@lincyaw/skills 1.0.0-rc.89 → 1.0.0-rc.90
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@lincyaw/skills",
|
|
3
|
-
"version": "1.0.0-rc.
|
|
3
|
+
"version": "1.0.0-rc.90",
|
|
4
4
|
"description": "Presmium's bundled authoring, animation, presentation and plugin-development skills for coding agents",
|
|
5
5
|
"license": "Apache-2.0",
|
|
6
6
|
"repository": {
|
|
@@ -36,7 +36,7 @@
|
|
|
36
36
|
],
|
|
37
37
|
"dependencies": {
|
|
38
38
|
"jiti": "^2.7.0",
|
|
39
|
-
"@lincyaw/kernel": "1.0.0-rc.
|
|
39
|
+
"@lincyaw/kernel": "1.0.0-rc.90"
|
|
40
40
|
},
|
|
41
41
|
"presmium": {
|
|
42
42
|
"skills": [
|
package/skills/index.json
CHANGED
|
@@ -12,7 +12,7 @@
|
|
|
12
12
|
"references/publishing.md",
|
|
13
13
|
"references/writing.md"
|
|
14
14
|
],
|
|
15
|
-
"version": "1.0.0-rc.
|
|
15
|
+
"version": "1.0.0-rc.90",
|
|
16
16
|
"hash": "0c47458d29bb2028b093409c03087dc1a20340270525a452d4f771c4257812f9"
|
|
17
17
|
},
|
|
18
18
|
{
|
|
@@ -536,6 +536,7 @@
|
|
|
536
536
|
"gallery/transitions/scenes/motion/wipe.ts",
|
|
537
537
|
"gallery/transitions/scenes/remotion/series-transition.ts",
|
|
538
538
|
"gallery/transitions/scenes/remotion/wipe.ts",
|
|
539
|
+
"references/acceptance.md",
|
|
539
540
|
"references/lengths.md",
|
|
540
541
|
"references/libraries.md",
|
|
541
542
|
"references/procedural.md",
|
|
@@ -545,8 +546,8 @@
|
|
|
545
546
|
"references/team.md",
|
|
546
547
|
"references/timeline.md"
|
|
547
548
|
],
|
|
548
|
-
"version": "1.0.0-rc.
|
|
549
|
-
"hash": "
|
|
549
|
+
"version": "1.0.0-rc.90",
|
|
550
|
+
"hash": "3a2f0fb11e1564895c0eef4a7dd9ae4e17d0449202f867f2e22e25fad6009a9b"
|
|
550
551
|
},
|
|
551
552
|
{
|
|
552
553
|
"name": "presmium-develop",
|
|
@@ -556,7 +557,7 @@
|
|
|
556
557
|
"references/plugin.md",
|
|
557
558
|
"references/testing.md"
|
|
558
559
|
],
|
|
559
|
-
"version": "1.0.0-rc.
|
|
560
|
+
"version": "1.0.0-rc.90",
|
|
560
561
|
"hash": "96422f5cef6a97bba23c9c67b7fb0ee491b18f01f789e262ca24b274e7fabaa1"
|
|
561
562
|
},
|
|
562
563
|
{
|
|
@@ -566,7 +567,7 @@
|
|
|
566
567
|
"SKILL.md",
|
|
567
568
|
"references/publication.md"
|
|
568
569
|
],
|
|
569
|
-
"version": "1.0.0-rc.
|
|
570
|
-
"hash": "
|
|
570
|
+
"version": "1.0.0-rc.90",
|
|
571
|
+
"hash": "dc37b5117120adbc5fc16afc67da8ced0d91992be88751996df133cca596b91c"
|
|
571
572
|
}
|
|
572
573
|
]
|
|
@@ -27,6 +27,8 @@ A teaching segment, an explainer or paper video, a promo, a product demo and a d
|
|
|
27
27
|
|
|
28
28
|
[Story](references/story.md) is the process, each stage ending in a check: brief, thesis, central picture, three storylines for the person to pick, a storyboard of rest frames with what carries each boundary, the script, an animatic the person approves, motion, review against the spec and with cold viewers. [Storyboards](references/storyboards.md) has filled-in examples for popular science, papers, courses, data stories and products; [Scenarios](references/scenarios.md) the usual pictures for each subject.
