# Four beats, two visual choices

Implemented 2026-10-05 at `visual-lesson.html`. Direction and supplied summary: Dominique Devereaux / Loptr Lab. Implementation assistance: OpenAI Codex. Uses the existing static HTML/CSS/JavaScript stack; no new service or dependency.

## Source status

Candidate reference: [Ren - KUJO BEAT DOWN | Behind The Scenes](https://www.youtube.com/watch?v=HiPY_BcyaoM), RenMakesStuff. The listing was found, but this implementation has not inspected the video/transcript. Video content, speaker attribution, timestamps and exact quotations remain **pending**. A supplied summary is the sole content evidence for the candidate claims below. It does not establish reproduction permission. The page links externally on user choice; no video, lyrics, audio, screenshots or transcript are packaged.

| Candidate paraphrase from supplied summary | Verification needed |
| --- | --- |
| Editing is compared with musical flow; editing reportedly occurred in a car after another shoot. | Speaker and timestamp; what was actually edited and when |
| Movement, breathing and camera pans reportedly align with delivery. | Timestamp and distinction between direction and completed edit |
| Confined-space lighting and equipment concealment influence filming. | Crew attribution and timestamp; do not infer audio/acoustic treatment |
| Clockwork Orange, comic-book and Gollum-like references reportedly inform styling/posture. | Exact attribution and distinction between reference and adaptation |
| Staged choir and kidnapping imagery reportedly represent aspects of the dispute. | Speaker, timestamp and explicit viewpoint label; no finding of theft, rights ownership or another party's motives |

Unanswered: specific plugins, vocal chain, microphone technique, compression settings, beat sourcing/license terms. “Not stated in the supplied summary” is correct. Do not convert visual observation, reaction commentary or this summary into Ren's verified statements. No first-person Ren bot or endorsement is implied.

## Demo

1. Open the lesson and check the evidence question. A named plugin and “no plugins” are unsupported.
2. Keep the original neutral frames and cue; Apply storyboard. A places the cue on beat 2; B on beat 3.
3. Advance four times. Compare the marked frame and cue in both versions. Return to beginning.
4. Change the cue beat to 4 and apply: B wraps to beat 1. Explain how that changes anticipation, without an expected emotional answer.
5. Add a short reflection. Inspect/copy or download JSON. Check that `source.verification` remains `pending`.
6. Keep a manual next-step note in Creator OS. Study JSON is not a PIXIE/Creator Session import. Upload no source media automatically. Roomy and MCP are not connected.

## Export and privacy

`tarantula-visual-study/v1` stores four bounded frame descriptions, original cue, cue beat and optional reflection; includes an unverified source reference and rights-review notice. No identities, timestamps of user behavior, device IDs, scores, recordings or credentials are collected. User text is rendered with `textContent`, not HTML. No import, executable chart data, fetch, local storage or autoplay. Download is explicit; copyable JSON remains available. Publishing is a separate human choice. The application cannot establish authorship or clearance.

## Most useful evidence to add

- Exact video timestamps for editing/musical-flow and movement/camera explanations, with speaker names and short paraphrases. Source timestamp verification must precede any verified lesson answer.
- A small self-authored storyboard or cleared four-beat recording for the demo. Record contributor credit, reuse scope and consent before sharing media.
- An iPad Safari/VoiceOver/Files test record: OS/browser, steps, focus/readout, large text/narrow layout, download/copy behavior and next fix. Avoid personal/health data.
- A de-identified feedback example: what changed between A/B and what next experiment the learner chose. No real community posts are copied without permission.

## Validation boundary

Node tests cover cue placement, unchanged frames, bounds, explicit wrap and pending-source export. Browser testing covers the original default and changed cue, evidence feedback and readable preview. These tests do not verify the source footage, physical devices, real-time sync, rights, iPad or VoiceOver. No changes to Duet, Inpatient, MIDI behavior or Creator OS data schemas.