@aswless_854771076/ai_short_studio_cli 0.1.98 → 0.1.100

package/skills/using-vvicat-ai-short-studio-cli/references/hypit/creation/project-files.md

@@ -0,0 +1,246 @@
+# Project files
+
+Read this when starting, resuming, or reorganizing a video project. The layout below is a working
+convention, not syntax: Hypit resolves ordinary paths and does not assign meaning to these directory
+names.
+
+## Establish the project boundary
+
+Understand the current request and keep its intended result in [Brief](brief.md). Supplied paths,
+the selected production's notes and their explicit references establish which work belongs here.
+A request to continue or revise a named production carries that work forward. A new adaptation
+starts from the supplied reference and requested changes, with its own target and working files.
+When the target is clear, choose a suitable project location and tell the user where the work lives.
+Clarify the intended target when the request leaves genuinely different productions possible.
+
+The current shell directory may be the Skill or executable installation rather than a video project.
+Keep discovery tied to the question being answered:
+
+| What is needed | Where the useful evidence lives |
+| --- | --- |
+| An executable or prepared local service | `hypit paths`, package-manager records, the selected Profile and Provider-documented tool locations |
+| Surface syntax or reusable behavior | Installed vocabulary, the owning package's documentation and a relevant component example |
+| This production's inputs and completed work | Supplied files, its notes, Sources, Runs and selected Result repository, including explicitly linked shared assets |
+
+Follow a path beyond the project when a supplied location, recorded dependency or documented tool
+location explains its purpose. A nearby project with a similar name or subject does not establish
+that connection. If the expected input is missing, report the missing item and resolve its location
+or prepare the new material the commission needs. This keeps discovery useful without turning
+the machine's other productions into an implicit asset library.
+
+Judge connected earlier work by its contribution: a reference, reusable behavior, or material that
+fits the intended person, product, words and presentation. Learning a component from an example
+carries its mechanics into the new work; casting, prompts, assets and accounts remain this
+production's choices. Explain useful reuse alongside what the requested changes require. Finding
+a similar finished video answers an availability question, not whether this commission is complete.
+A proposal to substitute a different existing video belongs in the conversation with the user.
+
+Run commands from the video project's root. A small `package.json` makes that root explicit and
+can later hold the project's component dependencies:
+
+```json
+{
+  "name": "workshop-video",
+  "version": "0.0.0",
+  "private": true,
+  "type": "module"
+}
+```
+
+The CLI uses the nearest `package.json` above the command's working directory, or that directory
+itself when none exists. Passing a Source or Runtime path does not select a different project.
+Sources can live below the root, with file and Source imports relative to the declaring file.
+
+| Boundary | When to set it explicitly |
+| --- | --- |
+| `--workspace <directory>` | Choose the project root for Sources, Runtime selection, project packages and Results when running from another directory |
+| `--asset-root <directory>` | Admit assets stored elsewhere while keeping Source imports in their workspace |
+| `--package-root <directory>` | Resolve project packages from another installation location |
+
+Relative command-line paths start at the shell's current directory. `--workspace` selects the project
+without rebasing the Run, Source or `--runtime` argument. For example, from outside a project:
+
+```bash
+hypit paths --workspace /path/to/video-project
+hypit studio --run /path/to/video-project/build.svrun --workspace /path/to/video-project
+```
+
+`paths` shows the resolved project and where its Runtime selection came from. Source imports and asset
+references inside files remain relative to their declaring file.
+
+For example, `hypit check authors/main.svml --asset-root /path/to/shared-media` admits intentionally
+referenced shared media. Keep the same relevant boundaries for subsequent commands. An ordinary
+project uses its own package installation; reserved `@hypit/*` packages come from the selected
+Distribution. [Distribution](../environment/distribution.md) explains locating that executable,
+and [component vocabulary](../production/vocabulary.md#let-ordinary-package-management-own-distribution)
+explains installing project packages.
