136 BPM demo sound; no audio file or microphone permission is needed.
- Visitors can choose a scene, 16:9, 9:16 or 1:1 frame and demo, local-file or live-audio starting path, then open the exact configuration in the free Studio. Homepage, guide and MCP links now lead to the catalog. One shared metadata list supplies the catalog, MCP and studio scene names.
- The catalog uses fixed first-party counts for preview starts and Studio handoffs. A marked Studio link separately counts arrival and first successful audio activation without visitor IDs, scene IDs or media. These independent counts do not establish a per-person conversion rate.
- This release improves product discovery and the first-run path; indexing, search rankings, AI citations and conversion lift remain to be measured after reporting delay.
Guide visitors into the studio and open the visual catalog to MCP
- The homepage now follows a clear three-step start: choose a music-video, live-visual or scene-exploration path, see a matching preview and open the corresponding free-studio flow. Its full-page sections use real browser recordings and actual 16:9, 9:16 and 1:1 output frames.
- The existing instrument is available at /studio/. Guided links can preselect a scene and output shape or show the local-file and live-input steps. The instant demo starts after a click in Studio, as required by browser audio policy.
- A public, read-only MCP endpoint and connection guide let compatible assistants browse the twelve scene descriptions, inspect capabilities and plan a studio link. Neither the private repository nor visitor media is shared through MCP.
- This release changes navigation and discoverability. Search ranking, AI citations and first-run conversion effects require later measurement.
Open video creation from the YouTube guide
- The YouTube visualizer guide now puts a clear Open the free studio to record action beside its setup summary. It leads to the existing studio, where visitors can choose local audio, output shape and browser-supported recording settings.
- The guide keeps its existing title, heading, canonical URL, format limits and real export example. The new link uses the same first-party guide-to-studio action count as other guide links; it collects no selected audio or visitor identity.
- This is a navigation and first-run hypothesis. Search clicks, rankings, completed recordings and conversion effects still need later evidence.
Make the free studio clear on the first screen
- The opening canvas now identifies PHASE//VOID as a free audio-reactive visualizer above the existing hero message. Instant demo and Open MP3 / WAV remain the first actions.
- The homepage title and description now name MP3, WAV and live audio, local files, projector use and 16:9, 9:16 and 1:1 recording. The wording matches the actual free studio and its documented limits.
- Search impressions, clicks, ranking or first-run activation effects are not established by this wording change; they need later measurement.
Notify search engines only about changed pages
- The IndexNow release workflow now requires exact changed page paths from the canonical sitemap. Preview shows the selected batch before any notification, and submission verifies the published ownership key and sitemap.
- The homepage, techno scene guide, Signal rupture page and this release history were updated after the last notification. Their four URLs can now be submitted as one small, auditable batch without resending unrelated pages.
- An accepted notification only confirms receipt. Search indexing, rankings, AI citations and new visits still require later independent evidence. All public studio tools remain free.
Show first-party output on the studio homepage
- The homepage now shows still frames from existing browser recordings of Signal rupture, portrait and square Interference, and shared-tab live input. Each example links to the corresponding setup, moving result and limitations.
- The example frames load lazily, with explicit dimensions and descriptive alternatives. The Signal rupture link preselects its free studio scene; its public test WAV loads only after a separate Play original WAV click.
- This improves the site's visible product evidence and navigation; the tiny, delayed GSC and Bing samples do not yet show a search or conversion effect.
Try recorded techno scenes with their original sound
- The techno comparison and Signal rupture watch page now open a matching studio scene directly, with a 16:9 canvas and the original 160 BPM public test loop available from one clearly labeled button.
- The 4.6 MB test WAV is requested only after a visitor clicks Play original WAV. The ordinary generated demo and local MP3/WAV import remain separate, free options; the site does not upload a visitor's media.
- Authenticated 26 September reports show 15 GSC impressions through 24 September and two Bing impressions through 25 September, both with zero clicks. Those delayed, tiny samples do not establish an effect of this release or AI citation gains.
Start the free visual demo in one click
- The opening studio screen now leads with Try instant demo while keeping local MP3/WAV import beside it. The demo creates its own sound in the browser and needs no file, microphone permission or account.
- The two first-run actions have at least 44-pixel-high targets on phone and tablet layouts. Local Chrome checks at 390, 768 and 1440 pixels verified the layout and a successful demo activation.