|
|
29
29
|
|
|
30
|
+
For a complete teaching explainer or paper video, read [Planning and acceptance](references/acceptance.md) before scene code. Save `storyboard.md` with both the step sequence and the element design tables: each object's meaning, visual encoding, start/end states, operation, narration trigger, result timing, evidence and continuity. Rest-frame keywords alone are not a motion plan. Each core explanation must visibly establish the situation, perform the operation, expose its result and let it be read. If a reference work is specified, analyse its operations and object/spatial continuity before adapting its look. Automatic workflows choose and review internally; honour a person's explicit preview request without adding routine approval stops.
|
|
31
|
+
|
|
30
32
|
A piece too large for one agent is made from one storyboard file every agent reads, with a term table that gives each term its definition and first page, shared scene modules that only the coordinator changes, edits through the MCP author actions, a reading of its own words every agent does against the storyboard, review rounds appended to the storyboard and a cold read by a fresh agent in three passes: [Team](references/team.md).
|
|
31
33
|
|
|
32
34
|
## 5. Motion, rhythm and sound
|
|
@@ -45,6 +47,8 @@ Narration is structured in `narrations`; steps name narration IDs. Cue targets u
|
|
|
45
47
|
|
|
46
48
|
Validate the edited document (`presmium check` or `check_document` also bundles each scene and reports one that doesn't compile, with its file and line), then look at the actual result. With the MCP server, `capture_slides` gives the rest frame of each step of a page and, with `at`, frames that many milliseconds into a step's animation, drawn by the step clock as a lecture export draws them, so they are reproducible; `measure` returns the boxes of elements (a CSS selector) in slide pixels, for positions and overlaps. Capture the steps in any order you like, since each picture is made from its step and time alone; `arrive: 'backward'` checks that going back to a step shows its frame, and `capture_reading` shows the scene on the reading page, also at a phone width (390). Without the MCP server, build only the relevant form, open it, jump forward and backward in arbitrary order, and inspect the static fallback if it matters (exports print it). Do not claim an animation works merely because JSON validation or compilation succeeded, and do not write browser scripts for pictures the tool takes.
|
|
47
49
|
|
|
50
|
+
For complete explanations, apply the separate fact, layout, runtime and explanatory-motion checks in [Planning and acceptance](references/acceptance.md#4-review-four-different-things). Review the full timed explanations, beat transitions and arbitrary seeks. A contact sheet verifies stills; it cannot pass motion. The cold viewer must retell which object was operated on, what changed and how the result follows. Static cards, a whole figure with all conclusions permanently visible, or a shared entrance preset used instead of the core operation fail this check even when the data and typography are correct.
|
|
51
|
+
|
|
48
52
|
```sh
|
|
49
53
|
presmium check lesson --json
|
|
50
54
|
presmium dev lesson
|
|
@@ -0,0 +1,77 @@
|
|
|
1
|
+
# Planning and acceptance for explanatory animation
|
|
2
|
+
|
|
3
|
+
Read this before implementing or accepting a complete teaching explainer or paper video. It adds the object design and dynamic review that a rest-frame storyboard cannot supply. A logo sting, a deliberately static title card or a single emphasis effect does not need a worked mechanism; its surrounding explanation does.
|
|
4
|
+
|
|
5
|
+
## 1. Extract the reference before adapting it
|
|
6
|
+
|
|
7
|
+
When a person specifies an earlier work, inspect its actual scene code and timed output before deciding the new piece's look. A still or a cover cannot establish its motion style. Record in `storyboard.md`:
|
|
8
|
+
|
|
9
|
+
| Reference passage | Object and meaning | What the operation demonstrates | Space, reading order and camera | What survives into the next beat | Rule to adapt |
|
|
10
|
+
|---|---|---|---|---|---|
|
|
11
|
+
|
|
12
|
+
Describe operations such as a sample being selected, the same point moving to a new encoding, a constraint changing a path, or a result being measured against a fixed baseline. Record entrances, exits and pauses as part of that explanation. Palette, typography and branding are a separate part of the comparison.