+
+## Keep reference truth and target truth separate
+
+A reference archive says what an existing piece does. A production says what the new piece should do
+and contains the files that make it. One reference can inform several productions, and one production
+can draw from several references; ordinary relative links express those relationships without a
+central project index.
+
+```text
+project/
+├── references/
+│   └── <reference>/
+│       ├── source.*
+│       ├── ANALYSIS.md
+│       ├── TIMELINE.md
+│       ├── transcript.json
+│       ├── evidence/
+│       ├── PROGRESS.md          # only while understanding is in progress
+│       └── drafts/              # optional observation experiments
+├── productions/
+│   └── <target>/
+│       ├── BRIEF.md
+│       ├── TREATMENT.md
+│       ├── RECREATION.yaml      # reference-led language, audio, picture-text and shot policies
+│       ├── ASSETS.md            # continuity-bearing inventory, versions and selected setting boards
+│       ├── SHOTS.md             # atomic shots, boards, reference manifests and continuity states
+│       ├── PROGRESS.md
+│       ├── authors/
+│       ├── recipes/
+│       ├── runs/
+│       ├── assets/
+│       └── drafts/
+├── assets/                      # inputs shared by several targets
+├── packages/                    # project Author Packages shared by targets
+├── hypit.runtime.json
+└── package.json
+```
+
+A target may have several Author Sources, Recipe Sources, and Runs. Re-running one Run creates another
+Build, not another target. Keep files for one target together so their relationship is readable from
+the filesystem rather than inferred from matching names elsewhere in the project.
+
+For example, one understood reference can support two distinct target works without being copied:
+
+```text
+references/viral-ad/
+productions/product-version/
+productions/short-version/
+```
+
+Each production links to `../../references/viral-ad/` from its own Brief or Treatment and keeps its own
+Sources and Runs.
+
+The CLI finds the project from the command's working directory as described above; Source and Run files may live in any
+subdirectory inside it. Shared assets and packages can stay at project root. Target-specific assets
+stay with that target. A reference file that is also intentionally used in the new film can be
+referenced in place; its documentary role does not require a duplicate.
+
+## Give each document one job
+
+- `ANALYSIS.md` is the current whole-piece understanding of a reference: form, story, recurring
+  systems, relationships, function, and why the work holds together.
+- `TIMELINE.md` connects what happens when with what it does for the viewer. Preserve the content,
+  placement, entry, changes, persistence and exit of the reference's visual systems, their word or
+  action relationships, source times and useful evidence paths. Concurrent picture, speech, Caption,
+  Typography, MG, Effect and Audio behavior belong to the same connected account. These are reference
+  timeline notes; the target's actual placements belong to the `time:Timeline` declaration in Source.
+- `transcript.json` is word-level speech evidence. It is evidence, not the director's interpretation.
+- `evidence/` contains only media worth reopening, with ordinary human-readable names.
+- `BRIEF.md` preserves the user's goal, facts, constraints, requested changes and
+  [agreed paid scope](brief.md#brief-preserves-user-authority).
+- `TREATMENT.md` is the director's current answer to the Brief: the intended new piece in complete
+  creative terms, before implementation details.
+- `RECREATION.yaml` exists for reference-led work and keeps source/dialogue/subtitle languages,
+  audio sources, permitted picture text, subtitle finishing and shot segmentation as separate fields.
+  Follow [Recreation policy and P0 gates](../playbooks/craft/recreation-policy.md); never collapse these
+  policies into a generation prompt.
+- `ASSETS.md` inventories characters, scenes, props and recurring voices whose identity or state must
+  survive cuts, including unresolved facts, persistent versions and selected setting-board paths.
+- `SHOTS.md` records atomic shot intent, storyboard or previz choice, start/end continuity states and
+  the ordered reference manifest used to compile each final model request.
+- `authors/`, `recipes/`, and `runs/` are the exact production implementation.