- The visible homepage FAQ and AI-readable product facts explain the same first-run path. Search rankings, AI citations and visitor conversion gains remain unproven; all public tools remain free.
Measure the shared-tab path to the visual pulse
- A repeatable production benchmark sent 96 uneven synthetic impulses through native Chrome shared-tab audio capture into Rave laser at 1280 × 720. All 96 reached the WebGL pulse threshold; the interval from estimated source audio-output-device sample time to shader submission measured 45.5 ms median and 54.2 ms at the 95th percentile across three runs.
- The live guide and AI-readable product facts publish the method, per-run results and a sanitized 96-event CSV. The release checks the CSV fingerprint and that its measured values agree with the guide.
- This is a digital-path measurement on Chrome 153, not physical mixer-to-projector or perceived sound-to-screen latency. The visual engine is unchanged; search rankings, citations and conversions remain unproven. All public tools remain free.
Explain the three ways visuals follow a beat
- The homepage FAQ, product facts and AI-readable discovery text now describe the actual pulse choices: detected audio transients, manually aligned BPM for live input, and track-position timing for local MP3/WAV files. A direct link leads to the existing Beat sync setup guide.
- The explanation separates visual pulse timing from Auto scene and cue deadlines. It states that BPM is supplied, not detected, and that the up-to-250 ms visual adjustment does not remove physical sound-to-screen latency.
- A fresh GSC URL inspection found the About page indexed despite the older aggregate report listing it as discovered. The tiny GSC and Bing performance samples still do not establish a ranking, traffic or AI-citation improvement. All public tools remain free.
Give indexed videos accurate publication times
- Google Search Console confirmed that the Signal rupture watch page and its video were indexed after a successful 25 September crawl. The video was recognized as watch-page content; this does not establish impressions, clicks or a ranking.
- Search Console also flagged two non-critical uploadDate warnings on the valid VideoObject. All eight playable video records now use an ISO 8601 publication datetime with an explicit time zone, based on their first successful Cloudflare deployment minute or first verified production observation.
- The SEO build checks datetime formatting and the visible media, poster and sitemap match. GSC and Bing performance reports still predate these releases; all public tools remain free.
Show moving portrait and square exports
- The aspect-ratio guide now plays two tested Interference takes with sound: native browser recordings at 1080 × 1920 and 1080 × 1080, served as smaller 720-pixel MP4 copies. The posters come from those files, and the existing 16:9 export remains linked from the guide.
- Each clip uses the same original 160 BPM test WAV. The saved videos were checked for exact 9:16 and 1:1 dimensions, changing frames and non-silent stereo audio; the page distinguishes recording size from playback-copy size.
- The clips show two tested browser exports, not guaranteed performance on every device or physical audio-to-projector latency. This update does not establish a search ranking, AI citation or traffic gain. All public tools remain free.
Give Signal rupture its own video watch page
- The Signal rupture recording now has a dedicated video-first page with its original 160 BPM WAV, repeatable input steps and clear recording limits. The three-scene techno comparison still links to and embeds the same playable clip.
- The video sitemap lists this recording under its watch-page URL, and that page names the video as its main entity. A fixed, first-party count records clicks from this watch page to the studio; the other five reviewed recordings retain their existing pages and metadata.
- Google Search Console showed the techno comparison page indexed on 25 September but its selected video excluded as supplementary content rather than a watch page. This release addresses that specific page-purpose signal; Google has not yet crawled or indexed the new page or its video, and no ranking, citation or traffic gain is claimed. All public tools remain free.
See whether Product Hunt reaches the free studio
- The verified Product Hunt website link includes ref=producthunt. Opening that exact link now records a fixed studio-arrival label, and the first successful file, live-input or demo start records a separate fixed activation label.
- Only fixed action and page labels reach the first-party endpoint. The query string, referral URL, visitor identity and media are not analytics fields; the Privacy page explains the new counts and their limits.
- These are independent per-page-load counts, not unique visitors or a joined conversion rate. The marker can be copied, and the launch was still scheduled when this change was prepared. Existing GSC and Bing reports predate this release; no search, AI-citation or revenue effect is claimed. All public tools remain free.
Keep portrait recordings closer to their selected frame rate
- The studio now advances a scheduled render deadline while recording, avoiding frame-rate loss from resetting the deadline on each display callback. The selected recording frame rate, output size and native detail remain under the same controls.