|
|
13
|
+
|
|
14
|
+
Adapt the rules to the new subject. Do not copy a sequence that no longer explains its evidence, add arbitrary motion to match a cut count, or treat one series' reference as the only acceptable style for all animation. If the timed output is unavailable, state that limitation and use the source timeline as partial evidence rather than claiming the visual match was verified.
|
|
15
|
+
|
|
16
|
+
## 2. Plan the elements and their operations
|
|
17
|
+
|
|
18
|
+
Before scene code, save the step sequence and both linked tables below in `storyboard.md`. Resolve them for every core explanation, including figures, labels, counters, connections, highlights and camera regions that affect understanding. Group repeated objects when they share the same rule; identify the objects whose individual identity matters. This is a design record: teaching text, parameters and step bindings still belong in `lecture.json`.
|
|
19
|
+
|
|
20
|
+
**Element design** establishes what a viewer is following:
|
|
21
|
+
|
|
22
|
+
| Stable object ID | Meaning and visual form | Colour role and spatial convention | Initial state | Data or illustrative basis |
|
|
23
|
+
|---|---|---|---|---|
|
|
24
|
+
|
|
25
|
+
**Operation design** establishes how each explanation happens:
|
|
26
|
+
|
|
27
|
+
| Step / semantic beat | Object IDs | Operation, path or morph and its meaning | End state | Narration phrase, cue or state-time fraction | When the result appears | Objects carried into the next beat |
|
|
28
|
+
|---|---|---|---|---|---|---|
|
|
29
|
+
|
|
30
|
+
- Give objects stable semantic identities, such as `training-tray`, `candidate-trajectory-2` or `reference-line`. Use scoped selectors in the implementation; these planning IDs do not require SVG `id` attributes.
|
|
31
|
+
- Specify the operation, not merely “animate”: which candidates remain after a filter, where one token travels, which coordinates or shape change, what a brace measures, or which source region a camera frames. Give the end value or geometry and its evidence.
|
|
32
|
+
- Identify the phrase or cue that causes the operation. Without word alignment, author a fraction of that state's duration and verify it against actual narration; see `timeline.md`. Do not apply the same fractions to every scene regardless of its words.
|
|
33
|
+
- State the result's reveal condition: after the last selected item lands, after the measurement reaches the baseline, or alongside the newly framed evidence. Readouts must agree with the state being shown. An observed experimental score can be revealed after its configuration is established; do not pretend to compute it from an illustrative token count.
|
|
34
|
+
- State what persists across the boundary and where it is at the end and start. When a cut is intentional, explain what the viewer must retain. Replacing the picture with another complete diagram does not establish object continuity.
|
|
35
|
+
- Use a short “held” entry for intentionally static branding, source text or reading time. A table of final-frame keywords with unspecified actions is incomplete and must be resolved before implementation.
|
|
36
|
+
|
|
37
|
+
Before polishing, make a rough timed preview of the core operations. The strip of resting frames checks structure; the moving draft checks whether the chosen objects can actually explain the claim.
|
|
38
|
+
|
|
39
|
+
## 3. Show the operation, not only the final answer
|
|
40
|
+
|
|
41
|
+
Each core explanation has a visible **setup → operation → result → hold**. A meaningful operation can contain several semantic beats: first isolate the candidates, then filter them, then carry the survivors into a count. Choose the number and pace from what the viewer must understand; do not require an identical number of beats, constant movement or one animation preset per paragraph. A deliberate reading hold is useful once the relevant operation and result have been shown.
|
|
42
|
+
|
|
43
|
+
Use a concrete case before aggregating. Keep its important objects through the changes, then show how the same rule applies to a set or a chart. Explain how a representation changes: if a token becomes a chart point, retain its colour and identity through the move. A new representation should not require the narration to rebuild the relationship from scratch.
|
|
44
|
+
|
|
45
|
+
For a paper figure, guide the eye through **axes and encoding → current object → comparator → conclusion**. Frame or highlight the region currently being discussed, keep the necessary baseline visible, and move the focus when the explanation moves. The original figure remains evidence; overlays and camera work must preserve its values, axes and uncertainty. If reconstructing a data view, use sourced values and make the correspondence to the original explicit. A camera zoom without a reading purpose is decoration.
