AI turns documents or topics into real, native PowerPoint decks—with native shapes, transitions and animations, data-backed charts and tables on demand, audio narration from speaker notes, and support for your own .pptx templates. · by Hugo He
refactor(image-layout): make compact guidance always-on
hugohe3ca165fb8f4e7
diff --git a/docs/technical-design.md b/docs/technical-design.mdindex fc7e84e9..3146e183 100644--- a/docs/technical-design.md+++ b/docs/technical-design.md@@ -518,16 +518,16 @@ Several architectural decisions shape this phase:## Image-Text Layout: Primary Structures + Modifier Layers-The optional library of *how an image can be placed on a slide* (full vocabulary in [`references/image-layout-patterns.md`](../skills/ppt-master/references/image-layout-patterns.md)) offers 99 stable-id techniques in two freely composable layers:+Whenever the image/formula branch is active, its compact placement vocabulary in [`references/image-layout-patterns.md`](../skills/ppt-master/references/image-layout-patterns.md) is read alongside the layout math. It offers 100 stable-id techniques in two freely composable layers:- **Primary Structures** (container layouts / image-as-canvas + native overlay / multi-image compositions) — the page's bones. One or more per page; cross-Primary combinations like *side-by-side comparison + image-as-canvas annotation* are legitimate.-- **Modifier Layers** (non-rectangular clips / overlays & masks / texture / special techniques) — finish. Any number per page, stacked on top of the Primary.+- **Modifier Layers** (non-rectangular clips / overlays & cutouts / texture / special techniques) — finish. Any number per page, stacked on top of the Primary.-**Why the library does not impose a catalog or layer quota.** It expands compositional choice rather than defining legal output. A page may use one or several Primary structures and any useful Modifiers, cite no id at all, or use a free-form composition when that better serves the narrative and hierarchy.+**Why always reading the library does not impose a catalog or layer quota.** It expands compositional choice rather than defining legal output. A page may use one or several Primary structures and any useful Modifiers, cite no id at all, or use a free-form composition when that better serves the narrative and hierarchy.-**Why the layers are physically separated, not just tagged.** Patterns are organized with Primary structures first and Modifiers second so a reader who opens the optional library can navigate by construction role. Numbers remain stable identifiers (`#38` is still image-as-canvas + annotation cards regardless of where it sits in the file), so existing references across `spec_lock.md`, `design_spec.md §VIII`, executor logs, and historical examples keep resolving.+**Why the layers are physically separated, not just tagged.** Patterns are organized with Primary structures first and Modifiers second so the active role can navigate by construction role. Numbers remain stable identifiers (`#38` is still image-as-canvas + annotation cards regardless of where it sits in the file), so existing references across `spec_lock.md`, `design_spec.md §VIII`, executor logs, and historical examples keep resolving.-**Why composition intent flows through Strategist's resource list.** The `Layout pattern` column in `§VIII Image Resource List` carries one non-empty free-form suggestion and may optionally cite stable ids from the library; `Crop Policy` separately records `adaptive` or `no-crop`. This preserves a useful starting point across session re-entry without making the library or its numbering mandatory. Executor may resize, reflow, reposition, rebalance, replace the suggestion, or use another composition when that communicates better. Resource identity, must-use/content obligations, `no-crop`, and explicit user/template constraints remain binding; only changing those requires an upstream Design Spec update.+**Why composition intent flows through Strategist's resource list.** The `Layout pattern` column in `§VIII Image Resource List` carries one non-empty free-form suggestion and may optionally cite stable ids from the library; `Crop Policy` separately records `adaptive` or `no-crop`. This preserves a useful starting point across session re-entry without making any catalog entry or id mandatory. Executor may resize, reflow, reposition, rebalance, replace the suggestion, or use another composition when that communicates better. Resource identity, must-use/content obligations, `no-crop`, and explicit user/template constraints remain binding; only changing those requires an upstream Design Spec update.**Why true hard constraints stay upstream.** Cross-cutting SVG authoring and PPTX-compatibility exceptions live in the authority set routed by [`shared-standards.md`](../skills/ppt-master/references/shared-standards.md). The layout patterns file points to that router rather than restating the contract, so each rule still has one owning module and no stale duplicate in the pattern catalog.diff --git a/docs/zh/technical-design.md b/docs/zh/technical-design.mdindex 804083cb..c393de22 100644--- a/docs/zh/technical-design.md+++ b/docs/zh/technical-design.md@@ -488,16 +488,16 @@ Generate 执行以 [`workflows/generate-pptx.md`](../../skills/ppt-master/workfl## 图文版式:Primary 主结构 + Modifier 修饰层-「图片**可以怎么放上幻灯片**」的可选灵感库(完整词汇在 [`references/image-layout-patterns.md`](../../skills/ppt-master/references/image-layout-patterns.md))提供 99 条稳定编号技法,分成两层并可自由组合:+只要图片 / 公式分支被触发,就会把 [`references/image-layout-patterns.md`](../../skills/ppt-master/references/image-layout-patterns.md) 的精简版式词汇与布局计算规范一起读入。词汇库提供 100 条稳定编号技法,分成两层并可自由组合:- **Primary 主结构**(容器布局 / 图作画布 + 原生覆盖 / 多图组合)—— 页面的骨架。一页可一个也可多个;跨 Primary 的组合,如「侧边对比 + 图作画布的注解卡」,是合规的。-- **Modifier 修饰层**(非矩形裁剪 / 遮罩与叠加 / 纹理 / 特殊技法)—— 装饰层。一页可叠任意多个,附着在 Primary 之上。+- **Modifier 修饰层**(非矩形裁剪 / 叠加与挖孔 / 纹理 / 特殊技法)—— 装饰层。一页可叠任意多个,附着在 Primary 之上。-**为什么灵感库不设置目录或层数配额。** 它用于扩展构图选择,不定义合法输出。一页可以由一个或多个 Primary 构成,按需叠加 Modifier,也可以完全不引用编号,直接使用更适合叙事和层级的自由构图。+**为什么常驻读取灵感库仍不设置目录或层数配额。** 它用于扩展构图选择,不定义合法输出。一页可以由一个或多个 Primary 构成,按需叠加 Modifier,也可以完全不引用编号,直接使用更适合叙事和层级的自由构图。**为什么物理拆分两层,而不是只打标签。** 灵感库按 Primary 在前、Modifier 在后组织,便于按构造角色查找。编号仍是稳定 id(`#38` 永远是「图作画布 + 注解卡」,不论它在文件里的物理位置),因此 `spec_lock.md`、`design_spec.md §VIII`、历史 executor 输出和过往示例里的既有 `#<id>` 引用照样解析。-**为什么构图意图走 Strategist 资源列表。** `§VIII 图片资源列表` 的 `Layout pattern` 列承载一句非空自由格式建议,也可以按需引用灵感库的稳定编号;`Crop Policy` 独立记录 `adaptive` 或 `no-crop`。这让可用的构图起点通过 lock 投影在 session 重入后继续存在,但不要求完整读取灵感库或使用编号。Executor 可以调整尺寸、位置、流向与权重,也可以替换建议或使用其他构图。资源身份、必用 / 内容义务、`no-crop` 和显式用户 / 模板约束仍具有约束力;只有改变这些边界才需要先更新 Design Spec。+**为什么构图意图走 Strategist 资源列表。** `§VIII 图片资源列表` 的 `Layout pattern` 列承载一句非空自由格式建议,也可以按需引用灵感库的稳定编号;`Crop Policy` 独立记录 `adaptive` 或 `no-crop`。这让可用的构图起点通过 lock 投影在 session 重入后继续存在,但不要求使用编号。Executor 可以调整尺寸、位置、流向与权重,也可以替换建议或使用其他构图。资源身份、必用 / 内容义务、`no-crop` 和显式用户 / 模板约束仍具有约束力;只有改变这些边界才需要先更新 Design Spec。**为什么真正的硬约束留在上游。** 跨切的 SVG 创作与 PPTX 兼容性例外属于 [`shared-standards.md`](../../skills/ppt-master/references/shared-standards.md) 路由的权威集。版式词表只指向该路由,不再复述合同;每条规则仍只有一个所属模块,词表里也不会留下过期副本。diff --git a/skills/ppt-master/references/executor-base.md b/skills/ppt-master/references/executor-base.mdindex c12d807e..70cb994b 100644--- a/skills/ppt-master/references/executor-base.md+++ b/skills/ppt-master/references/executor-base.md@@ -9,7 +9,7 @@ Always-loaded Executor authority for flat SVG page authoring and behavior shared| `pptx_structure.mode: structured` | [`executor-structured.md`](./executor-structured.md) || Any data chart, chart catalog selection, or text-grid table | [`executor-chart.md`](./executor-chart.md) || A page will use a preset pattern fill or evaluate native chart/table replacement | [`native-data-interface.md`](./native-data-interface.md) before deciding eligibility or emitting metadata |-| Any image/formula; a cited `#<id>` or optional composition recall also triggers the library | [`executor-image.md`](./executor-image.md); conditionally [`image-layout-patterns.md`](./image-layout-patterns.md) |+| Any image/formula | [`executor-image.md`](./executor-image.md) + [`image-layout-spec.md`](./image-layout-spec.md) + [`image-layout-patterns.md`](./image-layout-patterns.md) + [`svg-image-embedding.md`](./svg-image-embedding.md) || Any `Status: Sourced` web image | [`executor-web-image.md`](./executor-web-image.md), after `executor-image.md` || Effective Speaker Notes outcome is enabled after all SVG pages pass | [`executor-notes.md`](./executor-notes.md) |diff --git a/skills/ppt-master/references/executor-image.md b/skills/ppt-master/references/executor-image.mdindex ecfb79eb..cbd8a463 100644--- a/skills/ppt-master/references/executor-image.md+++ b/skills/ppt-master/references/executor-image.md@@ -27,9 +27,14 @@ Handle images by status; enum and lifecycle: [`svg-image-embedding.md`](svg-imag**Template-bundled images**: [`apply-template-workspace.md`](../workflows/stages/apply-template-workspace.md) copies them into project `images/`. Outside `mirror`, reference `../images/<name>` and never copy a template SVG's bare sibling href: the rendered page lives in `svg_output/`. `mirror` ([`executor-structured.md`](./executor-structured.md) §1.1) keeps hrefs verbatim; export resolves them against `images/`.-**Reference — layout catalog is optional recall**: Load [`image-layout-patterns.md`](./image-layout-patterns.md) for a cited `#<id>` or useful recall; free-form suggestions need no lookup.--**Default — action-bearing image composition (may override only for an upstream-documented restraint or explicit user/template/content boundary)**: On a `hero_page` or other image-led page, first understand the suggestion's communication job, then develop an image/content or image/shape relationship from hierarchy, copy, asset ratio/focus, and deck rhythm; do not mechanically reproduce its named pattern. Position, size, crop, and a legibility-only scrim alone do not constitute that relationship. A split or full-bleed remains valid only for deliberate restraint recorded in §IX `Layout` / `Images` or an explicit boundary. Preserve role/source, must-use, crop/content, and explicit user/template constraints; expression-only changes need no upstream rewrite.+**Layout ownership**: The image branch always arrives with+[`image-layout-spec.md`](./image-layout-spec.md) and+[`image-layout-patterns.md`](./image-layout-patterns.md) already read. Use the+catalog as vocabulary, not a quota or geometry lock. Executor owns final SVG+geometry and may deepen, simplify, or replace the preferred pattern—including+with a plain split or full bleed—while preserving resource role/source,+must-use status, crop/content boundaries, and explicit user/template+constraints. Expression-only changes need no upstream rewrite.**Reference — motion-ready image layering, not a constraint**: For adopted §IX or an explicit focus, comparison, evidence, reveal-order, or cross-page requirement, decide during SVG authoring whether the final composition needs separate visible units. Keep ordinary stable framing/background static and wrap each independently revealed or continuing Slide-local unit in a descriptive direct-root `<g id>`; structured atoms/slots retain their boundaries. Existing units or a page transition may suffice. The motion stage owns effects, pairing, order, and timing.diff --git a/skills/ppt-master/references/image-generator.md b/skills/ppt-master/references/image-generator.mdindex 823072e9..4d98a63f 100644--- a/skills/ppt-master/references/image-generator.md+++ b/skills/ppt-master/references/image-generator.md@@ -16,7 +16,7 @@ AI images exist to serve the deck's communication goal. Pick whatever combinatio| `page_role` | Use ||---|---|-| `local` | Image occupies a region of an SVG page (left half, right column, hero band, accent corner). Composition is the AI's call — fill the region as the page design wants |+| `local` | Image occupies a prepared SVG region. The AI composes inside that bitmap/container; it does not choose the page region or final SVG geometry || `hero_page` | Image is the page's main voice — cover, chapter divider, mood transition, single-number hero, closing quote. SVG above may be minimal or empty |**Two text policies** (orthogonal to page_role):@@ -34,13 +34,13 @@ AI images exist to serve the deck's communication goal. Pick whatever combinatio- **In-image text is only for words that will not need editing later** — visual keywords, decorative lettering, mood words. Editable text (titles that may be reworded, subtitles, dates, authors, captions, body) belongs in SVG. Changing one in-image word costs an image regeneration; one SVG word costs a keystroke.- Prompts are one coherent prose paragraph, not tag soup (a model-output reality, not an aesthetic choice)-Everything else is the AI's judgment per page. No mandated padding, no type-locked text_policy, no scenario whitelists for hero_page.+Everything else inside the prepared bitmap is the AI's judgment per page. No mandated padding, no type-locked text_policy, no scenario whitelists for hero_page.---## 2. Style and Composition Inputs-Every AI image uses one deck-wide rendering, the deck's stable color anchors/semantic behavior, and a per-image type / composition. Only rendering is a separate image-direction decision.+Every AI image uses one deck-wide rendering, the deck's stable color anchors/semantic behavior, and a per-image type / internal composition. Only rendering is a separate image-direction decision.| Dimension | Decides | When fixed ||---|---|---|@@ -71,7 +71,7 @@ Every AI image uses one deck-wide rendering, the deck's stable color anchors/sem### Step 1 — Load the dimension indices-Read the two index files that own user-visible image direction and per-image composition.+Read the two index files that own user-visible image direction and per-image internal composition.```read_file references/image-renderings/_index.md@@ -122,6 +122,8 @@ Derive color behavior from the available roles and image context: background / sFor each `Acquire Via: ai` row, use Strategist-owned §VIII/lock by default or the main agent's transient Quick roster. Explicit values remain binding; Quick resolves omissions automatically.+`Layout pattern` is a page-realization preference and is not copied wholesale into the bitmap prompt. Any generation-time subject direction, focal placement, quiet region, or overlay-safety requirement must therefore be present in the row's `Reference`, the matching §IX block, or the Quick roster's visual intent.+1. **Determine `page_role`** — the owning row's explicit value wins; a blank or omitted value resolves to `local`. In Default Generate, `hero_page` must be Strategist-explicit; in Quick Generate, the main agent may resolve it while building the transient roster.2. **Determine `text_policy`** — the owning row's value wins when set. **Declared-inference fallback for a blank or omitted value**: pick `none` or `embedded` from the row's `Purpose`, `Reference`, and page intent based on whether in-image text serves the page. Long body / data / lists stay in SVG.3. **Determine type or free composition** — an Illustration Sheet omits manifest `type` and follows §4.3's grid composition. For another local structural infographic, use one of the 11 types only when the `_index.md` offers a real match; otherwise omit type and author the intended structure directly with §4.1 E. A local single-subject/portrait image omits type and uses §4.1 A/B inside its actual region. A `hero_page` omits type and uses §4.1 A/B/C/D/E.diff --git a/skills/ppt-master/references/image-layout-patterns.md b/skills/ppt-master/references/image-layout-patterns.mdindex 21feb112..a39ed722 100644--- a/skills/ppt-master/references/image-layout-patterns.md+++ b/skills/ppt-master/references/image-layout-patterns.md@@ -1,436 +1,258 @@-# Image-Text Layout Pattern Library+# Image and Formula Layout Pattern Catalog-An optional vocabulary library for ways images can be placed on a slide. Open it when a page would benefit from more composition ideas; ordinary natural-language layout suggestions remain valid without consulting or citing the library. When using it, start at **High-Yield Patterns** below.--Every entry has a name plus a short technical hint. Common techniques get a single line. Less obvious or easily forgotten techniques get a short paragraph — not a full tutorial, but enough that a model unfamiliar with the project can implement it without guessing. This is an inspiration library, not a legality boundary or teaching document; it sets no usage, id, family, or coverage quota.--> **Numbers are stable optional identifiers, not sequence.** The file is split into **Part 1 — Primary Structures** (#1–#19, #38–#56, #73–#81, #88, #92–#94) and **Part 2 — Modifier Layers** (#20–#37, #57–#72, #82–#87, #89–#91, #95–#100). Numbers jump within each Part because Primary structures were grouped first; existing references to `#38`, `#48`, etc. anywhere in the project still resolve correctly. **High-Yield Patterns** is a router over those same numbers; use it when entering by page situation.+Compact composition vocabulary for prepared images, illustrations, and rendered formula assets. Use the patterns as options, not as a checklist.----## Core Principle — Two Layers--Almost every pattern below is an instance of one underlying split:+## 1. Catalog Boundary-> **The image carries atmosphere, world-building, emotional weight. Native SVG shapes carry information, data, editable text.**--This is the single most underused move in image-heavy decks. The default reflex is to place image and text in adjacent rectangles. The far more powerful move — especially for content-rich pages — is to let the image **be the canvas** (often full-bleed) and draw native vector elements (annotation cards, flow nodes, KPI tiles, leader lines, network diagrams, dashboards) directly on top.