- In one complete two-minute Chrome recording at 1080 × 1920 and requested 30 fps, the saved WebM had 3597 video frames at 29.976 average fps, versus 3331 at 27.765 fps before this change. Video timestamp gaps over 50 ms fell from 72 to one; a 129 ms gap near the start remains.
- This one-device result does not establish perfect frame pacing, physical live latency, audio-to-image onset sync or identical behavior on other hardware. Public tools remain free.
Repeat the MP3 and WAV input test
- The MP3/WAV guide now lets visitors play and download the exact two input files behind its existing browser demonstration, with a SHA-256 fingerprint for each and a short repeatable file-input protocol.
- The MP3 is a conversion of the same locally synthesized WAV, while the shown frames came from separate playback moments; the guide does not describe them as a frame-by-frame codec comparison. It links to the separate three-scene techno benchmark.
- The same-day GSC and Bing reports still end before version 1.18.0. This release does not establish indexing, ranking growth, AI citations or conversions; the public studio remains free.
Repeat the techno visualizer test with its original sound
- The techno guide now includes the original locally synthesized 24-second, 160 BPM stereo WAV behind its three playable scene recordings, with an audio player, download link, format details and SHA-256 fingerprint.
- A short reproduction protocol lets visitors bring the same input into the free studio, select a scene and compare the changing structures. The recordings do not preserve every scene control, so the guide does not promise a pixel-identical result.
- This is a new public benchmark workflow, not evidence of search indexing, AI citations or increased traffic. All public tools remain free.
Keep funnel events on the right release
- The Cloudflare Analytics Engine binding now labels new first-party events with the deployed 1.17.3 release instead of the stale 1.17.0 configuration value.
- The release check now requires Wrangler's runtime version to match the public package and release history, and treats Pages Functions and Wrangler configuration changes as release-affecting. CI will reject a future mismatch before deployment.
- Earlier event rows may carry an ambiguous release label and the known synthetic QA event is not organic adoption. Independent action counts do not establish a guide-to-studio conversion rate, completed download, sent enquiry or revenue.
Make first-party measurement clear everywhere
- The product facts, AI-readable discovery text and privacy links now explain the same fixed-label event counting already described on the Privacy page. Media, filenames and project brief text are not analytics fields; ordinary Cloudflare hosting requests still occur.
- The custom-visuals brief form now states that generating a brief or opening an email draft counts a fixed action without sending the brief contents. Sending an email remains the visitor's choice.
- This corrects product information and does not claim a change in search ranking, indexing, AI citation or conversions. The public studio remains free.
Help search engines understand real scene recordings
- Six first-party playable demonstrations now have matching video metadata on their guide pages and in the canonical sitemap: three techno scenes, MP3 input, shared-tab live audio and a YouTube-oriented export.
- Each video entry identifies the actual media file and poster with a distinct title, description, publication date and verified duration. The same evidence remains visible to visitors in the page player and accompanying explanation.
- This improves the clarity of video discovery signals; it does not claim video indexing, a search rich result, higher rankings or AI citations. The studio and all public tools remain free.
Measure the path from guide to studio
- First-party, privacy-minimal event counts now distinguish which of six guides leads into the free studio, initial file/live/demo activation, a video-download request, brief generation and an email-draft request. These actions are different from a completed download or a sent enquiry.
- Only fixed event and page labels are sent to a same-origin Cloudflare Pages Function. The server validates the labels and stores them with UTC day and release version in Analytics Engine; it does not store media, filenames, brief content, referral URLs or visitor identifiers as analytics fields.
- The privacy page explains the new measurement and its limits. Google Search Console's latest visible period still has 11 impressions and no clicks, so this release does not claim a search ranking, citation or conversion improvement.
Choose the techno language
- A new techno and rave guide compares three current first-party scene recordings—Signal rupture, Interference and Solar flare—driven by the same original 160 BPM stereo WAV. Each clip comes from a saved browser video with sound, then a smaller MP4 conversion for accessible playback.
- The guide helps DJs and video creators choose a visual structure, plan contrasting cues, select output shapes and test a projector or short export. It states the browser and hardware limits and does not claim automatic BPM detection, measured physical latency or real laser control.
- The new guide is linked from the studio information, guide index, related guides and first-party Markdown directory. Its distinct creative task is recorded in the dated search intent map; this release does not claim search ranking or AI citation gains.
Prove the file path
- The MP3/WAV guide now demonstrates the core local-file workflow with an original stereo 160 BPM signal loaded separately as PCM WAV and MP3 into the production studio. It shows WAV-driven changes in the Interference scene and a playable MP3-driven recording with sound.