|
|
46
|
+
|
|
47
|
+
| Fails as the core explanation | Corresponding explanatory treatment |
|
|
48
|
+
|---|---|
|
|
49
|
+
| Static cards say “one trajectory”, “more trajectories” and “better score”; the voice supplies the entire process | Follow candidates from one question into a training tray, visibly select the relevant set, update its sourced aggregate, then locate that configuration's score against the fixed reference |
|
|
50
|
+
| A whole paper figure and all conclusions are visible from the first frame while the voice reads them | Establish the axes, isolate the first relevant column, carry the highlight to the next column, measure their relationship to the reference, then reveal the conclusion and hold |
|
|
51
|
+
| Every page draws a line, fades in a group and pops up all numbers at the same fractions | Design each operation from its meaning: filter candidates, move a value to a new scale, extend a curve from observations, or change a constraint and show the resulting path |
|
|
52
|
+
| A dot grid fills randomly while a counter changes, with no explanation of what a dot means | Introduce one item and its mapping, apply the same rule to the set, then add only the newly covered items and update the matching count; mark an illustrative mapping as such |
|
|
53
|
+
| A new complete diagram replaces the previous one at each step | Retain the relevant objects and their positions, transform or move them into the next configuration, and change only what the new comparison controls |
|
|
54
|
+
| The logo, colours and title card match the reference, but the explanation is still a slide deck | Adapt the reference's way of operating on objects, staging comparisons and carrying identity across shots, as recorded in the reference table |
|
|
55
|
+
|
|
56
|
+
Entrance effects can support these operations. They cannot substitute for them. More animation, more steps or a longer timeline is not evidence of better explanation.
|
|
57
|
+
|
|
58
|
+
## 4. Review four different things
|
|
59
|
+
|
|
60
|
+
Report these results separately in the storyboard and delivery evidence. A pass in the first three does not imply a pass in the fourth. Mark missing observations as unverified rather than turning compilation or a contact sheet into an overall approval.
|
|
61
|
+
|
|
62
|
+
| Review | Evidence and question |
|
|
63
|
+
|---|---|
|
|
64
|
+
| Facts | Are the values, units, baselines, uncertainty and conditions supported? Are illustrative objects distinguished from measured data? |
|
|
65
|
+
| Layout | Are essential objects and labels legible at the delivery size, with captions enabled, without overlap or accidental cropping? |
|
|
66
|
+
| Runtime | Does the saved version compile and play with the intended audio and timing? Do direct, backward and arbitrary seeks reconstruct the same states without callback history or missing objects? |
|
|
67
|
+
| Explanatory motion | Does the viewer see the planned operation happen, follow the same objects through it, and understand how the visible result follows? Does the motion and spatial continuity match the chosen reference rules? |
|
|
68
|
+
|
|
69
|
+
For explanatory motion:
|
|
70
|
+
|
|
71
|
+
1. Review each core explanation's full timed preview at the intended pace, including the setup, operation, result and hold. Check its complete time range rather than only an entrance and a middle frame. If continuous playback cannot be observed, report that limitation; timed frame samples and source inspection provide partial evidence, not proof that playback was watched.
|
|
72
|
+
2. Sample the start, the before/during/after states of each planned operation and the settled result using native `capture_slides` with `at`. Choose times from the beat plan, not the same fractions for every shot. Compare against actual narration; a correct final picture does not repair a premature answer or an unexplained transformation.
|
|
73
|
+
3. Inspect both sides of every core boundary. Check that carried objects are recognizable in position, scale, colour and meaning; identify intentional cuts. Inspect the exported video around those boundaries as well as the preview.
|
|
74
|
+
4. Select steps directly and in a different order, seek into their operations and return backward. Compare the same timestamps. Stable outcomes and intermediate states must come from step and time alone.
|
|
75
|
+
5. Give an independent cold viewer the timed material and narration, without the source or explanatory notes. Ask: “Which object was being followed? What was done to it? What changed as a result? What comparison or rule supports that result?” Have them retell the chain and predict an appropriate next change. A viewer who repeats the captions or calls it “clear” has not demonstrated this understanding.