+| Boundary | Rule |+|---|---|+| Selection | **Reference — not a constraint**: use any listed pattern, combine compatible patterns, or author a clearer free-form composition. No ID, family, or coverage quota applies |+| Stable IDs | `#1`–`#100` are lookup handles. Keep their meanings stable; their numeric order is not a workflow |+| Primary vs modifier | A Primary Structure supplies page geometry. Modifier Layers change crop, contrast, depth, framing, asset treatment, or continuity |+| Asset ownership | Patterns consume prepared project-local assets. This catalog does not authorize acquisition or processing during SVG realization |+| Exact information | Keep authoritative text, data, labels, and annotations native when they must remain exact or editable |-Anything that must remain editable, numerically or semantically exact, or styled to the deck's exact typography belongs in the SVG layer regardless of what the image looks like underneath. Script alone never decides ownership.+| Technical concern | Owner |+|---|---|+| SVG legality | [`shared-standards-core.md`](./shared-standards-core.md) |+| Image embedding and crop policy | [`svg-image-embedding.md`](./svg-image-embedding.md) |+| Container sizing and aspect-ratio math | [`image-layout-spec.md`](./image-layout-spec.md) |+| Supported effects | [`svg-effects.md`](./svg-effects.md) |+| Boolean shape authoring | [`native-shape-authoring.md`](./native-shape-authoring.md) |+| Accurate chart overlays | [`executor-chart.md`](./executor-chart.md) |+| Page and object motion | [`animations.md`](./animations.md) |----## High-Yield Patterns — Optional Starting Points--The patterns below are efficient ways to expand a page beyond familiar splits. Nearly all of them are **one `<image>` plus geometry** — no extra asset, no generation cost, no second render — and they are the SVG equivalents of what PowerPoint users reach for under Merge Shapes.--**Reference — not a constraint**: use this router when it adds a useful composition option. A plain split, equal grid, bare whitespace, or an unlisted free-form construction remains valid when it serves the content.--| Page situation | Reach for | Produces |-|---|---|---|-| One ordinary photo must carry a cover or a chapter divider | `#90` scrim with shapes cut out + `#86` contour echo | Three elements turn a stock image into a designed page; the cut contour is where the page's character comes from |-| The supplied image does not fit the canvas | `#89` same image twice — sharp cutout over a receded copy | Subject at full fidelity in any aspect ratio; no stretching, no letterbox bars, no second asset |-| Several peer images belong to one frame | `#92` split tiling — one parent cut into interlocking cells | Edges interlock exactly; the group still reads as one object |-| One image should span detached containers as one edit object | `#82` one image shattered across separated shapes | Merge-Shapes look; the photo runs continuously behind the gaps |-| Those same-source containers must remain independently editable or animated | `#100` same-source addressable crops | Several native picture objects share one source coordinate system without slice assets |-| A panel needs a real opening onto what is behind it | `#83` panel with a hole punched through it | True subtraction — survives a gradient, a texture, or a second image behind the panel |-| A photo row needs depth without 3D | `#94` embracing arc row, or `#93` containers arrayed along a curve | A perspective wall reproduced in 2D from scale + vertical offset alone |-| A flat scrim reads as a sheet of paint over the photo | `#98` grid scrim with per-cell opacity | The overlay reads as panelled glass or a contact sheet, felt rather than drawn |-| A busy photo has no clear focus | `#99` selective desaturation, or `#96` cutout subject re-laid over its own photo | Focus without cropping; the subject can then overlap a title, a panel, or a grid line |-| Text needs legibility but a solid scrim would kill the photo | `#97` frosted-glass panel | The photo's colour and composition stay visible through the panel |-| An image grid looks like a stock template | `#88` non-rectangular tessellation with 1–3 cells left empty | The empty cells are where the title and body copy live |-| A subject should escape its container | `#85` subject breaking out + `#96` | Depth with no shadow at all |-| One place should be recognized across consecutive pages | `#87` one image panned across pages | The deck reads as one continuous scene; `-t morph` uses heuristic matching, while explicit `morph.pairs` makes the camera pan deterministic |+## 2. Situation Router-**Reference — not a constraint**: when citing a modifier-only result, also name the content-appropriate Primary that supplies the page bones. Free-form suggestions may describe the complete relationship without catalog ids.--**Hard rule — registration is what makes this family work**: in `#82`, `#85`, `#87`, `#89`, `#96`, `#97`, and `#100`, the image stays anchored to the *union* of its containers, or the copies share one source coordinate system. A few pixels of drift reads as a printing error. `#84` alone breaks registration on purpose.--**Prepared-asset gate**: select `#96` only when a registered cutout PNG already exists, `#97` only when its blurred crop exists, and `#99` only when its desaturated copy exists. If not, keep the original asset and fall back to a native-shape treatment such as `#30` / `#29`; do not invent an image-processing step during execution.--**Reference — not a constraint**: repeated plain splits may be a reason to consult this library, but are not evidence of failure. Neither §VIII nor the final deck must cite an id or cover a pattern family.--Each entry above is specified in full at its own number below; the table routes, it does not restate.+| Page need | Pattern options |+|---|---|+| Quiet, direct evidence | `#16` negative space, `#19` framed figure, `#47` small multiples, `#48` comparison |+| One visual should become the page canvas | `#38`–`#46` native overlays |+| One source should span unusual geometry | `#82` one picture, `#100` addressable pictures, `#89` sharp subject over receded copy |+| Several visuals should read as one system | `#50` grid, `#77` mosaic with text cell, `#88` tessellation, `#92` split tiling, `#93` curve array, `#94` depth row |+| A foreground needs an opening or reveal | `#83` true hole, `#90` cut scrim, `#95` background-registered fill, `#68` text subtraction |+| Text needs contrast without discarding the visual | `#29` directional scrim, `#33` spotlight, `#97` prepared frosted panel, `#98` grid scrim |+| A cover, divider, or promotional page needs image-led structure | `#73`–`#81` |+| Consecutive pages should share one visual world | `#87` cross-page pan opportunity |----# Part 1 — Primary Structures--Pick one or more of these as the page's bones. Cross-primary combinations are encouraged (see Composition Guidance).--## Container Layouts (where the image sits)--1. **Full-bleed background with floating title** — `<image x=0 y=0 width=1280 height=720 preserveAspectRatio="xMidYMid slice"/>` + scrim `<rect>` for legibility + overlay `<text>`.--2. **Left-third image + right text body** — `<image x=0 y=0 width=~427 height=720>` on the left; text area in the remaining width; optional right-edge gradient fade for smooth transition.+## 3. Primary Structures-3. **Right-third image + left text body** — mirror of #2.+### 3.1 Single-Image Structures-4. **Right image bleeding off the canvas edge** — `<image>` width extended past viewBox; text on left with a rightward gradient fade so the image emerges from the text area without a visible boundary.+1 · 73. **Full-bleed title structures** — **#1** floats a title over one canvas-filling image; **#73** uses a poster-like side or lower-corner title stack without a title card.-5. **Top-band image + bottom multi-column text** — `<image x=0 y=0 width=1280 height=~340>` at the top + bottom-fade gradient + 2–3 evenly spaced text columns below.+2–3. **Side image and text** — **#2** places the visual left and copy right; **#3** mirrors that relationship.-6. **Bottom-band image + top title + middle text** — mirror of #5 with the image at the bottom and a top-fade gradient.+4. **Edge-bleed image** — extend the visual beyond one canvas edge so it enters or exits the page instead of sitting in a box.-7. **Top-and-bottom symmetric split** — image occupies 50% (top or bottom) with a divider line or thin gradient band separating the halves.+5–7. **Horizontal split structures** — **#5** uses an upper image band with content columns below; **#6** places the image band below a title/content field; **#7** gives image and content balanced top/bottom fields with a deliberate seam.-8. **Z-pattern serpentine** — three rows, image on the left in rows 1 and 3, on the right in row 2 (or alternating). Each row roughly 1/3 canvas height; visual flow zigzags down the page.+9. **Central image in a 3×3 field** — put the visual at the center and use surrounding cells for labels, evidence, or small data.-9. **3×3 grid with central image** — nine cells; center cell holds the image, the other 8 hold text blocks, color swatches, or small data widgets.+10. **Centered image with radial callouts** — place one focal visual centrally and route native callouts outward.-93. **Containers arrayed along a curve (fan, arc, ring)** — N image containers distributed along an arc or wave, each rotated to sit square to the curve at its own position. Reads as motion and hierarchy at once, and it is the backbone of fan spreads, ring layouts, dial/roulette pages, and arched photo rows.+11. **Diagonal visual/content transition** — use a diagonal edge or directional fade whose contour supports the page's reading direction.- **Geometry** — place container `i` of `n` on a circle of radius `r` about `(cx, cy)`:- ```- θᵢ = θ_start + i × (θ_span / (n − 1))- xᵢ = cx + r·cos(θᵢ) yᵢ = cy + r·sin(θᵢ)- rotationᵢ = θᵢ + 90° (tangent-aligned; drop this for upright containers)- ```- Use `transform="rotate(rotationᵢ xᵢ yᵢ)"` on each container group. A `θ_span` of 60–120° reads as a fan; 360° with `θ_start = -90°` gives an evenly spaced ring. For a wave instead of an arc, sample the wave's own path and use its local tangent as the rotation.+12. **Receded image with oversized type** — push the image into the background and make typography the dominant foreground.- **Two things to get right**: keep radius and angular step *constant* — an eyeballed fan reads as a mistake, not a flourish; and when containers are tangent-aligned, images inside must not inherit the rotation blindly (a sideways face is the failure mode). Counter-rotate the image inside its container, or keep the containers upright and let only their positions follow the curve.+13 · 18. **Vertical image columns** — **#13** is a slim strip beside giant horizontal type; **#18** is a full-height sidebar beside a general content field.-10. **Centered image with radial callouts pointing outward** — image (often circular via `clipPath`) at canvas center; multiple `<line>` leader lines + small `<circle>` endpoints + offset text labels in surrounding space.+14 · 76. **Mid-page image bands** — **#14** runs a banner between content above and below; **#76** places native copy in a calm part of a wider image belt while keeping the heading outside.-11. **Diagonal split with directional gradient (not hard polygon cut)** — full-bleed `<image>` + overlay `<rect fill="url(#grad)">` whose gradient axis runs along the diagonal, plus a `<line>` to make the divider read. Do NOT hard-clip: polygon cuts give stair-stepped edges on text panels.+16. **Negative-space dominant** — keep the visual and copy compact so whitespace carries hierarchy.-12. **Faded image as backdrop with oversized overlay text** — `<image>` + heavy semi-transparent `<rect fill="bg-color" fill-opacity="0.5–0.7">` over it + huge `<text>` (80–120px) on top. Image becomes texture; text is the subject.+19. **Framed image with caption** — float one image in whitespace with a restrained frame and native caption.-13. **Narrow vertical image strip + giant horizontal title** — `<image x=0 y=0 width=200–280 height=720>` + thick divider `<rect>` + large `<text>` (60–90px) in the remaining width.+81. **Illustration-as-layout field** — let a large illustration or cutout set the page rhythm; place copy in its calm regions.-14. **Horizontal banner strip cutting through mid-section** — `<image y=middle width=1280 height=200–280>` with edge fades; text blocks above and below the band.+### 3.2 Image-as-Canvas with Native Overlay-15. **Multi-image montage with bold text spanning across** — `<image>` tiled with 2–4px gaps + large `<text>` (60–100px) in a `<rect fill-opacity="0.5–0.7">` band spanning the montage, so the text stays legible across every tile beneath it.+Use the prepared visual as the spatial field while native SVG carries the information layer.-16. **Negative-space dominant — small image, mostly whitespace** — image and text together occupy less than 40% of the canvas; rest is empty.+38. **Annotated evidence** — place compact annotation cards with routed leaders over the visual.-17. **Picture-in-picture inset** — large `<image>` background + small `<image>` overlaid inside it with a `<rect>` frame.+39. **Process through a scene** — connect numbered flow nodes along meaningful geometry in a real or illustrated scene.-18. **Image as full-height sidebar column** — narrow `<image x=0 y=0 width=~200–280 height=720>`; rest of canvas is content area.+40. **Contextual metrics** — place native KPI tiles in calm regions of the visual.-19. **Image floating in whitespace with thin frame and caption** — `<image>` + thin `<rect fill="none" stroke="…">` frame around it + `<text>` caption below.+41. **Engineering overlay** — add measurement lines, end ticks, module tags, and exact labels.-## Image-as-Canvas + Native Overlay (the most underused family)+42. **Interface overlay** — add translucent UI panels, progress indicators, badges, and native arcs.-This is the family that opens up the largest design space and the one AI is most likely to skip. The shared pattern: image fills the slide (or a large region), native SVG elements are layered on top to carry the actual information. None of the overlay elements need to be generated by the image model — they are vector primitives you draw yourself.+43. **Accurate chart over visual context** — draw the chart natively, treat the image as context only, and follow [`executor-chart.md`](./executor-chart.md).-38. **Background image + annotation cards with Shape-first leaders** — full-bleed `<image>` + 2–4 small info cards (`<rect rx>` + icon + title + one-line text) placed in the image's calm regions. Point to each subject with a straight `<line>` by default, or an authored native bent/curved Connector when its stock contour fits. Use a custom Bézier leader only when neither can route around the subject faithfully. Card text and leader lines remain editable; image is the scene.+44. **Architecture or network overlay** — draw native nodes, connections, icons, and labels over the scene.-39. **Background image + flow nodes drawn over the scene** — the image is a real or rendered scene (workshop, control room, landscape). On top, connect numbered `<circle>` stops with straight `<line>` segments or exact native bent/curved Connector contours. Use a custom dashed route only when the workflow must follow meaningful scene geometry those shapes cannot express. Each node = number + icon + label. The flow is fully editable; the image is atmosphere.+45. **Hotspots with sidebar legend** — pair numbered points on the visual with a matching native legend.-40. **Background image + floating KPI metric cards** — full-bleed image (often an operations photo) + dark scrim + multiple `<rect>` cards in negative-space regions. Each card = icon + small label + large metric number. Image gives context; cards give the data.+46. **Detail lens** — border one sub-region and place a native caption nearby.-41. **Background image + measurement lines and module tags (engineering overlay)** — used on technical / blueprint / cross-section images. Draw measurement lines with end-caps (`<line>` + perpendicular ticks) spanning a feature, with a centered label box reading dimensions or part names. Add tagged callouts with `<rect>` + monospace text. Reads as engineering drawing markup.+### 3.3 Multi-Image Structures-42. **Background image + glassmorphism UI panels** — image is the visual world; on top, draw UI elements (semi-transparent panels, progress arcs, status badges, indicators). Panels use `fill-opacity="0.6–0.8"` + thin light-color strokes; use exact native `arc` / `blockArc` presets when they fit, and custom `A` geometry only for data-defined arcs they cannot express. Looks like a live dashboard floating above the scene.+8. **Z-pattern serpentine** — alternate image and text positions down successive bands to create a zigzag reading path.-43. **Background image + native data chart on top** — AI image generation cannot produce accurate data charts. Solution: use an AI-generated dashboard image as **visual reference only** (clearly labeled as such in a caption), and draw the actual chart with native SVG primitives (`<line>` axes, `<path>` series, `<circle>` data points) directly on or next to it. Required marker if exporting: `<!-- chart-plot-area: x_min,y_min,x_max,y_max -->` inside the chart group.+15. **Montage with spanning type** — tile several visuals and run one legible title treatment across the assembled field.-44. **Background image + native network/architecture diagram** — same logic as #43 but for structural diagrams. Image provides atmosphere or visual anchor; the actual nodes, connections, and labels are SVG circles, lines, icons, and text — all editable.+17. **Picture-in-picture inset** — overlay one framed inset image over a larger image; the sources may differ. Use `#62` when an overview and detail must share one source.