+- `PROGRESS.md` is a short photograph of the work now: the live question, what remains to examine or
+  make, next useful action, real blockers, active Build ids and reusable Results. Keep cost information
+  relevant to the remaining work here when it affects the agreed budget. When handing work
+  over, retain the relevant Run and Runtime Profile, exact Build id and public Output names needed
+  to continue. Run Candidates own
+  the actual reuse choices; the note points to them. Established conclusions belong in their owning
+  document instead.
+
+For copyable `ASSETS.md` and `SHOTS.md` fields, the ordered reference table and the ready/blocked gate,
+read [Shot production contracts](../playbooks/craft/shot-production-contracts.md).
+
+Rewrite these files when the current truth changes. They are not logs. `PROGRESS.md` may exist beside
+a reference or a production because either kind of work can span conversations; it does not mark a
+stage and can disappear when there is nothing useful to hand over.
+
+Use these files as the working memory across conversations. Write discoveries and decisions into
+their owning documents as the work develops, preserving both the explanation and its concrete
+details. Let provisional accounts grow with the understanding, keeping detailed evidence once and
+linking to its owner. Share the meaning of discoveries in conversation while the files retain the
+full account. Resume by reading the current project files and relevant Skill pages; use Progress to find
+the next question and the saved evidence, Sources, Runs or Results to continue it. A new discovery
+updates the account it changes so the next session inherits the improved understanding.
+
+## Drafts and Results are different things
+
+`drafts/` holds unpublished authoring material that is not part of the current Source: trial copy,
+temporary designs, or media deliberately exported for further work. A Build output remains in that
+Build's Result even if a later Run does not use it. The absence of a later reference already says it
+was not adopted; it does not need to be copied into `drafts/`.
+
+When an output is deliberately exported into the project, place it according to its new role. It may
+become an `asset`, a `draft`, or direct input to another production.
+
+## Hand over an editable production
+
+A finished MP4 is a viewing deliverable. To continue making the piece, the recipient also needs the
+authored work and the produced values that its Runs select:
+
+- Sources, Recipes, Runs, project notes and referenced input assets, keeping their relative layout;
+- project component source or installed-release dependencies, `package.json`, its lockfile and any
+  tarballs referenced by `file:` dependencies;
+- the completed Results used by `build-record` Candidates, including their media and Composite value
+  documents, plus the project Result repository selection;
+- the intended Runtime configuration, with account access configured on the receiving machine through
+  its own Credential Store.
+
+For the default filesystem repository, preserving `.hypit/results/` intact with the project is the
+straightforward handoff. Include the hidden directory and keep its date/Build subdirectories. Some
+Results forward an Output to its original owning Build, so copying only the latest Build directory
+can omit media still in use. A custom filesystem or S3 repository needs the corresponding files or
+access and an explicit destination selection; changing `hypit.results.json` does not move them.
+
+External input files remain live dependencies, including when referenced inside structured Outputs.
+The Node workspace records their resolved file addresses. When moving to another machine, provide
+those inputs and update affected references to their new locations. Copying Result directories alone
+does not collect external files; an explicit `hypit get` export does collect the selected Output's
+referenced bytes.
+
+The recipient installs the selected Distribution and project dependencies, configures their Runtime,
+and inspects the received Results. Planning the intended Run shows whether its reuse choices still
+resolve and what new work remains. Active execution stays with the originating Runtime until that
+attempt ends; an editable handoff carries authored work and required Results rather than the originating
+Runtime's execution state.
+
+`hypit get` is useful when handing over a particular output. Exported image/video/audio files can be
+selected as file Candidates. A Composite export contains `value.json` and its resource files for
+inspection and transport; that document uses the Result value format, whereas a Run `<value>` accepts
+a StoredValue wrapper. To retain a SemanticTake's structured reuse, keep its Result available and use
+`build-record`. [Builds and Results](../production/builds.md) explains repository selection and export.
+
+## Resume from present facts
+
+After interruption or context compaction, read the current Brief, Treatment, reference archive,
+Source, Recipe, Run, and the short `PROGRESS.md` that applies. Then inspect the project's actual Build
+Results and Runtime status. Continue from usable outputs already present; loss of conversation context
+is not a reason to submit the same paid work again.