|
|
76
|
+
|
|
77
|
+
A contact sheet is useful for facts, layout and the broad story. The still-frame cold read in `team.md` remains an early check; supplement it with the dynamic review above before accepting a complete explanatory animation. Record specific unresolved moments by timestamp and object, the intended operation and the required change. Fix a missing mechanism at the storyboard stage, not by adding effects to its finished card.
|
|
@@ -14,11 +14,11 @@ When the person explicitly requests automatic completion or no stops, choose the
|
|
|
14
14
|
| 1. Thesis | "看完后,观众明白 X,因为他们看到了 Y" (After watching, the viewer understands X because they saw Y), X in 25 汉字 or fewer; the same claim as And–But–Therefore | Someone who has seen nothing restates it; the ABT has exactly one But |
|
|
15
15
|
| 2. Central picture | Three candidate pictures; keep the one in which X becomes a visible property that changes. Draw the money frame: the one still that proves X | Shown for 3 s to a cold viewer, the money frame alone gets something close to X back. If it needs a caption, keep looking |
|
|
16
16
|
| 3. Shape | Three storylines that differ in the idea the picture is made of (one element transforming through everything, one camera move through scale, a chain reaction, a split that becomes one), not three orders of the same cards; one of them bold. Each with its shape, its form (below), its first 2 s, its beats, what carries every boundary and what it rules out, look and sound; never the last piece's look or a gallery film's beats by default. Recommend one. Then for the chosen one: where its turn falls (usually 40–60% in); one spatial convention; one meaning per colour; the misconception to confront and the parts to name first; a term table: each concept's one name and the synonyms not to use | The person picked a storyline (stop and wait, unless they asked for the file or for no stops); the turn is a change in the picture, not a sentence |
|
|
17
|
-
| 4. Storyboard | One row per step
|
|
17
|
+
| 4. Storyboard | One row per step plus the element design and operation tables in [Planning and acceptance](acceptance.md): initial situation, visible manipulation, result, hold, and the objects carried across boundaries | The strip reads silently, and each core explanation has an executable object-and-operation plan; keywords or final frames alone do not pass |
|
|
18
18
|
| 5. Script | The narration of each step; when its words come; cues | Read aloud with a timer: each step's narration fits its state, or the state is lengthened on purpose |
|
|
19
|
-
| 6. Animatic | The
|
|
19
|
+
| 6. Animatic | The rest-frame strip with narration for the overall structure, plus rough timed operations for the core explanations; build and export | Timing and structure pass; the rough operations make the argument visible. A contact sheet alone cannot approve explanatory motion. The person approves the requested preview, or the automatic workflow reviews it internally |
|
|
20
20
|
| 7. Motion | The movement between rest frames, one transition at a time, by `lengths.md` | Every motion's why says what it means (cause, the same object, converging); delete the ones whose why is "looks nice" |
|
|
21
|
-
| 8. Review | Cold viewers, cheapest test first (below)
|
|
21
|
+
| 8. Review | Cold viewers, cheapest test first (below), then full timed explanations, boundaries and arbitrary seeks by [Planning and acceptance](acceptance.md#4-review-four-different-things); finally the export and delivery note | Understanding and explanatory motion both pass; viewers can retell the object, operation and result as well as the thesis. Export the final after the requested creative review, or internal review in an automatic workflow |
|
|
22
22
|
|
|
23
23
|
**Kinds of piece.** Name the kind in the brief; it sets the defaults below, and the person's brief or a reference overrides any of them. A piece that mixes kinds takes each part's row.
|
|
24
24
|
|
|
@@ -97,13 +97,13 @@ The strongest pieces in the gallery have one picture that is the argument: in th
|
|
|
97
97
|
|
|
98
98
|
**Forms** say how shots join, whatever the shape: one continuous take; a chain where each shot's last object becomes the next one's first; an endless zoom; match cuts on shape or motion; a split screen that becomes one; one frame that rebuilds itself. A film uses one form throughout.