-45. **Background image + numbered hotspots with sidebar legend** — small numbered `<circle>` markers placed on the image at points of interest. A sidebar (left or right) lists "1. … 2. … 3. …" with corresponding descriptions.+47. **Small multiples** — arrange same-kind images in one evenly spaced row with identical containers and caption structures so peers can be compared.-46. **Background image + bordered "lens" rectangle highlighting a sub-region** — full-bleed image + a bordered `<rect fill="none" stroke="accent" stroke-width="3"/>` framing a sub-region + caption nearby. Frame draws the eye to one detail without occluding the surrounding context.+48. **Before/after or A/B comparison** — place two equally sized image containers side by side and label both states explicitly.-## Multi-Image Compositions+49. **Asymmetric collage** — balance one dominant visual with smaller supporting visuals using consistent gaps.-94. **Embracing arc row (2D substitute for a 3D perspective wall)** — a row of images or cards where the centre element is largest and each step outward shrinks and drops, so the tops trace an arc and the row appears to curve toward the viewer. This is what PowerPoint decks build with 3D rotation (perspective left / right, X-axis 330° / 30°) for logo walls, certificate rows, and photo shelves — and it is reproducible in 2D, which matters because 3D transforms are outside the SVG contract ([`svg-effects.md`](./svg-effects.md) §6.8).+50. **Equal-cell tiled grid** — use equal containers when equality and scanability are the message.- **Construction**: for element `k` steps from the centre, apply `scale = 0.88ᵏ` and offset `y` downward so every element's *top* edge lands on one shallow arc; keep the horizontal step constant. Mirror the sequence left and right of the centre. Add a soft ground shadow or a reflection fading downward to seat the row. Bottom-aligning instead of top-arcing gives the flatter "shelf" variant.+51. **Irregular mosaic** — pack different-sized tiles into one coherent field.- The depth cue is entirely **scale + vertical offset + consistent light**; do not reach for skew or a fake 3D tilt, which fail closed on export. Three to seven elements is the working range — beyond that the outermost ones shrink into illegibility.+52–53. **Filmstrip and stack** — **#52** aligns a horizontal sequence by height while allowing content-driven widths; **#53** aligns a vertical sequence by width with annotations on a shared side.-47. **Small multiples — 3–6 same-kind images in an evenly spaced row** — identical containers, identical caption blocks (title + one line). Not a generic grid: the identical framing *is* the message, because readers compare across panels only when the structure is constant.+54. **Overlapping image stack** — use z-order and restrained offsets to create a layered print or archive feel.-48. **Side-by-side comparison (before/after, A/B, then/now)** — two `<image>` of equal size in 50/50 split with thin divider `<line>` and "before" / "after" labels.+55–56. **Diptych and triptych** — **#55** pairs adjacent images; **#56** aligns three distinct sources, unlike baked **#26**.-49. **Asymmetric collage** — one large `<image>` + 2–3 smaller `<image>` arranged around it; sizes vary, gaps consistent.+74. **Image-navigation table of contents** — turn sections into visual navigation cards with native numbering and summaries.-50. **Tiled grid (2×2, 2×3, 3×3) with equal cells** — `cell_size = (canvas - total_gap) / cols`; consistent `gap=2–20px`.+75. **Asymmetric dual-image chapter banner** — pair a compact image with a wider image and anchor them with a native section marker.-51. **Mosaic** — irregular tile sizes packed together with or without thin gaps; each image clipped to its tile's rect.+77. **Photo mosaic with a text cell** — reserve one mosaic cell for copy so absence of a photo creates hierarchy.-92. **Split tiling — one parent shape cut into interlocking cells** — the most-used construction in real image-heavy decks, and the counterpart to #82. Take one parent shape (circle, annulus, rounded rect, trapezoid, wave band), lay cutting lines across it (long bars, evenly distributed or fanned at different angles), and split it into cells. Each cell then holds a *different* image. Because every cell comes from one parent, the edges interlock exactly — no gaps, no overlaps, and the group still reads as one object.+78. **Ambient image, evidence image, and text panel** — let one visual establish mood and another provide concrete proof.- | Parent + cutters | Result |- |---|---|- | Circle + 2 crossed bars | Quadrant wheel |- | Annulus + radial bars | Ring segments |- | Wave band + vertical bars | Rhythmic strip |- | Trapezoid + slanted bars | Perspective row |+79. **Ribbon-header image cards** — give peer image columns distinct native ribbon or chevron headings.- **Authoring**: compute each cell's contour and write it as its own `<path>` clip — the geometry is deterministic, so derive the cells rather than eyeballing them. `shape_boolean_svg.py render <svg-file> --operation fragment --source <id> --source <id> --id <result-id>` returns exactly these interlocking regions as separately addressable paths. Give every cell the same stroke (2px, background color) so the cuts read as designed seams.+80. **Side hero with staggered evidence cards** — pair a full-height hero field with supporting cards that step through the opposite side.- **Choosing between #92 and #82 / #100**: different images per cell (#92)- are peers. One registered source means one edit object (#82) or independent- same-source objects (#100).+88. **Non-rectangular tessellation** — tile clipped geometric cells and reserve selected cells for native copy or color.-52–53. **Filmstrip / stack** — a sequence of `<image>` with thin consistent gaps: horizontal, equal height and varying widths (**#52**), or vertical, aligned by width with shared annotations down one side (**#53**).+92. **Split tiling** — fragment one parent contour into interlocking cells, each holding a different image as an independent object.-54. **Overlapping image stack** — `<image>` elements with overlapping `x/y` positions; each subsequent one in front (z-order by document order); often combined with slight rotation for layered photo-print look.+93. **Containers arrayed along a curve** — distribute containers consistently along an arc, wave, or ring; keep image orientation intentional.-55–56. **Diptych / triptych** — two images abutting 50/50, vertical or horizontal (**#55**), or three side-by-side at equal or 2:1:2 widths (**#56**), with an optional thin divider `<line>`. Distinct from #26, where the panels live inside one image file, and from #48, where the pairing carries a before/after argument.--88. **Non-rectangular tessellation (honeycomb, diamond, chevron array)** — a tiled field of hexagons, diamonds, or slanted parallelograms, each cell holding its own image via `clipPath` (#23) and separated by a consistent 2–3px stroke in the background color, which reads as the grid's mortar. The non-rectangular counterpart to #50 / #51.-- **Geometry**: a flat-top hexagon of width `w` and height `h` is `M x+0.25w,y L x+0.75w,y L x+w,y+0.5h L x+0.75w,y+h L x+0.25w,y+h L x,y+0.5h Z`. Tile it by stepping `0.75w` horizontally and offsetting alternate columns by `0.5h` vertically.-- **Leave cells deliberately empty**: fill 1–3 tiles with a flat or gradient deck color instead of a photo. A fully-populated honeycomb reads as a stock template, and the empty cells are where the title and body copy live. Keep the identical stroke on the empty cells so they read as designed rather than as a missing image.--## Imported Deck Patterns (image-led promotional pages)--These patterns come from polished image-text decks where photos define the slide skeleton instead of sitting inside generic cards. Treat them as layout vocabulary for travel, product, venue, hospitality, real-estate, event, and brochure-style decks.--73. **Full-bleed poster image + side title stack** — title stack on the left or lower-left third, no title card; scrim only where the image is busy.--74. **TOC image-navigation cards** — 3–5 vertical image cards, each with a translucent overlay, chapter number, title, one-line summary. A visual preview of the deck, not a text list.--75. **Asymmetric dual-image chapter banner** — one small + one wide image across the upper half; chapter title below, anchored by an oversized section number.--76. **Mid-page image belt with native text inset** — wide image strip through the middle 45–60%, key text inside its calm region, heading above.--77. **Photo mosaic with a text cell** — irregular grid with one cell reserved for copy. The missing photo is the hierarchy; do not fill every slot just because a grid exists.--78. **Ambient banner + evidence photo + text panel** — atmospheric image above, concrete evidence photo below, copy on a tinted side panel. One image sets mood, the other proves it.--79. **Ribbon-header image cards** — 3 columns, colored ribbon or chevron title above each image, prose below.--80. **Side hero image + staggered evidence cards** — full-height image in a side column; 2–4 smaller cards staggered vertically opposite it rather than gridded.--81. **Illustration-as-layout field** — a large vector or cutout illustration acts as the image region and sets spatial rhythm, with text in its calm areas. For when a photo would be too literal but the page still needs image-scale mass.+94. **Embracing arc row** — create depth with a center-weighted scale and vertical-offset rhythm while keeping the objects two-dimensional.----# Part 2 — Modifier Layers--Stack any of these freely on top of a Primary structure. Multiple Modifiers per page is the expected case, not the exception.--## Non-rectangular Image Shapes--20–23. **Basic shape crops** — `<clipPath>` holding one shape, referenced by `<image clip-path="url(#id)"/>`: `<circle>` (**#20**), `<rect rx ry>` (**#21**, `rx` sets roundness), `<ellipse>` (**#22**), `<polygon points>` (**#23**, keep every vertex inside the image's display rect). #24 supersedes all four whenever the contour is curved or organic.+## 4. Modifier Layers-24. **Custom path crop (blob, leaf, silhouette)** — use `<clipPath><path d="…"/></clipPath>` only when circle, ellipse, rounded-rect, and polygonal crops cannot faithfully express the silhouette. PowerPoint export translates the necessary custom contour to `custGeom` and survives roundtrip.+### 4.1 Crop, Boolean, and Same-Source Geometry-25. **Layered paper-cut stack** — clip each image layer under the image-only contract in [`shared-standards-core.md`](./shared-standards-core.md) §1.2; draw vector layers directly in their final geometry. A small conditional shadow on each layer can create physical separation.+20–23. **Basic crops** — **#20** is a circle, **#21** a rounded rectangle, **#22** an ellipse, and **#23** a bounded polygon.-82. **One image shattered across separated shapes (Merge Shapes look)** — clip-one `<image>` with one `<path>` containing disjoint closed subpaths. Size the-image over their union so the scene remains continuous; export yields one-picture with `custGeom`. Use `shape_boolean_svg.py render` `union` / `combine`-for non-trivial contours and obey-[`shared-standards-core.md`](./shared-standards-core.md) §1.2. Distinct from-#24 (one contour), #47–#56 (different sources), and #100 (several pictures).+24. **Custom-path crop** — use one authored organic or silhouette contour when basic crops cannot express it.-100. **Same-source addressable crops** — repeat one exact `href` in independent-nested crop wrappers with different source-unit `viewBox` values. They export-as separate native picture objects for editing and Morph while assembling one-registered scene without slice assets. Follow-[`executor-image.md`](./executor-image.md) §1. Unlike #82 this yields several-pictures; unlike #84 registration remains exact.+25. **Layered paper-cut stack** — clip image layers independently and draw vector layers in their final geometry.- **Registration construction**: choose one visible container union- `U = (ux, uy, uw, uh)` and one source region- `S = (sx, sy, sw, sh)`. For a container- `F = (x, y, w, h)`, derive its source-unit crop as- `Sx = sx + (x-ux)/uw × sw`, `Sy = sy + (y-uy)/uh × sh`,- `Sw = w/uw × sw`, and `Sh = h/uh × sh`. Use that result as the nested- wrapper `viewBox`; do not choose each crop by eye and do not apply- independent `cover`. This makes irregular heights and gaps behave like- windows cut from one continuous image while keeping every window a native- picture object.+62. **Full view plus zoom callout** — reuse one source for an overview and a differently cropped detail linked by native annotation.-83. **Panel with a real hole punched through it (Subtract window)** — a solid or tinted panel with a shape-cut opening that reveals the image below, PowerPoint's Merge Shapes 剪除.+67. **Painted knock-out** — cover part of an image with the matching background or another prepared visual; use this imitation only when the surrounding field makes it credible.- **Geometry**: one `<path>` containing both contours, running in **opposite directions**. Outer clockwise, inner counter-clockwise — e.g. panel `M 80,80 H 1200 V 640 H 80 Z` followed by hole `M 420,220 V 500 H 760 V 220 H 420 Z` (note the second one descends first, reversing the winding). Under nonzero winding the reversed subpath subtracts, producing a true hole, so the effect never needs `fill-rule` and stays inside the [`shared-standards-core.md`](./shared-standards-core.md) §1.2 boundary. Verified end-to-end: both subpaths survive into a single `<a:path>` in the exported `custGeom`. The `subtract` operation of `shape_boolean_svg.py render` emits this contour directly; follow [`native-shape-authoring.md`](./native-shape-authoring.md) §6.+68. **Text-as-subtraction** — reveal an image or field through glyph-shaped holes; materialize supported text Boolean geometry through [`native-shape-authoring.md`](./native-shape-authoring.md).- **Why not #67**: that pattern fakes the opening by laying a background-colored shape on top. It works only over a flat background and silently breaks the moment the page gains a gradient, a texture, or a second image behind the panel. A real hole also lets the underlying image be moved or swapped without recutting the panel.+83. **Panel with a true hole** — subtract an opening from a foreground panel so changing content behind it remains valid; follow [`native-shape-authoring.md`](./native-shape-authoring.md) §6.-84. **Deliberately misregistered fragments (Fragment look)** — the inverse of #82. Cut one image into pieces using several `<image>` elements that share the same source, each with its own clip, then **break the alignment on purpose**: offset a few px, rotate 1–3°, or nudge one piece's scale. The eye still assembles one photo, but the seams now read as intentional — misprint, torn paper, glitch.+84. **Deliberately misregistered fragments** — separate same-source fragments and break their alignment intentionally for torn, misprint, or glitch language.- Keep the displacement small and consistent in direction; large or random offsets stop reading as a decision and start reading as a rendering bug. The `fragment` operation of `shape_boolean_svg.py render` returns each atomic region as a separately addressable path when the pieces must be individually positioned; follow [`native-shape-authoring.md`](./native-shape-authoring.md) §6.+85. **Subject breaking out of a container** — layer a registered foreground subject across its frame boundary; use an already prepared cutout when the composition requires one.-85. **Subject breaking out of its container** — the subject sits half inside a card / grid cell / color panel and half outside its boundary. Two `<image>` elements from the same file: one clipped to the container (optionally tinted, #31), one clipped to only the escaping region, positioned so the two halves stay in perfect register. Produces depth with no shadow at all.+90. **Scrim with true cutouts** — subtract image-reveal openings from a full-canvas scrim; lettering and complex cuts follow [`native-shape-authoring.md`](./native-shape-authoring.md) §6.- Let the *subject* be what escapes, not a corner of background, and break out only once per page — a page where everything escapes has no frame left to break.+95. **Background-registered shape fill** — fill a stationary shape with the page background sampled in root coordinates so it impersonates a hole while remaining an object.-26. **Triptych baked into a single wide image** — one wide `<image width=1160 height=334>` whose internal composition already contains 2–3 scenes. Generate the triptych as one image (not three separate calls) when scene-to-scene consistency matters — the model preserves character identity, lighting continuity, and color grading far more reliably when panels are produced together.+82. **One image across detached shapes** — export one native picture with disjoint clip subpaths so one continuous scene spans every shape.-## Overlay, Scrim & Vignette Treatments+100. **Same-source addressable crops** — export several independent native pictures that share an exact source coordinate system; follow [`executor-image.md`](./executor-image.md) §1.-**Hard rule — visual masking is not SVG `<mask>`**: Masking in a design brief-names the intended appearance only. Realize it with crop/clip geometry,-scrim/overlay shapes, a real cutout path, or a baked-alpha asset; never emit-`<mask>` or `mask="url(...)"`.+The following three patterns are topologically different and are not interchangeable:-> **Default — focal-safe text contrast (may override when the image and treatment demonstrably remain legible).