+
+A command interruption or missing reply can leave the Worker running. [Builds](../production/builds.md)
+explains how to identify active work and continue through a new Run after a failed attempt.
+Completed Outputs remain in Results without being exported as files. Keep or add their Run Candidates
+when revising downstream work; neither the notes nor an unchanged output name selects them automatically.
+
+## Keep review notes with the work
+
+Studio's `FEEDBACK.json` belongs to the workspace and records timestamped comments for its Runs.
+[Working from Comments](../production/studio.md#revise-from-timestamped-comments) explains reading,
+resolving and preserving these shared notes. Carry settled creative choices into Treatment and
+execution progress into Progress; retain comments as the review conversation.

package/skills/using-vvicat-ai-short-studio-cli/references/hypit/creation/reference-video.md

@@ -0,0 +1,266 @@
+# Understanding a reference video
+
+Read this when a video or link is evidence for what the new work should preserve, adapt, or learn
+from. The useful result is a timecode-level semantic reading: what the piece communicates, and how
+its exact audiovisual choices make that communication work.
+
+For a reference supplied as a link, [video download](../production/video-downloads.md) explains
+`hypit media fetch`, its local preparation and the saved source file used by the tools below.
+
+## Establish the evidence boundary first
+
+Before interpreting the work, record the selected local file, checksum, duration, dimensions, frame
+rate and audio presence. Record what the user has authorized separately: analysis, close recreation,
+identity, voice, music, trademark, dialogue and visual style are distinct permissions. Unknown rights
+remain unknown and do not become permission through analysis.
+
+Give important observations stable evidence ids tied to timecodes, frames or transcript spans. Let the
+hook, beats, turns, payoff and transferable mechanisms cite those observations instead of becoming
+unsupported summaries. Distinguish what may be carried into the target, what must change and what must
+not be copied. A complete analysis accounts for the entire source duration and records any unsampled or
+uncertain interval as a limitation rather than silently treating a few representative frames as full
+coverage.
+
+## Read the whole through its details
+
+Watch the whole reference from opening to close. Follow its hook, argument or story, shifts of
+attention, payoff and intended viewer response. Read the spoken words with their times. Notice who
+establishes the performance, what the pictures contribute, and which Caption, MG, Typography, Effect
+and Audio systems persist or recur. Form a provisional explanation of why the piece works.
+
+Read that meaning through its concrete realization. Each designed element deserves an account of its
+content and appearance, spatial relationship, entry, active behavior, persistence and exit. Locate
+these changes against the words, pauses or actions they serve. Explain what the choice does for the
+viewer: establishes a person, contrasts two claims, accumulates proof, amplifies a reaction, makes
+a process legible, or hands attention to the next idea. Meaning directs the close look; the close
+look supplies the detail needed to recreate the expression. Close reading also reveals purposes and
+connections that the first whole-piece impression missed. An interpretation earns its usefulness
+by explaining the concrete choices; close observation becomes creative understanding when those
+choices connect to the piece's intention. Keep both growing together throughout the reading.
+
+When a choice is puzzling, widen the view: relate it to the surrounding words, nearby events, earlier
+appearances and later consequences. Consider what those connections could explain, then examine the
+relevant sequence. Read the Format or Craft for the relationship you are investigating; use it to
+sharpen the question and recognize what the reference is doing.
+
+A motion's name is a starting point. Inspect its path, scale or opacity change, pace, overshoot or
+settling, and relationship to neighboring events when those make the effect distinctive. Likewise,
+understand a graphic's hierarchy, typography, palette and spacing through what the viewer must read
+first and compare next. Identify the coherent behavior linking the observed frames, including the
+details that make its states and transitions work. A dimmer copy during an exit may be one object's fade, for example; inspect the sequence to
+understand the change.
+
+Whole-piece and close readings revise each other. An object that survives a camera cut belongs to a
+system with a longer lifetime than that shot. Follow its state across the piece: a new entry may
+replace an item, add to a collection, or temporarily cover it. Document the system once and locate
+its changes in time. Camera boundaries help navigation; meaning and continuity determine which
+things belong together.