|
|
99
99
|
|
|
100
|
-
**From a reference.** When the person brings a clip whose feel they want,
|
|
100
|
+
**From a reference.** When the person brings a clip whose feel they want, analyse it before building: the share of still frames, cut times and bursts (the ffmpeg commands in `lengths.md`, Build and export), the largest element's size, and what moves inside a beat. For an explanation, also record the recurring objects, their start/end states, each operation's meaning, the camera's reading order and what survives each boundary; use [the reference comparison](acceptance.md#1-extract-the-reference-before-adapting-it). Matching its logo, palette and title card does not match its teaching animation. Take the rules that explain the requested feel, adapting the sequence to the new subject rather than copying its cuts or inventing data to fit them.
|
|
101
101
|
|
|
102
102
|
Paper videos are designed to the venue's clock (SIGGRAPH's Fast Forward 20 s narrated or 10 s silent, a CHI preview 30 s, ECCV summaries 60–120 s, a Three Minute Thesis 3:00, journal video abstracts 4–5 minutes), say one thing to remember, show the baseline in the same frame and tag every number with its table. Result first suits short pieces and method or systems papers, not theory (open with the puzzle), clinical abstracts (the design comes first, because credibility depends on it) or a thesis talk (a human question). By length: 10 s is ABT compressed into setup, change, result, hold; 60 s is ABT, puzzle first or predict first; 3–6 minutes is 起承转合 or a chain of causes inside SCQA, with the misconception at the turn; promos use problem–agitate–solve, the sparkline or the heartbeat; data uses one picture over time shaped establisher → initial → peak → release (Amini et al., CHI 2015).
|
|
103
103
|
|
|
104
104
|
## 4. Storyboard the rest frames first
|
|
105
105
|
|
|
106
|
-
In Presmium a step rests on the final frame of its state, and that frame is also the slide and the still on the web page.
|
|
106
|
+
In Presmium a step rests on the final frame of its state, and that frame is also the slide and the still on the web page. The rest-frame strip anchors the storyboard; a complete teaching explanation also needs the element design and semantic beats that get it from one anchor to the next. Write those in `storyboard.md` before implementing scenes, using [Planning and acceptance](acceptance.md#2-plan-the-elements-and-their-operations). Keep teaching content and executable parameters in `lecture.json`; the storyboard records their design and review rather than becoming another input that regenerates the document.
|
|
107
107
|
|
|
108
108
|
- **One step = one state = one idea**: one or two sentences of narration over 6–15 s in a teaching segment, less in a promo. Mayer and Chandler's lightning lesson in 16 segments of one or two sentences, each ended by a Continue button, is the model: segmenting d = 1.36, naming the parts first d = 1.00 (Mayer and Moreno 2003, <https://www.uky.edu/~gmswan3/575/9_ways_to_reduce_CL.pdf>). On the web and on slides the reader is that button; in the video the hold after a payoff stands in for it.
|
|
109
109
|
- **Every rest frame makes sense alone**: a 3-second glance tells what it shows (Duarte's glance test; Disney's staging, checked in silhouette). The exception is a prediction step, whose still is an open question with options on purpose; its answer is the next step, and nothing in it gives the answer away (no updated label, no highlighted edge, no output, no blank shaped like the answer).
|
|
@@ -126,7 +126,9 @@ D. Steps | # | State | Start–end | Rest frame | What moves (one thing; caus
|
|
|
126
126
|
| Words come (with / after / cue) | On-screen text (one line at most) |
|
|
127
127
|
| Reads (what the eye reads, in order, each with its time window; windows never overlap) | Sound |
|
|
128
128
|
| Carries into the next (the element that survives the boundary and what it does, or "cut" and why) | Why |
|
|
129
|
-
|
|
129
|
+
ELEMENTS + OPERATIONS: fill the linked design tables in acceptance.md; a list of captions or rest frames is insufficient
|
|
130
|
+
E. Checks durations add up; the thumbnail strip reads silent; read-aloud timing; rough operations screened;
|
|
131
|
+
separate fact/layout/runtime/explanatory-motion results; full timed explanations and boundary/seek review
|
|
130
132
|
```
|
|
131
133
|
|
|
132
134
|
Put the filled worksheet in the piece's reading-view notes, as the gallery does: the next author sees why each beat exists. When several agents make one piece, the worksheet grows into the storyboard file they all work from (`team.md`).