** `preserveAspectRatio="xMidYMid slice"` center-crops whatever the source aspect ratio does not cover, so estimate the crop before placing text. Keep copy clear of the focal subject and maintain readable contrast across its full area. A gradient transition is valid when those conditions hold; use an opaque plateau or solid panel only when the image and softer treatment cannot guarantee them. When subject position is unresolved, prefer the opaque treatment rather than guessing.+| ID | Sources | Exported picture topology | Visual relationship |+|---|---|---|---|+| One-picture compound crop (`#82`) | One source | One native picture with disjoint clip subpaths | One continuous scene spans detached shapes; fragments are not independent picture objects |+| Addressable same-source crops (`#100`) | One exact source reference | Several independently addressable native pictures | Crops share one source coordinate system and remain in exact registration; follow [`executor-image.md`](./executor-image.md) §1 |+| Different-source split tiling (`#92`) | Different sources | Several independent picture objects in interlocking cells | The parent contour unifies peers; scene continuity across cells is not implied |-27. **Linear gradient scrim for text legibility** — `<linearGradient>` in `<defs>` (set `x1/y1/x2/y2` for direction) + overlay `<rect fill="url(#grad)">`. Most common is top-to-bottom darkening on full-bleed cover images.+### 4.2 Contrast, Color, and Depth-28. **Radial gradient vignette** — `<radialGradient cx cy r>` with dark outer stops; overlay `<rect>`. Focuses attention by darkening the periphery.+27. **Linear-gradient scrim** — add directional contrast while retaining image detail.-29. **Two-stop scrim — opaque on text side, transparent on focal side** — `<linearGradient>` with one stop at `stop-opacity="0.9"` and another at `stop-opacity="0"`. Use when text sits on one side and the image's subject on the other.+28. **Radial vignette** — darken the periphery to emphasize the central field.-30–31. **Flat overlay wash** — one `<rect fill-opacity>` over the image: neutral `#000000` / `#FFFFFF` around 0.4 for uniform darkening or lightening, the simplest scrim there is (**#30**), or a deck color at 0.15–0.25 to pull a foreign-looking photo toward the palette without regenerating it (**#31**).+29. **Opaque-to-transparent directional scrim** — protect copy on one side while keeping the focal side clear.-> **Sample the scrim color from the photo itself.** For any gradient scrim over an image (#27, #29, #31, #32, #90), take the solid end's hex from a dominant color *in that image* rather than defaulting to black or a deck color, and slide the gradient stop until the seam between scrim and photo disappears. A black scrim over a warm photo announces itself as a rectangle; a scrim in the photo's own shadow tone reads as part of the picture. This one substitution is the difference between a page that looks masked and one that looks composed.+30–31. **Flat washes** — **#30** uniformly darkens or lightens with a neutral wash; **#31** uses a palette color to integrate a foreign-looking image.-98. **Grid scrim with per-cell opacity** — instead of one flat or gradient scrim, cover the image with a grid of adjacent rectangles and give each cell a *slightly different* opacity (say 10–40 %, varied irregularly). The photo shows through unevenly, so the overlay reads as texture — panelled glass, a pixel field, a contact sheet — rather than as a sheet of paint. Text sits on the denser cells.+32. **Multi-hue gradient scrim** — shift color temperature or bridge image regions with a multi-stop field.- Keep the variation small and non-repeating: a regular light/dark alternation reads as a checkerboard, and a wide spread reads as broken rendering. Butt the cells exactly (no gaps, no strokes) so the grid is felt rather than drawn. Distinct from #50 / #88, where every cell holds its own image; here one image lies beneath one grid of glass.+33. **Radial spotlight** — keep a selected region clear while surrounding content recedes.-99. **Selective desaturation — colour only where it matters** — the whole image is muted while one subject stays in full colour, which fixes the focus of a busy photo without cropping it. Two registered copies: a desaturated (and usually darkened) version filling the frame, and the colour original clipped to just the subject region, sitting exactly on top.+36. **Shadow under an image panel** — lift the panel through a supporting shape; follow [`svg-effects.md`](./svg-effects.md).- **Both copies are baked assets** — there is no runtime colour filter on the native route ([`svg-effects.md`](./svg-effects.md) §6.12), so produce the desaturated file with a one-line Pillow `ImageEnhance.Color(img).enhance(0)` pass rather than reaching for `feColorMatrix`. Clip the colour copy along a real edge in the picture (the subject's own contour, per #96) — a rectangular colour patch over a desaturated field reads as an accident.+37. **Glow on an overlay shape** — emphasize an overlay boundary within the supported effect contract.-32. **Multi-stop scrim with hue shift** — three-or-more-stop `<linearGradient>` where stops are different colors (e.g. dark navy → transparent → warm orange). This re-grades the image's color world without regenerating — particularly useful when an AI image came back with the right composition but wrong color temperature.+57 · 60 · 61. **Receded image fields** — **#57** is a texture wash, **#60** ambient atmosphere, and **#61** a watermark behind body copy.-90. **Full-canvas scrim with shapes cut out of it (the cover / divider formula)** — the single highest-yield formula in this catalog, and the one real decks reuse most: a full-slide `<path>` whose outer contour is the canvas and whose inner subpath(s) are cut out using the opposite-winding rule from #83, laid over a full-bleed image. The scrim mutes the photo everywhere except through the cuts, so one ordinary image becomes a designed page. Three elements total: image, scrim, title.+66. **Fade into a solid background** — match the fade endpoint to the page background so the image edge disappears.- **The cut contour** — any of these, all authored as reversed inner subpaths in the same `<path>`:+91. **Faceted gradient form** — build a folded or ribbon-like object from adjacent facets with consistent light logic.- | Contour | Reads as |- |---|---|- | Wave, arc, ribbon (one soft curve across the page) | Editorial banner / horizon |- | Freehand closed curve (irregular, hand-drawn) | Organic torn-paper window |- | An array of hexagons / trapezoids / circles | Rhythmic screen, a window wall |- | Oversized numeral or letterform | Chapter marker (see caveat) |+98. **Grid scrim with varied cell opacity** — modulate one underlying image through a seamless grid of translucent cells.- An array of cuts is just several reversed subpaths in the same `d` — the same construction as #82, except here the image shows *through* the holes rather than being clipped *into* the shapes.+### 4.3 Placement and Framing- **Paint the scrim** with either a flat light fill at 0.15–0.25 opacity (white over a photo is the reliable default) or, for a directional reveal, a gradient that varies `stop-opacity` rather than color (`1 → 0.8 → 0`), so the image emerges progressively instead of through one hard boundary. Add a 1–2px stroke in the same light color on the cut edge to keep it crisp.+58. **Decorative corner fragment** — use a cropped image fragment as a secondary corner accent.- **Edge thickness**: to make the cut read as a physical opening, apply `feDropShadow` with `dx="0" dy="0"` and a small `stdDeviation` to the scrim path. Per [`svg-effects.md`](./svg-effects.md) §6.4 a zero-offset shadow is classified and exported as a **glow**, not a shadow — so use an accent or light color; black will read as diffuse haze rather than an edge. Never apply it to the `<image>` itself (#36).+59. **Image divider band** — replace a line between content regions with a narrow visual strip.- **Numeral / lettering**: keep the glyph as ordinary horizontal `<text>`, then use `shape_boolean_svg.py render --operation subtract` to resolve its exact font face and cut the shaped outline from the scrim ([`native-shape-authoring.md`](./native-shape-authoring.md) §6). The result is editable freeform geometry, not editable text. Do not hand-trace glyphs; use a baked RGBA fallback only when the font or text structure cannot resolve.+69. **Editorial rotation** — rotate an image or its container slightly when the style benefits from an informal print gesture.- **Motion pairing**: the scrim stays fixed while the image beneath drifts slowly (a 4–10s linear path, left, starting with the previous animation) — the cuts then behave like windows onto a moving world. That is an animation-stage decision, not page design; see [`animations.md`](./animations.md). It pairs with this pattern more often than with any other.+70–71. **Frames** — **#70** traces the image with one restrained outline; **#71** repeats nearby outlines for a layered photo-print treatment.-97. **Frosted-glass panel over the photo** — a legibility panel that is neither a flat scrim (#30) nor a full blur: a region of the image itself, blurred and lightened, sitting under the text while the rest of the photo stays sharp. It keeps the photo's color and composition visible through the panel, which a solid scrim destroys.+86. **Contour echo** — reuse a non-rectangular clip contour as an offset stroke instead of boxing it in a rectangle.- **Build it from a baked asset** — runtime blur does not survive native export ([#34](#), [`svg-effects.md`](./svg-effects.md) §6.12). Produce a blurred copy of the source (a Pillow `GaussianBlur` at a large radius, plus a brightness lift), then place that copy clipped to the panel contour and registered to the same position as the base image, so the blur lines up exactly with what is behind it. Add a thin light stroke and, if the style wants it, a slight lightening overlay.+### 4.4 Prepared-Asset Treatments- The panel must stay in register with the base photo; a frosted panel showing a *different* part of the scene is the classic tell. Pair with #95 when the panel should also carry a floating edge.+**Prepared-asset gate**: every treatment below consumes its named project-local asset; it does not authorize creation during SVG realization. In particular, `#96`, `#97`, and `#99` require their derivative pairs in advance. If a required asset is absent, return to the active workflow's preparation owner or choose a native treatment.-33. **Radial spotlight overlay — clear region surrounded by darkness** — cover the canvas with `<rect>` filled by a `<radialGradient>` whose inner stop is fully transparent and outer stop is opaque dark. Reads as a flashlight beam on the focal area. Use sparingly — it kills everything outside the spotlight.+26. **Baked triptych** — use one prepared wide source containing coordinated internal scenes; distinct from separate-image **#56**.-34. **Gaussian-blur backdrop** — blur the background in the source image, then layer sharp SVG content above it. Native filter export maps the supported blur graph to a glow/shadow effect; it does not preserve a blurred-image backdrop.+34. **Blurred-image backdrop** — use a prepared blurred asset; runtime image blur is not the backdrop mechanism.-35. **Duotone treatment** — two-color mapping of a photograph (e.g. deep navy shadows + warm cream highlights). Bake it into the source image; the native PPT route does not support a runtime duotone filter chain.+35. **Duotone photograph** — use a prepared two-color image treatment.-36. **Drop shadow under image panel** — `<filter><feDropShadow dx="0" dy="4" stdDeviation="6" flood-color="#000000" flood-opacity="0.10"/></filter>` applied to the image panel's backing `<rect>`. Standard depth lift; filters do not apply directly to `<image>` under the project contract.+63. **Transparent sticker or cutout** — use a prepared RGBA asset and preserve its open silhouette.-37. **Inner / outer glow on overlay shape** — `<filter><feGaussianBlur stdDeviation="6"/><feMerge/></filter>` on a shape, or simply a slightly larger blurred `<rect>` underneath the target.+64. **Artwork with embedded lettering** — use only for deliberate, fixed lettering that belongs to the artwork.-## Image as Texture / Atmosphere+65. **Text-free artwork with native labels** — keep authoritative or editable labels outside the asset as native SVG.-57 · 60 · 61. **Image pushed into the background** — the same move at three intensities: a full-bleed texture wash under the page (**#57**, overlay `<rect fill="bg-color" fill-opacity="0.7–0.85"/>`), low-contrast ambient atmosphere that is seen but never read (**#60**), or a watermark sitting behind body copy (**#61**). Suppress it with an overlay `<rect>` or a pre-dimmed asset — never a runtime filter.+72. **Soft image-to-image blend** — use a precomposited or baked-alpha asset when arbitrary images must blend per pixel.-58. **Image fragment as decorative corner element** — small `<image>` (often with `clipPath`) placed in one corner; not the focus, just visual seasoning.+89. **Sharp subject over a receded full-frame copy** — register two references to the same source; any blurred or desaturated derivative must already be prepared.-59. **Image as horizontal divider band** — narrow `<image height=80–150>` placed between two text sections instead of a `<line>` divider.+96. **Cutout subject re-laid over its source photo** — use a base photo plus its prepared transparent subject cutout, aligned in one coordinate system.-## Special Techniques+97. **Frosted-glass image panel** — use a base photo plus its prepared registered blurred crop beneath the text panel.-62. **Same image, two references — full view + zoom-callout** — reference the same image file twice in two `<image>` elements: one shows the full scene at normal size; the second uses `clipPath` (circle or rectangle) plus a larger display size to "zoom into" a sub-region. Connect them with a straight `<line>` or an exact native bent/curved Connector contour; use a custom Bézier only when the leader must avoid meaningful image content. Ring the zoom with a `<circle stroke>` so it reads as a magnifying lens. No special asset needed — the zoom effect comes from same-source-different-display.+99. **Selective desaturation** — use a prepared desaturated base plus a registered color subject layer.-63. **Transparent PNG sticker / cutout** — an RGBA PNG placed via plain `<image>`; the transparency lives in the file, so no `clipPath` is needed. Sources: `slice_images.py --alpha` output (see [image-generator.md](./image-generator.md) §4.3), an AI backend with native transparent output, or a user asset.+### 4.5 Cross-Page Continuity- Never box a cutout in a rectangle — that throws away the only thing it offers. Combine with #4 (bleed off the edge), #58 (corner fragment), #66 (fade into background), #69 (slight rotation), or #49 (asymmetric collage).--64. **Image with embedded text rendered by the AI** — text becomes part of the artwork: decorative lettering, artistic wordmark, hand-lettered keyword. Prompt with explicit text content — name the exact characters literally. Use for text that is part of the artwork and will not change. Authoritative titles and anything that must stay correct or editable go in the SVG `<text>` layer (#65).--65. **Image with NO text — labels added as native SVG** — generate the image with explicit "no text, no letters, no numbers, no signs" instruction (`text_policy: none`), then place all labels as `<text>` overlays. The right call when labels will be reworded, must stay exact, or carry data that must stay editable — pair with `#64` when stable visual identifiers (axis labels, subplot letters, unit symbols) belong inside the image instead.--66. **Image fading into the solid background** — soften the image's edge into the deck's background color via a `<linearGradient>` overlay whose end-stop matches the background hex exactly. The image's rectangular boundary disappears, producing seamless integration.--67. **Image with knock-out / cut-out shape** — overlay a shape filled with the background color or another image, creating the impression of a hole punched through the underlying image.--68. **Text-as-mask over image** — letterforms revealing an image, gradient, texture, or solid field through them. Author the foreground as a closed shape and the lettering as supported direct `<text>`, then materialize Subtract with `shape_boolean_svg.py`; this produces an editable freeform with real glyph-shaped holes, without SVG `<mask>`. Use a pre-rendered RGBA treatment only when the font/text cannot resolve or the lettering must remain a fixed artistic asset.--69. **Image rotated at a slight angle for editorial feel** — `transform="rotate(angle cx cy)"` on the `<image>` or its container `<g>`; 2–6 degrees typical. Adds dynamism without breaking layout.--70–71. **Frames** — a single `<rect fill="none" stroke="#color" stroke-width="2–6"/>` at the image edge (**#70**), or several nested outlines at slightly different sizes for a photo-print look (**#71**). When the image was cut to a non-rectangular contour, use #86 instead so the frame follows the cut.--72. **Baked-alpha image-to-image blend** — a genuinely soft blend between two images requires a precomposited bitmap or source images with baked alpha. An ordinary gradient overlay can conceal the join only when both images fade through the same solid bridge color; it is not a per-pixel mask and cannot blend arbitrary imagery.--95. **Shape filled with the page background itself** — the most-used trick in real decks and the one that has no obvious SVG name. A shape is painted not with a color but with *the page's own background, sampled at the shape's own position*, so it becomes invisible against the page while still being a real object that can carry an edge treatment.