+
+Speak aloud as the understanding develops. During a close look, explain what the discovered timing,
+placement or motion does for the viewer and how it informs the adaptation. Share representative
+frames or a word-labeled grid when they make that relationship tangible. Carry a provisional reading
+into the conversation, then explain how new evidence deepens or changes it. Articulating the
+connection also helps you notice where your explanation is still incomplete. Analysis and Timeline
+retain the detailed evidence; the conversation lets the user follow the developing interpretation
+and contribute their taste. Routine progress does not require another approval.
+
+Build an account of behavior that can become an editable production. A fixed screen image may be
+the evidence the passage needs; a changing comparison also needs its arrivals, states and handoffs.
+Retain what the objects do together as well as what each looks like. Several implementations can
+express the observed behavior, so the account need not guess the original author's source code or
+component structure.
+The useful result is an explanatory model for this commission: which relationships should remain
+when its person, product or wording changes, and which choices should be redesigned. Record the
+observed behavior before choosing how the target will implement it. A picture, text and motion that
+work together may become one component; repeated appearances may be states of that same object.
+
+Follow the whole reference and account concretely for each distinct system and its changes.
+Repeated behavior can share an explanation, with its variations and handoffs located in time.
+Carry established understanding into direction and composition while pursuing the questions still
+open. Preserve those questions at their source times so the next view can sharpen the account.
+
+## Make time visible
+
+For native reference recreation, review the shot cuts first, then use `hypit recreate transcribe`
+per shot when speech needs transcription. Its StarRouter `qwen3-asr-flash` result is text pending human
+review, without word-level timestamps. Inspect the speech against its shot and confirm the speaker and
+language. When the work needs word-aligned timing, select `hypit transcribe` with an alignment service
+such as WhisperX separately; local WhisperX may require model preparation. Keep either transcript
+beside its source and interpret it with the surrounding picture and supplied context.
+
+Pass a monolingual reference's spoken language explicitly: `--language zh` for Chinese, `en` for English,
+`es` for Spanish, or `ko` for Korean. For multilingual or code-switching material, first identify
+utterance boundaries, cut or extract each utterance, then run `--language auto` on each interval and
+confirm low-confidence results. Store the original words and confirmed language code per turn; the
+video prompt must repeat that code and exact source-language line instead of translating every turn into
+the dominant language or subtitle language. Actual alignment support belongs to the selected service;
+for local execution, prepare that language's resources as described in
+[local tools](../environment/local-tools.md). A Chinese passage can contain English brands and names while still using `zh`.
+Its timed Chinese characters help locate a phrase precisely; group those characters into meaningful
+phrases when describing the reference and writing the new Script.
+
+[Service selection](../environment/model-and-provider.md#choose-the-practical-capability-path-with-the-user)
+explains assessing local preparation and choosing the local or hosted path for this transcription.
+
+WhisperX's recognized spelling reflects the recognizer's interpretation of the audio. Coined words
+and unfamiliar names are especially prone to substitutions, so a spelling difference alone cannot
+establish a pronunciation error. The transcript helps recover the reference's content and locate
+its moments. For the new work, [Script pronunciation](script-and-time.md#write-the-intended-pronunciation)
+expresses the words and readings the generation request should perform.
+
+For speechless work, locate meaning through actions and changes in the scene.
+
+The local `hypit media` commands expose the source at the scale needed:
+
+- `probe` gives duration, dimensions, frame rate and audio presence;
+- `cut` saves a selected passage as a clip;
+- `frames` extracts chosen moments or a range at a chosen interval;
+- `tile` and `tiles` arrange time-labeled frames for inspecting change, with optional word context;
+- `boundaries` locates abrupt visual changes worth inspecting;
+- `fetch` saves a supported video link locally.
+
+For a close reading of motion, `frames --every-frame` and `tiles --every-frame` retain every original
+frame in the selected interval. Each interval is decoded continuously once, then the images are
+paginated at the requested cell size. Actual source timestamps preserve variable-rate footage.