|
|
@@ -139,7 +141,7 @@ Put the filled worksheet in the piece's reading-view notes, as the gallery does:
|
|
|
139
141
|
- **Speak the words; don't print them over moving pictures.** Narration with the picture beats text with it (modality d = 1.17); narration plus the same text on screen is worse than narration alone (redundancy d = 0.69). One or two key words beside the part they name are fine (spatial contiguity); captions are a separate track, not on-screen text.
|
|
140
142
|
- **Say it while it happens** (temporal contiguity d = 1.30): the narration and its motion share a step. `narrationStart: "with-animation"` (the default) for a mechanism; `"after-animation"` when the picture should land first and the words name it (a result, a surprise); a narration cue on the key word (`point` or `spot` to aim the eye, a pause timed in `production.cueTimings`) when the words lead.
|
|
141
143
|
- **Questions turn the acts**: "那有没有可能……直的呢?", "还剩最后一个问题". A 1–2 s silence before the turn, and 6–10 s with a spoken "先暂停,写下答案" after a prediction question in a video (the web view's prediction block waits by itself).
|
|
142
|
-
- **Not a slideshow.** A step whose parts are all on screen in its first quarter and then freeze reads as a slide:
|
|
144
|
+
- **Not a slideshow.** A step whose parts are all on screen in its first quarter and then freeze reads as a slide. In an explanation, showing labels later is only staging: the objects must also undergo the operation the words explain. Plan its semantic beats by [Planning and acceptance](acceptance.md#3-show-the-operation-not-only-the-final-answer), without adding movement merely to keep the screen busy. In a promo whose voice runs longer than its pictures, trim the script; don't stretch the motion.
|
|
143
145
|
- **Weed**: interesting facts off the causal chain, background music under a dense explanation, decorative clips (coherence d = 0.90). A tangent comes back to the question within its step or moves to the reading page.
|
|
144
146
|
- **Read d values as direction, not dose**: a 2025 meta-analysis of Mayer's studies found an average g = 0.37, smaller for primary-school and graduate learners, and the modality benefit narrows when learners set the pace.
|
|
145
147
|
|
|
@@ -170,7 +172,7 @@ Interactive explainers let the reader drag; a lecture video can't. Script the ma
|
|
|
170
172
|
|
|
171
173
|
## 9. Review from a cold seat
|
|
172
174
|
|
|
173
|
-
Two reviews, different seats. **Against the spec**: a fresh agent gets the brief, the requirements table, the LOOK line, the storyboard and the rules, but not the reasons given while making it, and returns each rule as met or not with
|
|
175
|
+
Two reviews, different seats. **Against the spec**: a fresh agent gets the brief, the requirements table, the LOOK line, the storyboard and the rules, but not the reasons given while making it, and returns each rule as met or not with evidence; an input it lacks is "can't verify", and each fix goes back to the stage it belongs to. For complete explanations, separate facts, layout, runtime and explanatory motion, and review the full timed material and operation boundaries by [Planning and acceptance](acceptance.md#4-review-four-different-things). **Against understanding**: the cold viewers below. Locate vague feedback ("节奏不对") in the actual material before changing it; a specific example from the person is already enough to investigate without another question.
|
|
174
176
|
|
|
175
177
|
Cheapest first; never ask "看懂了吗" (people say yes when they haven't), and never review with the team that made it: people tapping a song's rhythm expected listeners to name half the songs, and listeners named 3 of 120 (Heath and Heath, *Made to Stick*).
|
|
176
178
|
|
|
@@ -188,7 +190,7 @@ Before showing anyone, review the export yourself. Make the sheets with ffmpeg f
|
|
|
188
190
|
|
|
189
191
|
Hand the piece over with a delivery note in its reading view: the creative calls made, the storylines not built, what was measured and what was only looked at, the weakest shot, and the warnings kept on purpose.
|
|
190
192
|
|
|
191
|
-
A cold viewer saying "clear" is not a pass: Muller's clear video raised confidence and taught nothing. A changed prediction and a retold thesis are. A subagent that sees only
|
|
193
|
+
A cold viewer saying "clear" is not a pass: Muller's clear video raised confidence and taught nothing. A changed prediction and a retold thesis are. A subagent that sees only exported rest frames is a cheap first check of the story and words; it cannot approve explanatory motion. Give the dynamic reviewer the timed material and narration and require a retelling of the object, operation and resulting comparison, as well as the thesis. Fix the first step where anyone got lost and screen again. `team.md` (section 7) has the still-frame prompt, picture preparation and terminology checks; supplement them with [the dynamic acceptance review](acceptance.md#4-review-four-different-things).