-- **SVG form**: give the shape the same `<image>` as the page background, positioned in root coordinates exactly as the background is, and clip it to the shape contour (§1.2). Because the fill stays registered to the page rather than to the shape, the object reads as a hole in whatever is above it.-- **Registration boundary**: the sampled shape and page background must remain fixed in the same root coordinates. Moving, resizing, rotating, or morphing the sampled shape moves its pixels with it and exposes the seam; animate independent content above or below the stationary shape instead.-- Three things it buys you, all of which otherwise require a second asset:- - **A cut that keeps the scene continuous** — the shape "removes" a foreground panel and shows the background through it, with no seam even over a photo or gradient.- - **A stationary conceal/reveal patch** — it can cover one fixed region while independent content enters or leaves above or below it.- - **Edge-only forms** — the shape disappears but its stroke, glow, or shadow remains, giving a floating outline that appears cut into the page.-- Distinct from #83 (a panel with a real hole) and #90 (a scrim with cuts): those remove paint, this one *impersonates* the background. Reach for it when the thing above must stay a solid object.--96. **Cutout subject re-laid over its own photo** — the mechanism behind every "subject escapes the frame" page (#85), and worth stating on its own because it is a two-asset technique: keep the original photo as the background layer, and place a background-removed PNG of its subject on top, in perfect register.-- Once the subject exists as a free-floating layer, it can overlap anything drawn between the two copies: a title the subject stands in front of, a color panel it steps out of, a shape frame it breaks through, a grid line it crosses. The base photo can be tinted, desaturated, blurred (baked), or scrimmed as hard as the layout needs, because the sharp subject on top is what the eye reads.-- Register is everything — the cutout must sit exactly where the subject sits in the base image; a few px of drift reads as a printing error. Keep the cutout's own edge clean rather than adding a stroke, unless the design calls for the sticker look of #63.--89. **Same image twice — sharp cutout over a receded full-bleed copy** — the single best answer to "the photo is too narrow / too short for this canvas, and stretching distorts the subject". Reference the same file twice: the bottom copy fills the whole canvas (or panel) and is pushed back; the top copy is clipped to a shape (#82, #24, a slanted band, a folded contour) at native proportions and stays sharp. The subject reads at full fidelity while the background extends the frame to any aspect ratio — no stretching, no letterbox bars, no second asset.-- **Recede the bottom copy with what survives export**: a color-tinted or darkened overlay `<rect>` (#30 / #31) at 0.5–0.8, or a desaturated / lowered-brightness variant of the file. **Blur does not survive** — per #34 the native route does not preserve a blurred-image backdrop, so if the design depends on blur it must be baked into a second image file (a one-line Pillow `GaussianBlur` pass over the original is enough); never rely on a filter at export time. Keep both copies in register — same center, same crop logic — or the trick reads as two unrelated photos.--86. **Contour echo — the clip path reused as a stroke** — after clipping an image (#20–#25, #82, #83), reuse the *same* `d` as a `<path fill="none" stroke="accent"/>`, drawn slightly larger or offset a few px. The outline repeats the cut geometry instead of boxing it in a rectangle, which is what #70 / #71 do. One extra element, no new asset. Offset it in a single consistent direction across the page; an echo on every side reads as a border, not an echo.--91. **Faceted gradients for folded / dimensional form (origami, ribbon, folded band)** — build a folded or faceted object from several adjacent `<path>` facets, then give each facet its own `<linearGradient>` whose direction and lightness differ from its neighbours — one face catching light, the next in shade. The fold is created by the *lightness break between adjacent facets*, not by any shadow effect, so it survives export intact as ordinary shapes.-- Keep every facet on one hue and vary only lightness (a white → light-grey → white ramp across three facets already reads as a crease), remove all strokes so the facets meet seamlessly, and keep the light direction consistent across the whole object. Combine with #82 by using the assembled facet outline as the clip contour, which puts a photo inside the folded form. Do not reach for `<filter>` shadows to fake depth here — [`svg-effects.md`](./svg-effects.md) owns effect limits, and the gradient break is both cheaper and more reliable.--87. **One image panned across consecutive pages** — a single wide image referenced by 2–4 consecutive slides, each showing a different horizontal segment (same `<image>` file and container geometry per page, only `x` shifts). Static on its own, it makes the deck read as one continuous scene; the audience recognizes the place before reading a word.-- **Motion contract**: keep the same image file and compatible direct-root group/container geometry on every participating page. Exporting with `-t morph` alone leaves object matching to PowerPoint's heuristic; stable ids and compatible geometry improve the chance of a camera pan but do not prove it. When the pan must be deterministic, run the custom motion stage and declare the adjacent objects in `animations.json` `morph.pairs` ([`animations.md`](./animations.md) §2.1); the pair may bind different source/destination ids while preserving compatible object kinds. Changing the file or endpoint geometry still changes the visual action and may reduce an unpaired Morph to a cross-fade.+87. **Cross-page image pan opportunity** — show different regions of one wide image across consecutive pages so the audience recognizes one continuous place. Keep the source and framing relationship coherent; if motion is enabled, [`animations.md`](./animations.md) owns its implementation.----## Composition Guidance--A page is built by layering. Pick one or more **Primary Structures** (Part 1) as the page's bones, then add any number of **Modifier Layers** (Part 2) for finish. Both stack — the question on each page is "is the next layer still earning its place", not "have I exceeded a quota".--**Cross-primary combinations are encouraged.** A side-by-side comparison (#48) where each side is annotated with Shape-first leader cards (#38) is one page, not a violation. A 3×3 grid (#9) whose center cell is upgraded to an image-as-canvas with KPI overlay (#40) reads as one composition. The old reflex "one primary per page" tends to under-use the catalog — combine when the page asks for it.--**Reference — motion-aware layer vocabulary, not a constraint**: When focus, comparison, evidence, or reveal order serves the page, the Image-as-Canvas + Native Overlay and Multi-Image Compositions families may expose independently meaningful visible units. `#62` can separate full view from same-source detail; `#63` can isolate a cutout foreground; `#74` / `#77` / `#78` / `#80` can separate image-led navigation or evidence units. These are composition layers, not effect assignments, and no pattern owes animation. `#72` is a static image blend in the fully revealed page, not a PowerPoint page transition.--**Modifier stacking pattern that works in practice** — observed on real content pages combining one Primary with four Modifiers:+## 5. Composition-- one Primary from Part 1 (e.g. #48 side-by-side comparison)-- `#21` rounded-rectangle clipPath on the image (rx=6 or circle)-- `#27` top-edge linearGradient in the deck's accent color, opacity 0.55 → 0-- `#66` bottom-edge linearGradient fading to background color, opacity 0 → 0.95-- small color-block badge + reversed-out label replacing any opaque color bar that would otherwise sit over the image+**Reference — not a constraint**: start with the content relationship, choose a Primary Structure, then add only Modifier Layers that clarify hierarchy, focus, continuity, or material character. Combining primaries is valid when both remain legible; for example, `#48` may use `#38` annotations, and the center of `#9` may use `#40`.-Combine freely. The "AI-default" failure mode is the opposite: defaulting to bare #2 / #3 (left/right split) with no Modifier at all.+**Registration boundary**: registration-dependent effects succeed only when their declared coordinate relationship remains exact. Preserve registration for `#62`, `#82`, `#85`, `#89`, `#95`, `#96`, `#97`, `#99`, and `#100`; `#84` is the intentional exception.-**Reference — image-led promotional deck moves (not a constraint)**:--| Page intent | Pattern candidates |-|---|---|-| Cover / ending with strong atmosphere | `#73` + `#27` / `#30` only if contrast needs it |-| Visual table of contents | `#74` + `#30` / `#31` |-| Chapter divider | `#75` |-| Venue / destination overview | `#76` or `#78` |-| Many product/place photos | `#77` or `#50` when equality is the message |-| Service / feature comparison | `#79` |-| Benefits with one dominant proof image | `#80` |-| Light promotional page without photos | `#81` |--**Reference — not a constraint**: before adding another photo, consider whether one prepared image plus #82–#100 can express the idea more clearly. Registration and prepared-asset boundaries remain mandatory when the chosen technique depends on them.--**Cross-page through-line (recurring motif).** The patterns above are per-page, but a deck reads as *designed* when one illustration motif family recurs across pages—a cover anchor, section dividers repeating the motif (`#75`), and small `#63` spots threaded through the body. Keep one family (shared rendering / locked deck colors / subject world), vary scale and placement, and never turn recurrence into a quota.--## Hard Constraints--- Page chrome, body copy, captions, and data values that must remain exact or editable stay in SVG. Stable figure-internal identifiers, axis/unit labels, panel markers, or lettering that is deliberately part of the artwork may be image-owned under `text_policy: embedded`, regardless of script or length.-- Project-wide SVG compatibility rules start at [`shared-standards-core.md`](./shared-standards-core.md),- whose routing table names each conditional owner. This catalog neither- restates nor relaxes that contract; each pattern records only its- scenario-specific rendering choice.-----+**Formula placement**: treat a rendered formula as a prepared visual asset. Use whitespace patterns such as `#16` or `#19` for isolated derivations, `#10`, `#41`, or `#46` for annotated formulas, and `#47` or `#48` for comparisons; keep editable explanatory text native.-For sizing math (calculating container dimensions from image aspect ratio when using side-by-side intent), see [`image-layout-spec.md`](image-layout-spec.md). This file is the design vocabulary; that file is the dimension calculator.+All compatibility details remain owned by [`shared-standards-core.md`](./shared-standards-core.md) and its routed references.diff --git a/skills/ppt-master/references/image-layout-spec.md b/skills/ppt-master/references/image-layout-spec.mdindex eb82d9f6..fab789a4 100644--- a/skills/ppt-master/references/image-layout-spec.md+++ b/skills/ppt-master/references/image-layout-spec.md@@ -1,239 +1,184 @@-> See [`shared-standards-core.md`](./shared-standards-core.md) for common technical constraints.+> See [`svg-image-embedding.md`](./svg-image-embedding.md) for SVG image syntax and crop-policy enforcement.# Image Layout Specification-Sizing reference for side-by-side or multi-image pages. Use after the active resource owner's preferred composition and its communication job have been understood; this file never locks layout or crop policy.+Neutral geometry and review rules for every image or rendered-formula placement. This file calculates the selected composition; it never chooses a resource, pattern, or automatic left/right or top/bottom layout.-**Preferred pattern, authoring-owned realization**: Preserve the proposed communication job and binding resource/content/crop constraints, then deepen its expression for the actual page. Let original aspect ratio inform the container. Every slide using a `no-crop` asset keeps one complete visible instance; a same-slide same-source detail crop may supplement it. An `adaptive` asset may use `meet` or a focal-safe `slice`. Use the current hierarchy, copy volume, declared focal content, and deck rhythm to develop the geometry or choose another composition; do not mechanically transcribe the named pattern. A pattern-only change needs no upstream update.--> **Scope**: The ratio tables and formulas are calculation aids for a selected side-by-side or multi-image plan. Hero, background, accent, and other compositions stay outside this file. Layout never overrides the `no-crop` boundary owned by [`strategist-image.md`](./strategist-image.md) and [`executor-image.md`](./executor-image.md).+**When to run**: whenever an image or rendered formula will be placed. Use the current page composition to select its region first, then apply the relevant single-item, adjacent, overlay, or multi-item calculation below.----## Layout Decision Flow+## 1. Ownership and Inputs-```-1. Understand the narrative intent, hierarchy, communication job, and preferred primary/modifier ids from the active resource authority; do not translate the ids directly into geometry.-2. Deepen that concept into the actual composition from the current copy volume, asset dimensions and declared focus, crop safety, and deck rhythm. If the authoring-selected result is not side-by-side or multi-image, this spec does not apply.-3. Read the asset's `no-crop` boundary and original dimensions; calculate ratio (width/height).-4. Use the tables as candidate structures, not an automatic selector.-5. Calculate the image/text rectangles, then choose `meet` or focal-safe `slice` within the crop boundary.-6. Revise geometry or choose another composition when the result weakens hierarchy, legibility, or required image content.-7. Return to the active resource owner only for a different resource, role, must-use decision, crop boundary, or another binding constraint; SVG authoring owns pattern-only realization changes.-```+| Role | Owns |+|---|---|+| Default Strategist | Resource choice, semantic role, crop boundary, and preferred image/content or image/shape relationship |+| Image_Generator | Composition inside each generated bitmap for its planned container |+| Default Executor | Final SVG regions and geometry; may adapt the preferred relationship while preserving binding resource, content, and crop constraints |+| Quick Generate main agent | The planning and realization decisions above in one active context |-**When to run**: after `analyze_images.py` has produced current dimensions and a side-by-side or multi-image composition is under consideration. Skip this sizing reference for other page structures.-----+This specification and [`image-layout-patterns.md`](./image-layout-patterns.md) are the always-read geometry and composition vocabulary; [`svg-image-embedding.md`](./svg-image-embedding.md) owns embedding. After selecting a construction, conditionally load only its additional technical owner: [`svg-effects.md`](./svg-effects.md) for an adopted effect and [`native-shape-authoring.md`](./native-shape-authoring.md) for adopted preset or Boolean geometry.-## Layout Starting Points (side-by-side intent)+### 1.1 Geometry notation-| Image Ratio | Useful Starting Structure | Image Position | Description |-|-------------|-------------|----------------|-------------|-| > 2.0 (ultra-wide) | Top-bottom split | Top full-width | Image spans canvas width, height proportional |-| 1.5-2.0 (wide) | Top-bottom split | Top | Image width = content area width, height proportional |-| 1.2-1.5 (standard) | Left-right split | Left | Image height-first fit, width proportional |-| 0.8-1.2 (square) | Left-right split | Left | Image takes content area height, width proportional |-| < 0.8 (portrait) | Left-right split | Left | Image height = content area height, width proportional |+| Symbol | Meaning |+|---|---|+| `(x0, y0, W, H)` | Current selected page region |+| `(ws, hs)` | Measured source width and height |+| `R = ws / hs` | Source aspect ratio |+| `Q = W / H` | Selected-region aspect ratio |+| `g`, `gx`, `gy` | Gap between adjacent regions, columns, or rows |+| `ax`, `ay` | Horizontal and vertical anchor fractions in `[0,1]` |-> Boundary ratios are orientation cues, not thresholds. Let text volume, focal content, page hierarchy, and crop safety decide.+All dimensions must be finite and positive. Derive `R` from current measured source data rather than a requested or previously planned size.----## Dimension Calculation Formulas+## 2. Aspect-Ratio Placement-### Canvas Parameters (All Formats)+### 2.1 Contain-| Format | Canvas | Margins (L/R, T/B) | Content Area (W x H) | Title Height | Content Start Y |-|--------|--------|--------------------|-----------------------|-------------|----------------|-| PPT 16:9 | 1280x720 | 60, 60 | 1160 x 600 | 60px | 80px |-| PPT 4:3 | 1024x768 | 50, 50 | 924 x 608 | 60px | 70px |-| Xiaohongshu | 1242x1660 | 60, 80 | 1122 x 1500 | 80px | 100px |-| WeChat Moments | 1080x1080 | 60, 60 | 960 x 960 | 60px | 80px |-| Story | 1080x1920 | 60, 120/180 | 960 x 1620 | 80px | 140px |-| WeChat Article | 900x383 | 40, 40 | 820 x 303 | 40px | 50px |+Contain keeps the complete source visible inside `(W,H)`:-> Below, **W** = content area width, **H** = content area height (excludes title). PPT 16:9 example: W=1160, H=600.--### Top-Bottom Layout Calculation+```text+if R >= Q:+ w = W+ h = W / R+else:+ h = H+ w = H × R+x = x0 + ax × (W - w)+y = y0 + ay × (H - h)```-Image width = W = 1160 px-Image height = W / R = 1160 / R px-Text area height = H - image height - gap(20px)-Review: if the remaining text area cannot carry the planned copy legibly,-rebalance the rectangles or choose another composition while preserving binding-resource/content/crop constraints.