- The MP3 clip is a smaller MP4 conversion of an actual saved browser WebM. The guide identifies the source, browser, scene and test limits and links directly to the free studio without implying that different file formats or devices produce identical results.
- The first-party Markdown discovery directory points to the verified file example. Existing guide titles and topics remain stable while Google Search Console has too little query evidence to justify retargeting.
Show the live signal
- The live-visuals guide now shows a real shared-tab audio performance: an isolated Chrome tab played an original 160 BPM stereo WAV, PHASE//VOID captured its audio, rendered Interference, filled the projector window and saved a recording with sound.
- A user-controlled 960 × 540 MP4 demo is derived from the verified browser WebM recording. Its poster and video load separately; the page states the source, captured audio settings, browser, tested duration and limits without claiming physical mixer-to-screen latency.
- The first-party Markdown directories now link to the live demonstration, video export, aspect-ratio captures and troubleshooting guidance. All public studio functions remain free.
Watch the real take
- The YouTube guide now plays an original twelve-second WebM saved by PHASE//VOID from a local 160 BPM test WAV. The clip includes stereo sound and a changing Solar flare scene, so visitors can inspect a real result instead of only a still frame.
- The player loads on demand with a poster decoded from the same file. A direct download link remains available when a browser cannot play WebM; the guide states the tested browser, frame size, format and limits.
- This is first-party product evidence, not a claim of Google video indexing, universal codec support or identical performance on every device.
Show the real output
- The YouTube recording guide now shows an actual 16:9 frame decoded from a locally recorded PHASE//VOID WebM with original 160 BPM test audio. Its caption identifies the scene, source and limits rather than implying all devices will render the same result.
- The aspect-ratio guide now shows real landscape, portrait and square Solar flare captures from the same browser acceptance run, with dimensions and a clear distinction between recorded and live canvas captures.
- The SEO measurement baseline now separates Google Search impressions and indexing from Bing search and Bing AI citations. Google generative-performance data is marked unavailable while its property report is not visible, rather than being counted as zero.
- The active project goal now allocates about 80% of work to search visibility, AI answer visibility, distribution and conversion, while keeping the public studio free and reserving about 20% for product improvement. Rankings and citations remain measured goals, not release claims.
Tear open the sun
- Solar flare now fills the frame with evolving plasma arcs. A giant broken arc peels into unequal hooks, reconnects across a dark slit and breaks into opposing jets. The four arrangements evolve over twelve scene seconds at Motion 100% with Gentler motion off.
- Red and orange current sheets contain branching hot fissures and dark cellular gaps. Bright knots move through the structure, while tapering branches replace the previous sphere and its fixed outline.
- Bass changes the arc scale, mids bend the currents and highs brighten local knots. The selected audio or Beat sync pulse displaces the geometry. Symmetry changes every arrangement, Detail changes its internal texture, and Motion controls the transformation independently of BPM.
- World 09 keeps its Solar flare name and saved identity, with reduced default glow and a new thumbnail captured from the running visualizer. Portrait adapts the composition along the long axis; landscape, square, custom sizes, projector output and recordings use the same scene. Existing saved looks retain their own settings.
Fracture the light
- Prism ritual replaces solid metal blades with open prismatic light gates. An off-centre tunnel fractures into jagged wings, twists into a braided corridor and splits into two angular throats over twelve scene seconds at Motion 100% with Gentler motion off.
- Bass opens the gates, mids fold their edges and high frequencies send bright accents along them. The selected audio or Beat sync pulse displaces the geometry; the continuous transformation still follows Motion rather than detecting musical phrases.
- The light has sharp cores, spectral fringes and black spaces between layers. Portrait output rearranges the composition along its long axis; landscape, square, custom sizes, projector output and recordings use the same scene.
- World 04 keeps its Prism ritual name and saved identity, with a lower default glow and a new thumbnail from the running visualizer. Existing saved looks keep their parameter values. Symmetry changes the folds and spacing in every phase, and Detail changes the number of repeated frames.
Stay with the track
- Beat sync · track position is a new pulse source for local MP3/WAV files. Enter the track's BPM and press Resync on a beat; pulses then follow the file through pause, resume, seek, restart and looping, including recording from the beginning.