|
|
192
194
|
|
|
193
195
|
## The worked example: flow matching
|
|
194
196
|
|
|
@@ -30,13 +30,15 @@ Presmium 原生 MCP 是创建、编辑、检查、配音、导出和截图的首
|
|
|
30
30
|
|
|
31
31
|
## 2. 制作动画
|
|
32
32
|
|
|
33
|
-
读取 `presmium-author` 的写作与文稿规范,`presmium-animate`
|
|
33
|
+
读取 `presmium-author` 的写作与文稿规范,`presmium-animate` 的叙事、时间线、对应动画范例,以及[解析动画的设计与验收](../presmium-animate/references/acceptance.md)。遵从本系列工作区的固定片头、结尾、色彩和来源页规则。
|
|
34
34
|
|
|
35
|
-
-
|
|
35
|
+
- 编码前先写 `storyboard.md`:中心论点、术语、数字依据、步骤,以及链接规范中的元素设计与操作表。逐项写清对象身份、视觉语义、初末状态、操作路径或变形、讲稿触发点、结果出现时机及跨步保留关系;只列休止帧关键词不算完成分镜。内部选择主线并完成预演检查,不恢复逐阶段的例行用户确认;用户指定预览时遵从。
|
|
36
|
+
- 有指定参考作品时,先拆解其实际动作、空间和对象连续性,再设计新片。在本仓库的 `examples/papers/` 系列,第一期 `papers/hinted-scaling-laws/` 是逐步解析方式的参考:读其场景时间线并检查动态画面,不只复用品牌。其他工作区采用其指定参考,不把这一篇当作所有动画的唯一风格。
|
|
37
|
+
- 每个核心解释都展示设置、操作、结果与停留,按内容安排语义节拍;概念由具体对象和可见操作引入再命名,结果交代基线、指标和关键条件。静态卡片、原图加全部结论常亮、通用入场统一套用不能替代解析演示。
|
|
36
38
|
- 优先用论文原图做镜头、标注和对照,动画表达机制或因果,不堆说明卡片。无法复算的数字引用论文明确位置;示意与估算在教学画面中标明。
|
|
37
39
|
- 通过 MCP 保存当前文稿和场景,检查并看静态联系表及过渡帧。连续场景的主体保持存在,新增信息才入场;步骤能任意跳转和回退。
|
|
38
40
|
- 片头从完整系列封面开始,开场配音保留;用页面级 `captionVisible: false` 隐藏片头字幕,正文恢复字幕。不删讲稿或把字幕隐藏当成静音。
|
|
39
|
-
-
|
|
41
|
+
- 预演自检通过后继续制作。协调者统一编辑文稿,场景文件明确分工。联系表冷读检查事实、词语与布局;独立的动态审阅另看完整核心段落、操作边界与任意 seek,要求冷读者复述对象、操作及结果的关系。四类验收分别记录,不能用静态审阅或技术通过代替解析动画质量通过。
|
|
40
42
|
|
|
41
43
|
## 3. 配音与真实时长
|
|
42
44
|
|
|
@@ -48,7 +50,7 @@ Presmium 原生 MCP 是创建、编辑、检查、配音、导出和截图的首
|
|
|
48
50
|
|
|
49
51
|
## 4. 导出并验收成片
|
|
50
52
|
|
|
51
|
-
|
|
53
|
+
用当前文稿的原生视频导出动作,烧入中文字幕并导出联系表。检查实际视频文件、声音、首帧、正文字幕、关键结果帧、相邻换页和结尾,并按[动态验收](../presmium-animate/references/acceptance.md#4-review-four-different-things)核对分镜中的完整操作链、结果时机与对象连续性;事实、布局、运行正确不等于动画质量通过。源文稿、预计输出路径或静态联系表不算成片验收。成片须有完整配音;无声预演不是这条流程的最终交付。
|
|
52
54
|
|
|
53
55
|
## 5. 独立封面与简介
|
|
54
56
|
|