-```+Centered contain uses `ax = ay = 0.5`. SVG realization normally maps this to a legal `meet` anchor.-### Left-Right Layout Calculation+### 2.2 Fill-**Method 1 (height-first, suitable for portrait images)**:-```-Image height = H = 600 px-Image width = H x R = 600 x R px-Text area width = W - image width - gap(20px)-```+Fill covers `(W,H)` without distortion and crops overflow:-**Method 2 (width-constrained, for wide images converted to left-right)**:-```-Image width = W x 0.7 = 812 px-Image height = image width / R-Text area width = W - image width - gap(20px)+```text+if R >= Q:+ h = H+ w = H × R+else:+ w = W+ h = W / R++overflow_x = w - W+overflow_y = h - H+x = x0 - ax × overflow_x+y = y0 - ay × overflow_y```-**Review**: if the remaining text area cannot carry the planned copy legibly, rebalance the image/text rectangles or choose another composition while preserving binding resource/content/crop constraints.+Centered fill uses `ax = ay = 0.5`. SVG realization normally maps this to a legal `slice` anchor. Use fill only when the active crop boundary permits the computed loss and the anchor protects the declared focal content.----+### 2.3 Mode selection-## Layout Examples+| Need | Geometry |+|---|---|+| Complete source, formula, evidence, or edge content | Contain |+| Region coverage with a focal-safe crop | Fill |+| Complete source plus a detail view | One contain placement plus a separately justified crop |+| Irregular or repeated source windows | Apply the selected region math first, then load the owning crop/shape reference |-### Ultra-wide Image (ratio 2.45)+----```-Original: 1960x800, R=2.45 → Top-bottom split-Image: 1160x473, Text area: 1160x147 → 7:3 top-bottom-```+## 3. Single Image or Formula-### Standard Landscape (ratio 1.38)+Place a standalone item by applying §2 to its selected region. The region itself comes from the page hierarchy; source ratio determines the item geometry inside it, not the page structure.-```-Original: 1614x1171, R=1.38 → Left-right split-Image: 773x560 (left), Text area: 367x560 (right) → 7:3 left-right-```+For an item adjacent to another region, divide only the available selected region. Let `q_item` and `q_other` be positive visual weights for the image/formula and the other content.-### Wide Image Edge Case (ratio 1.75)+### 3.1 Horizontal adjacency-```-Original: 1820x1040, R=1.75-The resource owner compares top-bottom: image height=663, text area=-43 ❌-The resource owner recommends left-right: image 780x446 (left), text area 360x600 (right) → 7:3 left-right+```text+available = W - g+item_width = available × q_item / (q_item + q_other)+other_width = available - item_width```----+Both regions use height `H`. Place either region first according to the selected composition; no fixed share is implied.++### 3.2 Vertical adjacency-## Portrait Canvas Override+```text+available = H - g+item_height = available × q_item / (q_item + q_other)+other_height = available - item_height+```-Default selection table assumes **landscape or square canvas**. For portrait canvases (height > width), left-right splits leave both columns too narrow — use the override below.+Both regions use width `W`. Place either region first according to the selected composition.-| Canvas Orientation | Image Ratio | Useful Starting Structure | Reason |-|-------------------|-------------|-------------------|--------|-| Portrait (Xiaohongshu, Story) | > 1.5 (wide) | Top-bottom | Same as landscape canvas |-| Portrait (Xiaohongshu, Story) | 1.2-1.5 (standard) | Top-bottom | Left-right too narrow on tall canvas |-| Portrait (Xiaohongshu, Story) | 0.8-1.2 (square) | Top-bottom | Image fits well in top half |-| Portrait (Xiaohongshu, Story) | 0.5-0.8 (portrait) | Left-right | Portrait image on tall canvas works |-| Portrait (Xiaohongshu, Story) | < 0.5 (extreme portrait) | Left-right | Image takes one side, text the other |+### 3.3 Overlay and inset-> Square canvases (WeChat Moments 1:1): use the standard landscape rules.+An overlay keeps the image region and overlay region independently measurable. An inset selects a child region `(xi, yi, Wi, Hi)` inside the current region, then reapplies §2 using the same source ratio. Do not derive either region from an assumed percentage; size it from the actual hierarchy, copy, focal content, and required separation.----## Multi-Image Layout+## 4. Multiple Images-For slides with multiple images, divide the content area evenly using the formulas below.+### 4.1 Equal grid-### Grid Formulas+For `c` columns and `r` rows:-```-columns = number of columns-rows = number of rows-gap = 20px (PPT formats) or 30px (social formats)+```text+cell_width = (W - (c - 1) × gx) / c+cell_height = (H - (r - 1) × gy) / r-cell_width = (W - (columns - 1) * gap) / columns-cell_height = (H - (rows - 1) * gap) / rows+cell_x(col) = x0 + col × (cell_width + gx)+cell_y(row) = y0 + row × (cell_height + gy)```-### Common Patterns+Use equal cells when peer comparison is the message. Apply contain or fill independently to each source within its cell.-| Image Count | Layout | Grid | Description |-|-------------|--------|------|-------------|-| 2 (both landscape) | Side-by-side | 2x1 | Two equal columns |-| 2 (both portrait) | Stacked | 1x2 | Two equal rows |-| 2 (mixed) | 1 large + 1 small | Custom | Landscape top (full-width), portrait right-bottom |-| 3 | 1 large + 2 small | 1+2 | Left large (50% width), right column with 2 stacked |-| 4 | Grid | 2x2 | Equal-sized cells |+### 4.2 Weighted tracks-### Example: 2x2 Grid on PPT 16:9+For column weights `u[1]…u[c]` and row weights `v[1]…v[r]`:-```-W=1160, H=600, gap=20-cell_width = (1160 - 20) / 2 = 570-cell_height = (600 - 20) / 2 = 290+```text+available_width = W - (c - 1) × gx+available_height = H - (r - 1) × gy-Image positions:- (60, 80) 570x290 (650, 80) 570x290- (60, 390) 570x290 (650, 390) 570x290+column_width[j] = available_width × u[j] / sum(u)+row_height[k] = available_height × v[k] / sum(v)```-> Multi-image slides: decide `meet` or focal-safe `slice` per asset. On every slide using a `no-crop` source, keep one complete instance; a same-slide same-source detail crop may supplement it. Do not force every image into the same scaling mode merely for grid uniformity.+Use weighted tracks when one item is primary. A spanning item receives the sum of its tracks plus the internal gaps it crosses.------## Composition Checks+### 4.3 Free multi-item composition-| Check | Action |-|-----------|-----------------|-| Proportion does not reflect information weight | Rebalance image and text rectangles |-| Container conflicts with the native ratio | Change the container, choose `meet`, or use a focal-safe crop |-| Required pixels, labels, identity, or evidence would be cropped | Use a legal anchor with `meet` and recompose around the complete image |-| Text area cannot carry the planned copy legibly | Increase its area or choose another composition while preserving binding constraints |+For montage, arc, overlap, or another non-grid arrangement, assign one explicit region to every item and verify the union against `(W,H)`. Reuse one gap/rhythm system where separation is intended; overlap is explicit geometry, not a negative-gap accident.----## Handoff Fields--This spec only defines layout calculation. Keep computed fields in the active resource authority defined in [`svg-image-embedding.md`](svg-image-embedding.md):--| Field | Meaning |-|-------|---------|-| `Ratio` | Original image width / height |-| `Layout pattern` | Non-empty owner-resolved suggestion in free-form prose, optionally citing catalog ids; SVG-authoring realization |-| `Crop Policy` | `no-crop` requires one complete instance; `adaptive` lets SVG authoring choose `meet` or focal-safe `slice` |-| `Reference` | Optional calculated image/text rectangles, focal notes, and composition intent |-| Default `spec_lock.md images` value | `<path> | source=<Acquire Via> | pattern=<Layout pattern> | crop=<adaptive|no-crop>`; source/crop exactly project §VIII, while pattern preserves the normalized free-form suggestion and any optional catalog ids as a recommendation, not a geometry/realization lock |-| Quick Generate transient row | Keep the same source/pattern/crop meanings in active context; do not create a lock |+## 5. Rendered Formula Geometry-For SVG `<image>` syntax, path rules, `preserveAspectRatio`, external refs, and Base64 embedding: see [`svg-image-embedding.md`](svg-image-embedding.md).+Treat a rendered formula as an aspect-ratio source and apply contain within its selected mathematical region. Centering is the default geometric anchor; align to a nearby baseline or relation only when the page composition defines that relationship.-### SVG Image Embedding Examples+For `n` vertically stacked formula regions with equal lanes:-Complete display (`no-crop` assets such as data charts):--```xml-<image href="../images/xxx.png"- x="60" y="80" width="780" height="446"- preserveAspectRatio="xMidYMid meet"/>+```text+lane_height = (H - (n - 1) × g) / n+lane_y[i] = y0 + i × (lane_height + g)```-**Hard rule — no-crop placement**: On every slide using the source, retain one visible complete instance with one of the nine legal anchors plus `meet`, never `none`, and no `clip-path`, `mask`, clipping overflow, or nested `<svg>` viewport. An auxiliary same-slide detail or lens may crop the same source only while the complete instance remains visible. Definitions and hidden nodes are not placements; an image materialized through a visible local `<use>` is.--Crop-to-fill (an `adaptive` asset with a verified focal-safe crop):--```xml-<image href="../images/bg.png"- x="0" y="0" width="1280" height="720"- preserveAspectRatio="xMidYMid slice"/>-```+Contain each formula independently in its lane. When formulas are visual peers, a common effective scale may improve comparison; otherwise let their selected regions reflect their semantic weight and source ratios.----## Automation Tool--```bash-python3 scripts/analyze_images.py <project_path>/images # Infer project canvas; fallback PPT 16:9-python3 scripts/analyze_images.py <project_path>/images --canvas ppt43 # PPT 4:3-python3 scripts/analyze_images.py <project_path>/images --canvas xiaohongshu # Xiaohongshu-```--`--canvas` explicitly overrides the project-derived format; `ppt169` is only the fallback. The tool computes a top-bottom / left-right candidate, image display area, and text area from the formulas above. Treat its output as decision input; retain the composition actually selected for the page in the active authority.-----+## 6. Composition Checks-## Role Responsibilities+| Check | Required response |+|---|---|+| Computed width or height is non-positive | Re-select the page regions or reduce gaps |+| Contain leaves unusable residual space | Recompose the surrounding regions; do not stretch the source |+| Fill removes focal or required content | Change anchor, enlarge the region, or use contain |+| Adjacent text/content region cannot carry its material | Reweight or change the selected relationship |+| Equal cells imply equality that the content does not have | Use weighted tracks or a free composition |+| Peer images use inconsistent visual scale without meaning | Normalize their regions or make the hierarchy explicit |+| Formula symbols become unreadable at the intended viewing size | Enlarge its region or restructure the page |+| Gaps, alignments, or overlaps drift without purpose | Recalculate from the shared region and gap values |-| Role | Responsibility |-|------|---------------|-| **Default Strategist** | Run `analyze_images.py`, recommend a catalog pattern, select resources, and record the crop boundary |-| **Default Executor** | Choose the actual composition for the asset/page while preserving role, source, must-use, content, and `no-crop` constraints |-| **Quick Generate main agent** | Complete the same analysis/selection in active context before SVG authoring, then keep the chosen resource/crop boundary stable while realizing the page |+The final geometry must express the active page hierarchy, preserve the selected resource relationships, and remain valid under the conditionally loaded technical contracts.diff --git a/skills/ppt-master/references/modes/showcase.md b/skills/ppt-master/references/modes/showcase.mdindex 138e5107..132cddf7 100644--- a/skills/ppt-master/references/modes/showcase.md+++ b/skills/ppt-master/references/modes/showcase.md@@ -24,7 +24,7 @@ Titles are short and evocative — a phrase, not a sentence.- Generous negative space around the primary visual relationship.- Bold use of the deck's theme color for atmosphere (cover / chapter pages).-> Hero / full-bleed / breathing-page geometry lives in [`executor-base.md`](../executor-base.md) and the optional [`image-layout-patterns.md`](../image-layout-patterns.md) library; this mode decides *what each page makes primary*.+> Hero / full-bleed / breathing-page geometry lives in [`executor-base.md`](../executor-base.md) and the compact [`image-layout-patterns.md`](../image-layout-patterns.md) vocabulary loaded by the image branch; this mode decides *what each page makes primary*.---diff --git a/skills/ppt-master/references/strategist-image.md b/skills/ppt-master/references/strategist-image.mdindex bedf2815..fa77ce9e 100644--- a/skills/ppt-master/references/strategist-image.md+++ b/skills/ppt-master/references/strategist-image.md@@ -49,16 +49,14 @@ Follow `latex_render.py --help` for the manifest fields. The renderer writes dim## 4. Image Resource List-Add §VIII rows for the image resources actually planned from the confirmed source boundary and for every selected formula; a formula-only plan contains only formula rows. A permitted but unused source needs no row. Author each row's filename, dimensions/ratio, preferred layout suggestion, crop policy, purpose/type, acquisition, status, reference, and conditional AI fields as part of the complete Design Spec. `Acquire Via` is `ai`, `web`, `user`, `formula`, `placeholder`, or `slice`; status follows [`svg-image-embedding.md`](./svg-image-embedding.md). When a planned or explicitly required asset is not yet available, retain its row as `Pending` or `Needs-Manual`; never remove the row or change `Acquire Via` to make the Design Spec look complete. After §VIII passes final confirmation, project every placed row into `spec_lock.md images` as `<path> | source=<Acquire Via> | pattern=<Layout pattern> | crop=<adaptive|no-crop>` and omit unplaced Illustration Sheets. `source` and `crop` preserve the exact confirmed §VIII text; `pattern` preserves the non-empty free-form suggestion, including any optional catalog ids, while remaining preferred expression rather than locked geometry. References describe visual intent: AI uses subject + intent + composition without repeating rendering or HEX; web records exact subject, view/mood, focal/quiet region, and crop safety with positive quality cues; Image_Searcher later derives a separate short, specific provider query without rewriting this locked intent, while complete entity names or necessary disambiguation may use more words; formula preserves source LaTeX and placement intent.+Add §VIII rows for the image resources actually planned from the confirmed source boundary and for every selected formula; a formula-only plan contains only formula rows. A permitted but unused source needs no row. Author each row's filename, dimensions/ratio, preferred layout suggestion, crop policy, purpose/type, acquisition, status, reference, and conditional AI fields as part of the complete Design Spec. `Acquire Via` is `ai`, `web`, `user`, `formula`, `placeholder`, or `slice`; status follows [`svg-image-embedding.md`](./svg-image-embedding.md). When a planned or explicitly required asset is not yet available, retain its row as `Pending` or `Needs-Manual`; never remove the row or change `Acquire Via` to make the Design Spec look complete. After §VIII passes final confirmation, project every placed row into `spec_lock.md images` as `<path> | source=<Acquire Via> | pattern=<Layout pattern> | crop=<adaptive|no-crop>` and omit unplaced Illustration Sheets. `source` and `crop` preserve the exact confirmed §VIII text; `pattern` preserves the non-empty free-form suggestion, including any optional catalog ids, while remaining preferred expression rather than locked geometry.-**Prepared-user fast path**: For initial imported or user-supplied assets confirmed as `provided`, copy the exact `Filename` basename and derive `Dimensions` / `Ratio` from that row's EXIF-corrected `Width` / `Height` / native `AspectRatio` in the latest `analysis/image_analysis.csv`; `SourceDisplayRatio` is source-context metadata, not the bitmap crop ratio. Drop source-side directories, set `Acquire Via: user` and `Status: Existing`, and decide the remaining §VIII fields normally. Existing §VIII / lock / provenance-manifest records override this inference. Assets declared as `ai`, `web`, `slice`, `formula`, or manual fulfillment retain that provenance and advance through their own status lifecycle after entering `images/`; location never reclassifies them as `user / Existing`.--**Mandatory**: write one concise, non-empty, executable `Layout pattern` per non-formula row in ordinary language; optional ids from [`image-layout-patterns.md`](./image-layout-patterns.md) must be exact.+References describe visual intent: AI uses subject + intent + composition without repeating rendering or HEX; web records exact subject, view/mood, focal/quiet region, and crop safety with positive quality cues; Image_Searcher later derives a separate short, specific provider query without rewriting this locked intent, while complete entity names or necessary disambiguation may use more words; formula preserves source LaTeX and placement intent. Any subject direction, focal placement, quiet region, or overlay-safety requirement that must affect acquisition/generation belongs in `Reference` or the matching §IX block, not only in `Layout pattern`.