+Use `--ranges` with `tiles` to visit several intervals, and `--transcript` to read their word context.
+
+```bash
+hypit media tiles references/ad/source.mp4 --start 6.8 --end 8.4 --every-frame \
+  --columns 4 --rows 3 --cell 480 --transcript references/ad/transcript.json \
+  --to references/ad/evidence/list-motion
+```
+
+For the production being composed, [snapshot](../production/snapshots.md) captures its existing
+picture directly from Studio or compiled HTML. Align the comparison by the event's meaning and
+then inspect its concrete frames. Coarse and fine readings continue to inform each other: understand
+the whole work, establish the particular behavior, and revisit the whole explanation with that knowledge.
+
+Read grids along the developing explanation. When an aligned transcript exists, use its phrases and
+word times to locate picture changes; `--transcript` can keep those words beside the frames, and
+`--around` can locate a phrase. For a native recreation transcript without timestamps, inspect the
+shot interval directly instead of passing that text as word-timed evidence.
+
+Continuously adjust the viewed range, sampling interval and cell size to the question. Widen the
+range to understand an argument or a persistent system; narrow it and bring samples closer to trace
+an entrance, change or handoff; enlarge a frame to read its typography or spatial detail. These are
+scales of the same investigation. Move between them whenever a detail changes the explanation or
+the explanation directs attention elsewhere. A long reference can be read in manageable passages
+while keeping its whole development clear.
+
+Samples establish what appears at their times, with gaps between them. A wider overview can miss a
+brief event even when neighboring samples look unchanged. Follow the words and visual systems
+through those intervals at a useful density, accounting for their entries, changes, persistence and
+exits. A few representative stills cannot supply that temporal account.
+
+For a project that specifically needs word-aligned evidence, an alignment-service example is:
+
+```bash
+hypit transcribe references/ad/source.mp4 \
+  --language en --to references/ad/transcript.json
+
+hypit media tile references/ad/source.mp4 --start 0 --end 12 --every 1 \
+  --transcript references/ad/transcript.json \
+  --to references/ad/evidence/opening.jpg
+
+hypit media tile references/ad/source.mp4 --start 6.8 --end 8.4 --every 0.1 \
+  --transcript references/ad/transcript.json --columns 4 --cell 480 \
+  --to references/ad/evidence/list-change.jpg
+
+hypit media cut references/ad/source.mp4 --start 6.8 --end 8.4 --label-time \
+  --to references/ad/evidence/list-change.mp4
+
+hypit media frames references/ad/source.mp4 --at 6.9,7.3,7.8 --label-time \
+  --to references/ad/evidence/list-frames
+```
+
+Time labels identify positions in the input media. Give `--transcript` an aligned transcript of that same
+media; its word times share that clock. Word labels sit below the picture so the original Caption
+and MG remain visible. Keep useful evidence under `evidence/` and use descriptive names that make it
+easy to reopen the relevant question.
+
+To inspect a spoken phrase, use `--around "the phrase" --transcript references/ad/transcript.json`
+in place of `--start` and `--end`. `--padding` adds surrounding time; repeated phrases can be selected
+with `--occurrence`. Use `tiles` with `--columns` and `--rows` to page a dense sequence into readable
+grids. Keep the range broad enough to see the incoming and outgoing handoffs.
+
+For Caption, inspect each meaningfully different configuration: speaker treatment, placement,
+emphasis, Cue shape and animation. Follow transitions between configurations too. Repeated uses of
+the same behavior can share its description, with their differing content and times recorded.
+
+When two readings conflict, reopen the relevant source interval and make the disputed detail legible.
+State what is visible or audible separately from what you infer it means. Player controls and other
+viewing context belong to the viewing surface; distinguish them from the designed video content.
+
+## Preserve a connected account
+
+Write the understanding into the project as it develops. Preserve both the whole-piece explanation
+and the detailed reading, with source times and useful evidence paths so someone can pick up the
+same work from the files. Include what a choice does for the viewer alongside how it appears.
