| name | editorial-vision-studio |
| description | Universal visual direction engine for AI image, design, and layout work: model-agnostic decision pipeline (intent, analysis, visual language, planning, recovery/refinement) plus swappable adapters for GPT Image, Flux, Ideogram, and generic image backends. Use for visual concepting, image prompts, photo-to-design, posters, covers, zines, gallery prints, campaigns, brand key visuals, product/editorial imagery, social assets, website hero art, moodboards, presentation decks, Panter-style low-contrast recovery, multi-image series with a consistent visual system, or switching image models while preserving the same creative direction. |
Editorial Vision Studio
AI Creative Director for Visual Generation.
Philosophy: Do not decorate. Always interpret.
An extensible Editorial Design Engine: one decision pipeline, swappable model adapters. Every layer — intent, analysis, art direction, planning, recovery, compilation, review, iteration — is a separate module with a single job, and the contract between them is a model-agnostic spec.
Architecture: reference/architecture.md
When to Use
- User asks for AI image direction, image prompts, art direction, visual concepting, or prompt adaptation across models
- User uploads a photo and asks for photo-to-design, editorial poster, cover, zine, gallery print, campaign key visual, brand visual, product visual, or hero image
- User gives a theme only and wants a poster, social asset, zine, campaign, moodboard, or conceptual image
- User mentions low-contrast / gray photo recovery (Panter compensation)
- User specifies a style: Swiss, Kinfolk, MUJI, Brutalist, Wallpaper*, Purple, Apartamento, POPEYE
- User wants analysis → direction → prompt → image, not immediate generation
- User specifies model:
gpt-image, flux, ideogram — or asks to reuse direction with a different model
- User wants a set, not an image: campaign at all sizes, a carousel, a multi-page deck
- User wants a second image that matches the first — series consistency, brand system, "同一套視覺"
- User names a preset, or wants a specific fixed look reproduced exactly
Architecture: Decision Engine + Model Adapters
DECISION ENGINE (fixed) MODEL ADAPTER (swappable)
Intent → Style Gate VisionSpec / EditorialSpec
→ Analyzer → ↓
→ Visual Language → adapters/{model}.md
→ Art Direction → ↓
→ Planner → GenerationRequest → API
→ Recovery →
→ Compiler → VisionSpec
SHARED POST-LAYER
Reviewer → Optimizer → Generate
→ Evaluator → Iteration
│
one spec mutation, │
re-enter Compiler ◄─────────┘
(max 3 passes)
Core Pipeline
User Request
↓
Intent Engine → [prompts/intent.md](prompts/intent.md)
↓ (series? → [prompts/series.md](prompts/series.md) · memory? → [prompts/visual-memory.md](prompts/visual-memory.md))
Style Gate → [prompts/style-gate.md](prompts/style-gate.md) ← ask the user once, up front
↓
Visual Analyzer → [prompts/analyzer.md](prompts/analyzer.md) (skip if theme-only / prompt-only)
↓
Visual Language Engine → [prompts/visual-language.md](prompts/visual-language.md)
↓
Art Direction Engine → [prompts/art-direction.md](prompts/art-direction.md) ← 2–3 candidates, commit to one
↓
Visual Planner → [prompts/planner.md](prompts/planner.md)
↓
Recovery Engine → [prompts/recovery.md](prompts/recovery.md) + [recovery/](recovery/)
↓
VisionSpec → [spec/editorial-spec.schema.md](spec/editorial-spec.schema.md)
↓
Model Adapter → [adapters/registry.md](adapters/registry.md) ← swappable
↓
Prompt Reviewer → [prompts/reviewer.md](prompts/reviewer.md) ← is it legal?
↓
Prompt Optimizer → [prompts/optimizer.md](prompts/optimizer.md) ← is it well said?
↓
Image Generation
↓
Quality Evaluator → [prompts/evaluator.md](prompts/evaluator.md) → quality vector
↓
Iteration Engine → [prompts/iteration.md](prompts/iteration.md) ← fix one layer, loop (max 3)
↓
VisualManifest → [spec/visual-manifest.schema.md](spec/visual-manifest.schema.md)
Each layer does one job. Never analyze in Compiler. Never generate in Analyzer. Never re-decide in Iteration — mutate the spec and re-run the layer that owns the failure.
Quick routing: reference/decision-tree.md
Step 0: Intent Engine
Before analyzing pixels, resolve user goal → output family:
| User says | Intent | Allowed outputs |
|---|
| art book cover | Art Book | magazine-cover, gallery-print, poster |
| TEDx key visual | Event Campaign | campaign-poster, brand-key-visual, social-asset |
| skincare brand launch | Branding | brand-key-visual, product-editorial, social-asset, campaign-poster, website-hero |
| app hero image | Digital Product | website-hero, interface-asset, social-asset |
| zine page | Zine | zine, poster, editorial-spread |
| gallery print | Gallery | gallery-print, photo-abstract-diptych, poster |
| moodboard | Visual Concept | moodboard, poster, editorial-spread |
| 10-page deck | any family, purpose: presentation | presentation-deck |
Also resolve the six intent dimensions — subject, purpose, audience, emotion, platform, aspect ratio. Infer them; ask only when a missing one is load-bearing. They are what every later layer is graded against.
Detect scope here, not later: series_id for a set (prompts/series.md), memory_id for a continuation (prompts/visual-memory.md).
Read prompts/intent.md. Reject mismatched formats (e.g. gallery print for TEDx campaign).