-**Default — action-bearing image composition (may override when deliberate restraint is documented for the page)**: For a `page_role: hero_page` or otherwise image-led row, name a concrete image/content or image/shape relationship. Position, size, routine crop, and a legibility-only scrim alone do not constitute that relationship. A plain split or full-bleed treatment remains valid only when page-specific restraint is deliberate; state why in the matching §IX `Layout` / `Images`.+**Prepared-user fast path**: For initial imported or user-supplied assets confirmed as `provided`, copy the exact `Filename` basename and derive `Dimensions` / `Ratio` from that row's EXIF-corrected `Width` / `Height` / native `AspectRatio` in the latest `analysis/image_analysis.csv`; `SourceDisplayRatio` is source-context metadata, not the bitmap crop ratio. Drop source-side directories, set `Acquire Via: user` and `Status: Existing`, and decide the remaining §VIII fields normally. Existing §VIII / lock / provenance-manifest records override this inference. Assets declared as `ai`, `web`, `slice`, `formula`, or manual fulfillment retain that provenance and advance through their own status lifecycle after entering `images/`; location never reclassifies them as `user / Existing`.-**Reference — not a constraint**: open [`image-layout-patterns.md`](./image-layout-patterns.md) only when its vocabulary would expand the current options. Techniques needing a cutout, blurred crop, or desaturated copy require that prepared asset. Executor may adapt, replace, or decline the suggestion while preserving resource role, file/source, must-use status, crop boundary, content, and explicit user/template constraints; layout-only changes need no upstream rewrite.+**Mandatory**: write one concise, non-empty, executable `Layout pattern` per placed row, including formulas. Treat it as an adaptable page-realization preference: state the intended image/content or image/shape relationship when that relationship matters, while allowing a plain split, full bleed, contained formula region, or other restrained composition when it best serves the page. Optional ids from the already-read [`image-layout-patterns.md`](./image-layout-patterns.md) must be exact. Executor may adapt or replace the suggestion while preserving resource role, file/source, must-use status, crop boundary, content, and explicit user/template constraints; layout-only changes need no upstream rewrite.-Choose narrative intent before dimensions: hero/full-bleed, atmosphere/background, side-by-side, or accent/inline. Portrait and multi-image calculations belong to [`image-layout-spec.md`](./image-layout-spec.md). Write `Crop Policy: no-crop` whenever cropping could remove required pixels, labels, evidence, identity, or edge content; screenshots, charts, certificates/contracts, dense diagrams, logos, product markings, and formulas are common triggers rather than an exhaustive list. Otherwise write `Crop Policy: adaptive`: Executor may use complete display or a focal-safe crop, and the value never commands cropping. Formula rows use `Type: Latex Formula`, `Acquire Via: formula`, `Crop Policy: no-crop`, and `Rendered` or `Needs-Manual`.+Choose narrative intent before dimensions, then apply the already-read [`image-layout-spec.md`](./image-layout-spec.md) to the actual page region. Techniques needing a cutout, blurred crop, or desaturated copy require that prepared asset. Write `Crop Policy: no-crop` whenever cropping could remove required pixels, labels, evidence, identity, or edge content; screenshots, charts, certificates/contracts, dense diagrams, logos, product markings, and formulas are common triggers rather than an exhaustive list. Otherwise write `Crop Policy: adaptive`: Executor may use complete display or a focal-safe crop, and the value never commands cropping. Formula rows use `Type: Latex Formula`, `Acquire Via: formula`, `Crop Policy: no-crop`, and `Rendered` or `Needs-Manual`.Judge `text_policy` per AI row using [`image-generator.md`](./image-generator.md) §5.3; paper figures, academic schematics, panel comparisons, and data-axis graphics are positive triggers for reconsidering an all-`none` plan. Step 5 dispatches pending `ai` / `slice` rows to Image_Generator and pending `web` rows to Image_Searcher; formula rows bypass both.diff --git a/skills/ppt-master/scripts/analyze_images.py b/skills/ppt-master/scripts/analyze_images.pyindex 9a98c194..42aa4baf 100644--- a/skills/ppt-master/scripts/analyze_images.py+++ b/skills/ppt-master/scripts/analyze_images.py@@ -2,20 +2,12 @@"""Image Size Analysis Tool========================-Reports objective parameters (width, height, aspect ratio, category) for all-images in a folder. Intentionally does NOT prescribe a layout — the Strategist-decides narrative intent (hero / atmosphere / side-by-side / accent) per-references/strategist-image.md; this tool only supplies the numbers.--After resolving the canvas from the project, an explicit override, or the-ppt169 fallback, also reports the reference image/text area sizes that would-apply *if* an image is placed side-by-side with body text. Those numbers are-conditional on the Strategist picking the side-by-side intent.+Reports objective parameters for all images in a folder. It does not resolve a+canvas, prescribe a layout, or generate Strategist recommendations.Usage:python scripts/analyze_images.py <images_folder_path>python scripts/analyze_images.py projects/xxx/images- python scripts/analyze_images.py projects/xxx/images --canvas ppt43Output:- Analysis report displayed in console@@ -41,37 +33,11 @@ except ImportError:print("Error: PIL/Pillow not installed. Run: pip install Pillow")sys.exit(1)-try:- from config import CANVAS_FORMATS, LAYOUT_MARGINS-except ImportError:- CANVAS_FORMATS = {- 'ppt169': {- 'name': 'PPT 16:9',- 'width': 1280,- 'height': 720,- },- }- LAYOUT_MARGINS = {- 'ppt169': {- 'top': 60, 'right': 60, 'bottom': 60, 'left': 60,- 'content_width': 1160, 'content_height': 600- },- }--from project_utils import get_project_info, normalize_canvas_format-IMAGE_EXTENSIONS = {".jpg", ".jpeg", ".png", ".gif", ".webp", ".bmp", ".tiff", ".tif"}OFFICE_VECTOR_EXTENSIONS = {".emf", ".wmf"}REPORT_WIDTH = 100CATEGORY_WIDTH = 50-# Title area height and gap between image/text areas (px)-TITLE_HEIGHT = 60-LAYOUT_GAP = 20-# Minimum text area dimensions (px)-MIN_TEXT_HEIGHT = 150-MIN_TEXT_WIDTH = 280-ImageAnalysis = dict[str, object]@@ -219,7 +185,7 @@ def _result_from_manifest('ratio_source': 'manifest','format': Path(filename).suffix.lstrip('.').upper(),'has_transparent_pixels': None,- 'layout_hint': classify_ratio(ratio),+ 'category': classify_ratio(ratio),'filesize_kb': os.path.getsize(filepath) / 1024,}_apply_manifest_metadata(result, meta)@@ -238,11 +204,10 @@ def _result_from_manifest(def classify_ratio(aspect_ratio: float) -> str:- """Classify image aspect ratio into layout category.+ """Classify an image by its objective aspect-ratio range.- Thresholds aligned with image-layout-spec.md:- >2.0 ultra-wide, 1.5-2.0 wide, 1.2-1.5 standard landscape,- 0.8-1.2 square, <0.8 portrait.+ Ranges: >2.0 ultra-wide, 1.5-2.0 wide, 1.2-1.5 standard+ landscape, 0.8-1.2 near square, and <0.8 portrait."""if aspect_ratio > 2.0:return "Ultra-wide"@@ -256,77 +221,6 @@ def classify_ratio(aspect_ratio: float) -> str:return "Portrait"-def compute_layout_dimensions(- ratio: float,- content_w: int,- content_h: int,- gap: int = LAYOUT_GAP,-) -> dict:- """Compute image and text area dimensions following image-layout-spec.md.-- Returns dict with layout_type, image_w, image_h, text_w, text_h.- """- # Effective content height (below title)- H = content_h- W = content_w-- def _try_top_bottom() -> dict | None:- img_w = W- img_h = int(round(W / ratio))- text_h = H - img_h - gap- if text_h >= MIN_TEXT_HEIGHT:- return {- 'layout_type': 'top-bottom',- 'image_w': img_w,- 'image_h': img_h,- 'text_w': W,- 'text_h': text_h,- }- return None-- def _try_left_right_height_first() -> dict | None:- img_h = H- img_w = int(round(H * ratio))- text_w = W - img_w - gap- if text_w >= MIN_TEXT_WIDTH:- return {- 'layout_type': 'left-right',- 'image_w': img_w,- 'image_h': img_h,- 'text_w': text_w,- 'text_h': H,- }- return None-- def _try_left_right_width_constrained() -> dict:- img_w = int(round(W * 0.7))- img_h = int(round(img_w / ratio))- text_w = W - img_w - gap- return {- 'layout_type': 'left-right',- 'image_w': img_w,- 'image_h': min(img_h, H),- 'text_w': max(text_w, MIN_TEXT_WIDTH),- 'text_h': H,- }-- # Decision tree per image-layout-spec.md- if ratio > 1.5:- # Ultra-wide or wide → try top-bottom first- result = _try_top_bottom()- if result:- return result- # Fallback to left-right (wide-constrained)- return _try_left_right_width_constrained()- else:- # Standard landscape, square, portrait → try left-right (height-first)- result = _try_left_right_height_first()- if result:- return result- # Fallback to left-right (width-constrained)- return _try_left_right_width_constrained()--def _analyze_images(images_dir: str) -> tuple[list[ImageAnalysis], list[str]]:"""Analyze all image files in a directory.@@ -373,7 +267,7 @@ def _analyze_images(images_dir: str) -> tuple[list[ImageAnalysis], list[str]]:'ratio_source': 'native','format': image_format,'has_transparent_pixels': has_transparent_pixels,- 'layout_hint': classify_ratio(aspect_ratio),+ 'category': classify_ratio(aspect_ratio),'filesize_kb': os.path.getsize(filepath) / 1024}_apply_manifest_metadata(result, meta)@@ -411,25 +305,6 @@ def analyze_images(images_dir: str) -> list[ImageAnalysis]:return results-def enrich_with_layout(- results: list[ImageAnalysis],- canvas_key: str,-) -> None:- """Add computed layout dimensions to each result in-place."""- margins = LAYOUT_MARGINS.get(canvas_key)-- if not margins:- print(f"[WARN] No layout margins for canvas '{canvas_key}', skipping dimension calculation")- return-- content_w = margins['content_width']- content_h = margins['content_height']-- for img in results:- dims = compute_layout_dimensions(img['aspect_ratio'], content_w, content_h)- img.update(dims)--def print_results(results: list[ImageAnalysis]) -> None:"""Print the analysis report to stdout."""@@ -437,31 +312,26 @@ def print_results(results: list[ImageAnalysis]) -> None:print("Image Size Analysis Report")print("=" * REPORT_WIDTH)- has_layout = 'layout_type' in results[0] if results else False-- if has_layout:- print("\nNote: 'Img (SxS)' shows the image area *if* the Strategist chooses the")- print("side-by-side intent for this image. Decide narrative intent first — see")- print("references/strategist-image.md. Hero / atmosphere / accent intents ignore it.\n")- print(f"{'No.':<4} {'Width':<7} {'Height':<7} {'Ratio':<7} {'Source':<8} {'Refs':<5} {'Size':<10} {'Category':<20} {'Img (SxS)':<14} {'Filename'}")- else:- print(f"\n{'No.':<4} {'Width':<7} {'Height':<7} {'Ratio':<7} {'Source':<8} {'Refs':<5} {'Size':<10} {'Category':<20} {'Filename'}")+ print(+ f"\n{'No.':<4} {'Width':<7} {'Height':<7} {'Ratio':<7} "+ f"{'Source':<8} {'Refs':<5} {'Size':<10} {'Category':<20} {'Filename'}"+ )print("-" * REPORT_WIDTH)for i, img in enumerate(results, 1):ratio_source = str(img.get('ratio_source', 'native'))usage_count = int(img.get('usage_count', 1))- base = f"{i:<4} {img['width']:<7} {img['height']:<7} {img['aspect_ratio']:<7.2f} {ratio_source:<8} {usage_count:<5} {img['filesize_kb']:<10.1f}KB {img['layout_hint']:<20}"- if has_layout:- img_area = f"{img['image_w']}x{img['image_h']}"- print(f"{base} {img_area:<14} {img['filename'][:35]}")- else:- print(f"{base} {img['filename'][:40]}")+ base = (+ f"{i:<4} {img['width']:<7} {img['height']:<7} "+ f"{img['aspect_ratio']:<7.2f} {ratio_source:<8} {usage_count:<5} "+ f"{img['filesize_kb']:<10.1f}KB {img['category']:<20}"+ )+ print(f"{base} {img['filename'][:40]}")print("-" * REPORT_WIDTH)print(f"Total: {len(results)} images\n")- # Group statistics by aspect ratio (aligned with image-layout-spec.md thresholds)+ # Group statistics by objective aspect-ratio ranges.print("\nGroup by Aspect Ratio:")print("-" * CATEGORY_WIDTH)@@ -508,49 +378,6 @@ def print_results(results: list[ImageAnalysis]) -> None:print(f" ... and {len(native_only) - 10} more")-def generate_markdown(results: list[ImageAnalysis], canvas_key: str) -> None:- """Print a Markdown-ready image inventory section."""- print("\n" + "=" * REPORT_WIDTH)- print("Markdown Snippet for Strategist (Copy & Paste)")- print("=" * REPORT_WIDTH)-- has_layout = 'layout_type' in results[0] if results else False- fmt_name = CANVAS_FORMATS.get(canvas_key, {}).get('name', canvas_key)-- print(f"\n## Image Resource Inventory (Auto-scan Results — {fmt_name})\n")-- print("> Decide narrative intent per image (hero / atmosphere / side-by-side /")- print("> accent) per `references/strategist-image.md` before filling the table. The")- print("> `Img Area (SxS)` / `Text Area (SxS)` columns only apply if the chosen")- print("> intent is side-by-side; ignore them for hero / atmosphere / accent intents.\n")-- if has_layout:- print("| Filename | Size | Ratio | Category | Img Area (SxS) | Text Area (SxS) | Intent | Usage | Type | Status | Generation Description |")- print("|----------|------|-------|----------|----------------|-----------------|--------|-------|------|--------|-----------------------|")- else:- print("| Filename | Size | Ratio | Category | Intent | Usage | Type | Status | Generation Description |")- print("|----------|------|-------|----------|--------|-------|------|--------|-----------------------|")-- for img in results:- ratio_str = f"{img['aspect_ratio']:.2f}"- asset_kind = str(img.get('asset_kind', 'bitmap'))- image_type = "Office Vector" if asset_kind == "office_vector" else ""- status = (- "PPTX Native Only"- if asset_kind == "office_vector" and not img.get('svg_renderable', True)- else "Existing"- )-- if has_layout:- img_area = f"{img['image_w']}x{img['image_h']}"- text_area = f"{img['text_w']}x{img['text_h']}"- print(f"| {img['filename']} | {img['width']}x{img['height']} | {ratio_str} | {img['layout_hint']} | {img_area} | {text_area} | (to be filled) | {img.get('usage_count', 1)} refs | {image_type} | {status} | - |")- else:- print(f"| {img['filename']} | {img['width']}x{img['height']} | {ratio_str} | {img['layout_hint']} | (to be filled) | {img.get('usage_count', 1)} refs | {image_type} | {status} | - |")-- print("\n" + "=" * REPORT_WIDTH + "\n")--def _format_optional_number(value: object, digits: int = 2) -> str:"""Format a numeric value for CSV, leaving unavailable facts blank."""if not isinstance(value, (int, float)):@@ -558,16 +385,10 @@ def _format_optional_number(value: object, digits: int = 2) -> str:return f"{float(value):.{digits}f}"-def save_csv(- results: list[ImageAnalysis],- csv_path: str | Path,- include_layout: bool | None = None,-) -> None:+def save_csv(results: list[ImageAnalysis], csv_path: str | Path) -> None:"""Atomically save analysis results to a standards-compliant CSV file."""target = Path(csv_path)target.parent.mkdir(parents=True, exist_ok=True)- if include_layout is None:- include_layout = bool(results and "layout_type" in results[0])header = ["No","Filename",@@ -587,8 +408,6 @@ def save_csv("SizeKB","Category",]- if include_layout:- header.extend(["ImageArea_SxS", "TextArea_SxS"])temporary_path: Path | None = Nonetry:@@ -622,20 +441,8 @@ def save_csv(image.get("svg_renderable", True),image.get("pptx_native_supported", True),_format_optional_number(image["filesize_kb"], digits=1),- image["layout_hint"],+ image["category"],]- if include_layout:- image_area = (- f"{image['image_w']}x{image['image_h']}"- if "image_w" in image- else ""- )- text_area = (- f"{image['text_w']}x{image['text_h']}"- if "text_w" in image- else ""- )- row.extend([image_area, text_area])writer.writerow(row)os.replace(temporary_path, target)temporary_path = None@@ -646,37 +453,15 @@ def save_csv(print(f"\nCSV saved to: {target}")-def _resolve_canvas_key(images_dir: Path, override: str | None) -> tuple[str, str]:- """Resolve canvas from an explicit override, project context, or fallback."""- if override:- return normalize_canvas_format(override), "--canvas"-- project_dir = images_dir.parent if images_dir.name == "images" else images_dir- project_info = get_project_info(str(project_dir))- project_canvas = normalize_canvas_format(str(project_info.get("format", "")))- if project_canvas in CANVAS_FORMATS:- return project_canvas, "project"- return "ppt169", "fallback"--def main(argv: list[str] | None = None) -> int:"""Run the CLI entry point."""parser = argparse.ArgumentParser(- description="Analyze image sizes and compute PPT layout dimensions"+ description="Analyze objective image-file facts")parser.add_argument("images_dir",help="Path to the images directory")- parser.add_argument(- "--canvas",- help=(- "Canvas format override. By default, infer it from the project "- f"directory and fall back to ppt169. Available: "- f"{', '.join(sorted(CANVAS_FORMATS.keys()))}"- ),- )-args = parser.parse_args(argv)images_dir = Path(args.images_dir).resolve()@@ -688,31 +473,18 @@ def main(argv: list[str] | None = None) -> int:print(f"Error: Not a directory: {images_dir}")return 1- canvas_key, canvas_source = _resolve_canvas_key(images_dir, args.canvas)- if canvas_key not in CANVAS_FORMATS:- available = ", ".join(sorted(CANVAS_FORMATS.keys()))- print(f"Error: Unknown canvas format '{canvas_key}'. Available: {available}")- return 1-- fmt = CANVAS_FORMATS[canvas_key]print(f"Analyzing: {images_dir}")- print(- f"Canvas: {fmt.get('name', canvas_key)} "- f"({fmt.get('width', '?')}x{fmt.get('height', '?')