- Track sync uses your manual tempo, with the existing 250 ms earlier/later adjustment. It does not detect BPM, follow tempo changes automatically or remove device latency. Beat sync · manual BPM remains available for live audio and the demo, and Audio transients remains the default.
- Changing BPM preserves the current pulse phase; Tap or Resync sets a new alignment at the current file position. A new file, source change, reload or session import resets track alignment to beat 1 at file time zero. Session files save pulse mode, offset and BPM without file URLs or alignment timestamps.
- Auto scene and cue timing keep their independent performance clock. Track sync does not seek audio or move cue deadlines; its pulse is shared by the studio, projector and recording, with the existing Freeze, Blackout and Gentler motion controls.
Split the void
- Interference fills the frame with flowing acid contours. One slanted tunnel splits into unequal twin vortices, tears into an open wavefront and reconnects through a curved throat. Its four compositions evolve over ten scene seconds at Motion 100% with Gentler motion off.
- Bass opens the apertures, mids bend the contours and highs light traveling accents. Audio transients or the selected Beat sync pulse move the geometry as well as its highlights. Motion and Gentler motion still control the pace of the transformation.
- A stronger primary contour and quieter echoes keep black gaps between the light. The composition turns for portrait output and uses the same live controls in landscape, square, projector output and recordings.
- World 12 keeps its Interference name and saved scene identity, with updated default controls and a new thumbnail captured from the running visualizer. Existing saved looks retain their own parameter values.
Lock the pulse
- Beat sync is an optional pulse source driven by your manual BPM, Tap and Resync. Laser hits can follow that grid while measured spectrum, energy and audio-driven movement keep responding to the music. Audio transients remains the default.
- Visual timing shifts the grid pulse up to 250 milliseconds earlier or later without moving audio playback or cue deadlines. Negative values advance it; positive values delay it. This is manual alignment, not automatic tempo detection or removal of device latency.
- Pulse source and visual timing travel with session files and recent browser settings. Older sessions keep audio-transient mode. Loading a session does not start audio or override the current Gentler motion, Freeze or Blackout controls.
- Beat sync follows the current grid after a pause or missed frame, without replaying missed hits. It keeps pulsing through breaks while a source runs; re-align with Tap or Resync after seeking or a tempo change. Recording and projector output use the same selected pulse.
Music comes through
- Tab and system capture now request stereo without echo cancellation, noise suppression or automatic gain control. A short capture-buffer preference avoids the larger default buffer observed with unprocessed audio; browsers may choose different supported settings.
- The source label reports the actual channel count and whether the browser reports speech processing on or off. These properties contain no device IDs or captured source titles.
- The required, unused capture video requests a small five-frame-per-second stream. Supporting browsers keep the visualizer focused when sharing starts; all capture tracks still stop together on disconnect.
- This improves the live-input path without promising zero latency. Browser capture, audio hardware and display delays remain separate; manual BPM continues to control cue timing.
On the hit
- Rave laser now uses six moving light banks, colored beam cores and stronger directional changes on detected audio transients. Crossfire, tunnel rakes, fans and curtains evolve through a ten-second scene-clock cycle at Motion 100%.
- A faster audio-response path and immediate full-motion transient attack make short accents more direct. The analyser keeps listening on display callbacks skipped by video capture, retaining the strongest onset for the next rendered frame.
- Gentler motion is now visible directly below Motion. Its active setting shows the resulting motion percentage and offers Use full motion. Existing system and saved preferences remain respected.
- Full motion has shorter audio envelopes; Gentler keeps its softer response. Manual BPM still controls cue timing independently. Browser, source and device delays are not automatically compensated.
- A Boost FPS action appears after sustained slow live rendering. It enables Performance detail while keeping the canvas and recording dimensions; Output shows the internal render size and lets you restore Native resolution.
Light cuts through the dark
- Rave laser replaces Laser cathedral with sharp light beams, opening fans, crossfire, bursts and scanning curtains. Complexity changes the beam count; Symmetry adjusts their spread.
- Signal rupture replaces Vector sculpture with dense luminous filaments that evolve through vortices, flowing ribbons and ruptured apertures. Symmetry changes their folds; detected transients shear the field and accent travelling highlights.
- Both scenes respond to bass, mids and highs, adapt to landscape, portrait and square output, and keep Freeze, Blackout and Gentler motion. Scene evolution follows Motion speed; the manual BPM grid still controls cue timing independently.