+
+`ANALYSIS.md` carries the whole-piece model: what the work is trying to achieve, how its story and
+pacing work, what each visual or sound system contributes, which systems persist or recur, and how
+distant moments relate. Keep the reference's facts and your interpretation distinguishable in
+ordinary prose.
+
+`TIMELINE.md` carries time-locatable realization. Organize it into sections named by source-media
+time and a meaningful phase. Within each, connect the active objects and their detailed behavior to
+the words or actions they serve. An account should let someone find the event, understand its
+expression and implement an appropriate counterpart. For example:
+
+```md
+## 6.07–8.27 · The workload accelerates
+
+The spoken list reaches “videos / voiceovers / ads / scripts”. Each noun brings a new full-frame
+illustration and a marker-style word at its center, replacing the preceding pair. The lower spoken
+Caption gives way to these central labels so each example reads as one unit.
+
+The word and picture enter together on each noun; the labels pop to size, settle briefly, and leave
+with their picture on the next cut. The increasingly short holds make the workload feel excessive.
+The 6.8–8.4 clip and list-change grid show the handoffs; word times are in transcript.json.
+```
+
+These sections can overlap and contain finer timed notes where the work is dense. A persistent
+title, board, sound bed or Caption system can span several sections; refer to the same system and
+describe what changes. Keep exact text, meaningful colors, positions, motion phases and timings where
+they are needed to understand or reproduce the design. Reopen the media for facts still in doubt.
+
+Use production vocabulary by function. A-roll establishes the semantic performance even inside a
+small inset or behind a full-frame B-roll. Caption displays the speech; Typography carries independent
+writing. MG and UI can contain text of their own. A sound can support a continuing argument across
+several picture changes. Name the roles that clarify this particular work.
+
+## Turn understanding into new direction
+
+Source time locates evidence. Record both the original seconds and the expressive relationship:
+a reveal answers a question, an image illustrates a phrase, an exit makes room for the next claim,
+or an audio handoff begins the next speaker under the previous picture.
+
+The user's request determines how those relationships should live in the new piece. A different
+person or product may change the argument, copy, number of examples, graphic content, placement and
+duration. Use [transformations](transformations.md) to think through that adaptation. Prefer
+Selections for meaningful spans and Moments for events in the target Script, and let the accepted
+performance establish their time. [Script syntax](../production/script-syntax.md) owns that authoring language.
+
+For generated camera imagery, [image direction](../playbooks/craft/image-direction.md#two-ideas-behind-every-sentence)
+turns the observed appearance into a coherent styling and scene direction, with the decisive details
+the target needs. The reference notes retain the observations behind those choices.
+
+For the designed picture, use [component design](../production/component-design.md) to turn the
+target's relationships into useful objects, parameters and events. A system can span several shots;
+one shot can contain several independent systems. Choose the new component boundaries from shared
+behavior, with the details required to realize it. [Direction and its inputs](../production/system.md#give-each-part-the-direction-it-can-realize)
+connects the whole creative design to its separate material and composition instructions.
+
+The recorded account should explain the whole piece from opening to close and make its distinct
+visual systems and their changes locatable. It should be concrete enough to direct the new work:
+what to preserve or adapt, how it is expressed, and why it belongs. Continue investigating an
+unexplained relationship where it could change that direction.
+
+Keep the current question, passages or systems still to examine, and next useful action in
+`PROGRESS.md`. Write discoveries into Analysis and Timeline while they are fresh, and revise those
+accounts when the reading changes. When resuming, read these files and reopen the source at the
+recorded locations. The user's goal remains in Brief and your new design in Treatment.
+
+When the intended adaptation depends on a performed, physical or camera relationship unfolding in
+time, preserve a useful source excerpt as temporal evidence for [reference-directed generation](../playbooks/craft/video-direction.md#let-footage-carry-temporal-behavior-that-matters).
+The analysis explains what matters; the footage can carry that behavior into the model request.