}; {canvas_source})"- )results, errors = _analyze_images(str(images_dir))- enrich_with_layout(results, canvas_key)if results:print_results(results)- generate_markdown(results, canvas_key)else:print("No readable supported image files found in the directory.")analysis_dir = images_dir.parent / "analysis"csv_path = analysis_dir / "image_analysis.csv"- save_csv(results, csv_path, include_layout=canvas_key in LAYOUT_MARGINS)+ save_csv(results, csv_path)if errors:print(diff --git a/skills/ppt-master/scripts/docs/image.md b/skills/ppt-master/scripts/docs/image.mdindex c428b4ea..9e3c6aaa 100644--- a/skills/ppt-master/scripts/docs/image.md+++ b/skills/ppt-master/scripts/docs/image.md@@ -165,16 +165,29 @@ MINIMAX_API_KEY=your-api-key## `analyze_images.py`-Analyze images in a project directory before writing the design spec or composing slide layouts.+Analyze objective image-file facts in a project directory before writing the+design spec or authoring SVG.```bashpython3 scripts/analyze_images.py <project_path>/images-python3 scripts/analyze_images.py <project_path>/images --canvas ppt43```-Without `--canvas`, the tool resolves the project format and falls back to `ppt169`; the flag is an explicit override. The atomic CSV records EXIF-corrected native dimensions/`AspectRatio`, optional source `SourceDisplayRatio`, format, and actual transparent-pixel presence. Native ratio—not source display metadata—drives bitmap layout/crop. An empty folder rewrites a header-only report; unreadable supported files still refresh the report and produce a non-zero exit.--Use this as the default inventory and geometry source; it does not perform semantic image understanding. Generate planning follows the Strategist's context-first boundary: source context, captions / alt text / titles, filenames, user notes, and existing resource records come first. Only an already-selected provided/web asset whose focal-safe crop, overlay contrast, or quiet region remains materially ambiguous may be inspected for that placement; this never reopens selection or provenance, never bulk-opens the image folder, and never restores routine readback of AI-generated images.+The tool does not resolve a canvas or recommend a left/right, top/bottom, or+other slide layout. Its atomic CSV records EXIF-corrected native dimensions and+`AspectRatio`, the objective aspect-ratio category, optional source+`SourceDisplayRatio`, format, actual transparent-pixel presence, usage count,+and bitmap/vector capability facts. An empty folder rewrites a header-only+report; unreadable supported files still refresh the report and produce a+non-zero exit.++Use this as the default factual inventory; it does not perform semantic image+understanding or choose composition. Generate planning follows the Strategist's+context-first boundary: source context, captions / alt text / titles, filenames,+user notes, and existing resource records come first. Only an already-selected+provided/web asset whose focal-safe crop, overlay contrast, or quiet region+remains materially ambiguous may be inspected for that placement; this never+reopens selection or provenance, never bulk-opens the image folder, and never+restores routine readback of AI-generated images.## `image_search.py`diff --git a/skills/ppt-master/scripts/prompt_audit_manifest.json b/skills/ppt-master/scripts/prompt_audit_manifest.jsonindex 69cacd1f..76cf1e0c 100644--- a/skills/ppt-master/scripts/prompt_audit_manifest.json+++ b/skills/ppt-master/scripts/prompt_audit_manifest.json@@ -12,7 +12,7 @@"skills/ppt-master/templates/schemas/*.json"],"exclude": [],- "max_tokens": 402440+ "max_tokens": 392825},"file_budgets": {"AGENTS.md": 2400,@@ -215,7 +215,7 @@"stage.generate.executor.image"],"files": [],- "max_tokens": 43975+ "max_tokens": 46275},"route.generate.quick-generate.formula": {"description": "Quick Generate with manifest-driven formula rendering and formula-image realization.",@@ -227,7 +227,7 @@"files": ["skills/ppt-master/scripts/docs/image.md"],- "max_tokens": 47160+ "max_tokens": 49475},"route.generate.quick-generate.ai-two-types": {"description": "Quick Generate with AI resource preparation, representative rendering inputs, and two local image types.",@@ -238,7 +238,7 @@"stage.generate.executor.image"],"files": [],- "max_tokens": 67015+ "max_tokens": 69400},"route.generate.quick-generate.web-image": {"description": "Quick Generate with web resource preparation, provenance, attribution, and image execution.",@@ -249,27 +249,27 @@"stage.generate.executor.web-image"],"files": [],- "max_tokens": 51725+ "max_tokens": 54025},"route.generate.planning-image": {- "description": "Generate-PPTX planning context after a non-none image source is proposed or confirmed; layout-catalog recall remains optional.",+ "description": "Generate-PPTX planning context after a non-none image source is proposed or confirmed, including the compact layout math and vocabulary.","scope": "cumulative","include": ["route.generate.planning","stage.generate.strategist.image"],"files": [],- "max_tokens": 74850+ "max_tokens": 70275},"route.generate.planning-formula": {- "description": "Formula-only planning context with no non-formula image resource or layout catalog.",+ "description": "Formula-only planning context, including the shared image/formula layout math and vocabulary.","scope": "cumulative","include": ["route.generate.planning","stage.generate.strategist.image"],"files": [],- "max_tokens": 64975+ "max_tokens": 70275},"route.generate.planning-ai": {"description": "Generate-PPTX planning when AI image direction is available, including representative multi-source custom rendering synthesis.",@@ -289,7 +289,7 @@"registry": "image-renderings"}],- "max_tokens": 79100+ "max_tokens": 74500},"stage.generate.apply-template-workspace": {"description": "Conditional Step 3 workspace validation, installation, and fusion framework.",@@ -594,7 +594,7 @@"registry": "image-type-templates"}],- "max_tokens": 23050+ "max_tokens": 23125},"stage.generate.image.web": {"description": "Incremental web-image acquisition role.",@@ -785,23 +785,14 @@"max_tokens": 3100},"stage.generate.strategist.image": {- "description": "Conditional Strategist image/formula policy module without a resource-layout catalog.",+ "description": "Conditional Strategist image/formula policy with the always-read layout math and compact composition vocabulary.","scope": "incremental","files": [- "skills/ppt-master/references/strategist-image.md"- ],- "max_tokens": 2525- },- "stage.generate.strategist.image-layout": {- "description": "Optional image-layout catalog recall layered on image planning when the Strategist wants additional composition vocabulary.",- "scope": "incremental",- "include": [- "stage.generate.strategist.image"- ],- "files": [+ "skills/ppt-master/references/strategist-image.md",+ "skills/ppt-master/references/image-layout-spec.md","skills/ppt-master/references/image-layout-patterns.md"],- "max_tokens": 14675+ "max_tokens": 7825},"stage.generate.strategist.template": {"description": "Conditional Strategist module for an explicitly installed template workspace.",@@ -820,14 +811,15 @@"max_tokens": 6200},"stage.generate.executor.image": {- "description": "Conditional image embedding and layout execution rules; the optional layout catalog is not part of the required bundle.",+ "description": "Conditional image embedding and execution rules with the always-read layout math and compact composition vocabulary.","scope": "incremental","files": ["skills/ppt-master/references/executor-image.md","skills/ppt-master/references/image-layout-spec.md",+ "skills/ppt-master/references/image-layout-patterns.md","skills/ppt-master/references/svg-image-embedding.md"],- "max_tokens": 18425+ "max_tokens": 9425},"stage.generate.executor.web-image": {"description": "Conditional sourced-image attribution layered on image execution.",@@ -1012,7 +1004,7 @@"name": "image-layout-patterns","kind": "numbered_markdown","source": "skills/ppt-master/references/image-layout-patterns.md",- "entry_pattern": "^(?P<ids>\\d+(?:\\s*(?:[-–]|·)\\s*\\d+)*)\\.\\s+\\*\\*",+ "entry_pattern": "^(?:\\|\\s*`#)?(?P<ids>\\d+(?:\\s*(?:[-–]|·)\\s*\\d+)*)(?:`\\s*\\||\\.\\s+\\*\\*)","reference_terms": ["image-layout-patterns","image-text layout patterns"diff --git a/skills/ppt-master/workflows/generate-pptx.md b/skills/ppt-master/workflows/generate-pptx.mdindex 17165d7e..3fe08796 100644--- a/skills/ppt-master/workflows/generate-pptx.md+++ b/skills/ppt-master/workflows/generate-pptx.md@@ -176,9 +176,9 @@ Read references/strategist.md| Deterministic trigger | Additional Strategist reference ||---|---|| Step 3 installed an explicit Brand/Layout/Deck workspace | `references/strategist-template.md` |-| The core's proposed Stage 2 `image_usage` contains a source other than `none`, the user supplied an explicit non-`none` image constraint, or formula-worthy content activates formula planning | `references/strategist-image.md` before authoring image renderings, production detail, formula resources, or §VIII |+| The core's proposed Stage 2 `image_usage` contains a source other than `none`, the user supplied an explicit non-`none` image constraint, or formula-worthy content activates formula planning | `references/strategist-image.md` + `references/image-layout-spec.md` + `references/image-layout-patterns.md` before authoring image renderings, production detail, formula resources, or §VIII |-The core first chooses the proposed Stage 2 source ids. Load the image module before writing Stage 2 whenever that proposal is non-`none`; after confirmation, keep it active only for confirmed non-`none` sources or an active formula plan. A confirmed `none` path with no formula work writes no image rows. Bare template names and style language do not load the template module.+The core chooses proposed Stage 2 source ids first. Load this bundle for a non-`none` proposal; after confirmation, retain it only for confirmed non-`none` sources or an active formula plan. Confirmed `none` without formulas writes no image rows. Bare template names and style language do not load the template module.> ⚠️ **Mandatory artifact gates**: after final confirmation, author the complete `design_spec.md` from `templates/design_spec_reference.md`. After Gate 1, run enabled refinement and wait for approval; then author `spec_lock.md` from its reference, the approved Design Spec, and context. Create each new artifact once—no placeholder scaffold; `scaffold-*` commands remain manual-only. Schemas validate structure; semantic fidelity remains mandatory.@@ -210,7 +210,7 @@ If the user rejects the current recommendation before confirming it, regeneratepython3 ${SKILL_DIR}/scripts/confirm_ui/server.py <project_path> --wait-only --wait-stage stage1```-2. Read the Stage 1 result. Derive proposed image sources in core and load `strategist-image.md` before constructing Stage 2 when its trigger fires; apply `strategist-template.md` when active. Create `confirm_ui/recommendations.stage2.json` without changing Stage 1, then wait:+2. Read Stage 1. Derive proposed image sources, load the triggered image-planning bundle above, and apply `strategist-template.md` when active. Create `confirm_ui/recommendations.stage2.json` without changing Stage 1, then wait:```bashpython3 ${SKILL_DIR}/scripts/confirm_ui/server.py <project_path> --wait-only --wait-stage stage2@@ -249,7 +249,7 @@ For the normal/default `continuous` path, print no split-mode reminder and proce**Mandatory — spec-refinement note** (not another Confirm UI stage): after confirmation details and any split-mode line, append one localized 💡 line offering review of the complete Design Spec before the lock; any part may be revised in chat until explicit approval. Default OFF; only explicit chat opt-in or `refine_spec: true` runs [`refine-spec`](stages/refine-spec.md) after Gate 1. Confirm UI records the toggle; chat fallback prints the same line.-**Formula policy**: Stage 3 confirms `mixed`, `render-all`, or `text-only`. When the confirmed policy requires rendering formula-worthy content, load [`strategist-image.md`](../references/strategist-image.md) even if `image_usage` is `none`, and follow its formula-resource contract before filling the planning artifacts. `text-only` creates no formula image rows.+**Formula policy**: Stage 3 confirms `mixed`, `render-all`, or `text-only`. When rendering is required, load the image-planning bundle even if `image_usage` is `none`, then follow [`strategist-image.md`](../references/strategist-image.md)'s formula-resource contract. `text-only` creates no formula image rows.**Proactive production decisions**: Stage 3 records`proactive_speaker_notes`, `proactive_custom_animations`, and@@ -352,7 +352,7 @@ Workflow:- [ ] **Next**: open a fresh chat window and input `继续生成 projects/<project_name>` to enter the execution session via the [`resume-execute`](stages/resume-execute.md) stage.```-> On acquisition failure, follow [image-base.md](../references/image-base.md) §5 without halting. Web rows continue through materially different query/provider/license/URL strategies; after exhaustion, mark `Needs-Manual`, report, and continue.+> On acquisition failure, follow [image-base.md](../references/image-base.md) §6 without halting. Web rows continue through materially different query/provider/license/URL strategies; after exhaustion, mark `Needs-Manual`, report, and continue.---@@ -384,7 +384,7 @@ Read references/visual-styles/<resolved-id>.md # one preset id, or each `visu| `pptx_structure.mode: structured` | `executor-structured.md` + `pptx-structure-interface.md` || Any data chart/table, including mini or inset charts and sparklines | `executor-chart.md` || Preset pattern or supported native chart/table | `native-data-interface.md` before drawing |-| `spec_lock.md images` / §VIII has an image/formula row, or the template has bundled images | `executor-image.md` + `image-layout-spec.md` + `svg-image-embedding.md`; add `image-layout-patterns.md` only for a cited `#<id>` or optional composition recall |+| `spec_lock.md images` / §VIII has an image/formula row, or the template has bundled images | `executor-image.md` + `image-layout-spec.md` + `image-layout-patterns.md` + `svg-image-embedding.md` || At least one placed image has `Status: Sourced` | `executor-web-image.md` after the image branch || The locked style/current page calls for noncanonical or alpha paint, dash/cap/join, tracking/decoration/outline, gradient/filter/glow/shadow, path/transform/clipping, or another constructed effect | `svg-effects.md` before authoring that value or effect || §IX contains `Native shape suggestion`, or current page construction calls for a literal PowerPoint stock shape, a stock Connector contour, an explicit Merge Shapes operation/result (including a text, letterform, or numeral knockout), or a shape-built dimensional form (cylinder/pedestal, layered diagram, reflection, ground plane), or Executor is about to hand-author a freeform not already required by data geometry or the locked organic / hand-drawn style | `native-shape-authoring.md` before selecting or materializing that geometry. This trigger is image-independent: a text-only, data-only, or icon-only page reaches it the same way |diff --git a/skills/ppt-master/workflows/profiles/quick-generate.md b/skills/ppt-master/workflows/profiles/quick-generate.mdindex 6d65a20f..6531fa5a 100644--- a/skills/ppt-master/workflows/profiles/quick-generate.md+++ b/skills/ppt-master/workflows/profiles/quick-generate.md@@ -66,14 +66,13 @@ Before writing P01, resolve in active context:- the slide roster, canvas, visual direction, palette, typography, and wording;- a transient resource roster with page, filename, purpose, visual intent,- acquisition path, crop behavior, and status; for a hero page or other- image-led use, the visual intent includes an action-bearing composition;+ acquisition path, crop behavior, and status. For an image/formula, include+ its page relationship plus any subject position, focus, quiet region, or+ overlay-safety cue that must exist before SVG authoring;- the implementation path for each resource. An explicit user path wins;otherwise choose the registered automatic/default path without anotherinteraction.-**Default — action-bearing image composition (may override only for an explicit user/material boundary or a page-specific restraint decision)**: For a hero page or otherwise image-led resource, resolve one concrete image/content or image/shape relationship before SVG authoring. Position, size, routine crop, and a legibility-only scrim alone do not constitute that relationship. A plain split or full-bleed treatment remains valid only when the explicit boundary requires it or the agent records a page-specific restraint reason in active context; decide and continue without another interaction.-Prepare only the resource paths that the roster triggers:| Resource | Required preparation |@@ -104,9 +103,15 @@ silently replace it with unrelated material.Always read[`shared-standards-core.md`](../../references/shared-standards-core.md). Do notload `executor-base.md`: its persisted-plan prerequisites do not apply to this-profile. Load the conditional image/web/canvas references when needed, and load-`native-shape-authoring.md` for selected preset/Boolean construction such as-text knockouts.+profile. For any image/formula, always read+[`executor-image.md`](../../references/executor-image.md),+[`image-layout-spec.md`](../../references/image-layout-spec.md),+[`image-layout-patterns.md`](../../references/image-layout-patterns.md), and+[`svg-image-embedding.md`](../../references/svg-image-embedding.md); add+[`executor-web-image.md`](../../references/executor-web-image.md) for a sourced+web image. Load [`canvas-formats.md`](../../references/canvas-formats.md) only+for a non-default canvas, and load `native-shape-authoring.md` for selected+preset/Boolean construction such as text knockouts.Use one zero-padded filename width sized for the resolved roster, such as`01_cover.svg` through `12_end.svg` or `001_cover.svg` through `120_end.svg`.