- Updated scene thumbnails show the new visuals. Existing saved looks and cues retain their scene slots and settings, and now use the rebuilt visuals.
Perform with pads and knobs
- MIDI Learn connects one selected input in a browser with Web MIDI. Assign note pads or CC buttons to cue playback, individual cues, world changes, Tap, Resync, Freeze and Blackout.
- Absolute 7-bit CC knobs and faders control the nine visual parameters and color palette. Learning consumes the message without performing its action; held notes do not repeat a trigger.
- Mappings stay in this browser and travel with session files, without device names or IDs. Importing a session disconnects MIDI. Removing a cue also removes its MIDI assignment.
- MIDI actions pause while editing a dialog or text field and while the studio is hidden. Disconnect stops input handling. MIDI requires browser support and permission; physical-controller compatibility still needs testing on your setup.
Direct your set with a show sequence
- Arrange up to 32 named cues, each with a copied scene and its visual controls. Fire cues individually, run them in order, rename, reorder, duplicate or remove them, and keep the sequence in a session file.
- Choose incoming cuts or 0.25–4 second dissolves between two moving worlds. Hold each cue manually, for a number of seconds, or for a minimum number of bars on your manual BPM grid. Holds begin after transitions.
- Freeze pauses both moving worlds, the dissolve and the hold. Blackout covers the complete output. Cue playback and Auto scene take turns controlling the show; Stop lets the current dissolve finish.
- Importing a session or resetting a look now preserves current Freeze, Blackout and Gentler motion. Session import cannot replace settings while a recording is active.
Prism ritual becomes a kinetic aperture
- Prism ritual now uses tapered metal blades that evolve through a narrow seam, an open aperture, split wings and a spatial spiral. Its shape keeps changing with steady audio input.
- Symmetry selects 3–12 blades, and portrait output adapts the arrangement to a taller frame. The studio thumbnail shows the new scene.
- Bass moves the assembly, midrange tilts the blades and high frequencies brighten travelling edge lights. Motion controls the evolution speed; the manual BPM grid continues to control scene cuts independently.
Keep performance controls when recalling a look
- Recalling a saved look now preserves your current Blackout, Freeze and Gentler motion settings. It no longer unexpectedly reveals or unfreezes the output.
- The saved scene, palette and visual settings still load normally. Use the dedicated performance controls to change those three modes.
Set the pace of your scene changes
- Performance timing adds manual BPM from 30 to 300, Tap tempo (T) and Resync (R) to mark the first beat of a 4/4 bar.
- Auto scene can now change worlds every 1, 4 or 8 bars on your manual grid. The existing 16-second mode remains available, and Freeze pauses automatic changes.
- Timing settings travel with saved sessions. Audio playback and animation speed remain independent: this release does not detect a track's BPM or its musical phrases.
A version for every update
- The studio now shows its version number, linked directly to these release notes.
- This public history records what changed and when. Release notes are also available as a plain Markdown document.
Before public version numbers.
These entries document verified updates from before versioned release notes began with v1.1.0. They are listed by date.
Auto scene respects your performance
- Freeze now pauses automatic scene changes as well as the image. The control shows when switching is paused.
- Choosing a scene manually, unfreezing or returning to the studio gives the scene a fresh 16-second interval before the next automatic change.
An optional screen hold for live sets
- Keep screen awake in Output requests a screen hold where the browser and device support it, with a visible status.
- Screen hold starts off. It releases when the studio is hidden or closed, and retries on return only while you have it enabled.
More control over unsaved recordings
- An active or unsaved recording enables a browser leave warning. Starting another take requires a saved acknowledgement or explicit replacement.
- The preview and download controls appear before settings for the next take. A failed encoder start keeps the previous result available.
- A download click does not mark the file as saved. Check your downloaded file before acknowledging it; in-memory recordings still need to be saved before closing the tab.
Custom visual projects and a local brief
- A Custom visuals page explains separately scoped work with Balance IT for releases, live shows and installations. The public studio remains free.
- A local brief builder lets you review, copy or download your project outline, or open a draft in your email app. The page does not submit or store your brief.
What these notes cover.
Each new public release adds its version, date and the changes you can use. Dates refer to the update, not to the day you visit this page. Prototypes and planned features are not listed as available.
The studio version links here, so you can check which release is open in your tab. An already open studio keeps its loaded version until you reload it. Save any recording before reloading.
Prefer plain text? Read the release notes in Markdown. For current capabilities and limits, see product facts or the guides.