Step 0.5: Style Gate
The first question about how the image should look. Before analysing a single pixel, ask which look they want — a numbered menu of the compatible presets plus 「讓 AI 提案」. (Intent may have asked about purpose or platform first; that is a different question.)
Half a step because it computes nothing. It asks one question and turns the answer into a lock.
要什麼風格？
1. 米色明信片 —— 米白紙底，照片重畫成簡化色塊，大量留白。
2. 時代海報 —— 飽和油墨滿版，硬邊平面色塊，地名做成版面。
3. 紙雕明信片 —— 寫實照片，明信片上長出立體紙雕世界。1:1。
4. 自己描述 —— 你說想要的樣子，我拆解照片後照你的描述重畫。
5. 讓 AI 提案 —— 看過照片和用途後給你三個方向再選。
Nobody chooses a look twice unless they asked to see proposals. The gate emits a single field, style_gate.outcome, and Art Direction branches on that and nothing else:
| Gate answer | outcome | Art Direction |
|---|
| A preset | commit | Becomes the session's Visual Memory; auto-commits, asks nothing |
| 自己描述, or just typing a description | commit | Captured verbatim; the Analyzer deconstructs the photo, then prompts/style-brief.md rebuilds it to the description. A description is never answered with a menu |
| 「讓 AI 提案」 | offer | Offers 2–3 candidates. This is the one path with two prompts, and the user asked for it |
It skips — outcome: commit, no menu — when a preset or style is already named, a memory or series is active, a direction was committed earlier this session, or the run is unattended. It skips to outcome: offer when fewer than two presets fit the intent, because a one-item menu is not a choice. Unattended runs pick the highest-fit preset with intent-only weights and say which in one line — never silently.
Presets declare intended_layouts; the gate offers only those that intersect intent.allowed_outputs.
Show looks, not internal vocabulary. Nobody outside this repo knows what Kinfolk means.
The gate settles ground and medium only. Layout, composition, abstraction, and copy stay with the Planner.
Read prompts/style-gate.md.
Step 1: Visual Analyzer
Produce structured Image Report with star ratings and Editorial Score (0–100).
Dimensions: subject, clarity, contrast, saturation, composition, negative space, geometry, texture, lighting, emotion.
Read prompts/analyzer.md.
Step 2: Visual Language Engine
Derive Visual Language first, then style/palette/layout — not the reverse.
Examples: Museum → Swiss or MUJI + fine serif; Quiet Human → Kinfolk + cream/sage; Indie Memory → Zine + riso anchor; Poster Graphic → Travel Poster + flat inks.
Read prompts/visual-language.md. User style: override skips auto-derivation but Reviewer still validates DNA fit.
Step 3: Art Direction Engine
Do not generate the first plausible reading of the brief. Draft competing directions, each with a one-sentence thesis and a named trade-off, score them for fit, commit to one, and keep the runner-up.
Branch on style_gate.outcome and nothing else. The gate already resolved every case — preset chosen, free text, 「你決定」, explicit style:, memory, series, unattended, too few presets — into one of two values:
outcome | Behaviour |
|---|
commit | Build one direction, commit silently. Never offer candidates — that would be the user's second time choosing a look. Name the runner-up in one line when one exists |
offer | Present 2–3 candidates and ask. Two for a narrow intent family, three for a theme-only brief or Editorial Score <50 |
The layer always runs, even on commit — every spec needs an art_direction block with a thesis.
Every candidate must set a different ground. Eight of the eleven style DNAs resolve to a light paper field, so candidates chosen on style alone come back as three shades of ivory. Beyond that, candidates must differ on at least two of: render mode, visual language, layout family, abstraction level, typography weight. Palette swaps are not directions.
The committed direction is a hard constraint on the Planner. Switching to the runner-up later re-runs Planner onward only — never the Analyzer.
Read prompts/art-direction.md.
Step 4: Editorial Planner
Decide layout, typography direction, abstraction level inside the committed direction — not the final prompt.
Key rules (full matrix in prompts/planner.md):
- Portrait + negative space >50% → Magazine Cover
- Architecture + strong geometry → Swiss Poster
- Landscape + quiet mood → Gallery Print
- Street + human story → Documentary Zine
- Food/object + minimal → Product Editorial
If user specifies style: kinfolk, load styles/kinfolk.md DNA.
Step 5: Recovery Engine
Apply only when Image Report flags weakness. Each recovery is one atomic fix — see recovery/.
| Problem | Recovery |
|---|
| Low contrast / gray (saturation <30%) | Panter Mode: warm/cool conflict hues, high-sat anchor, wider tonal separation |
| Weak subject | Increase silhouette / scale |
| Flat lighting | Directional light |
| Busy background | Simplify geometry |
| Too many colors | Limit palette to 4 |
| No focal point | Editorial color anchor |
| No rhythm | Abstract panel |
Panter Mode: discard dull grays; boost warm to 75% / cool to 70% saturation; add 8% high-chroma anchor block; widen tonal separation and mark scale. Panter is a colour compensation and never adds texture on its own. See recovery/contrast.md.
Texture Permission — single source of truth: assets/texture.md. Three tiers: PRINT (riso/halftone/scan defects) is zine only; SURFACE (substrate character such as cotton paper) is allowed on CLEAN layouts whose style DNA rates Texture ★★★+; FLAT (zero texture words) covers the photo-abstract-diptych panel ground, interface-asset, the website-hero copy-safe area, the product-editorial background, and every page of a presentation-deck that contains a data page. Recoveries never raise a layout's tier.
Never redesign the entire image unless Editorial Score <50 (Concept Reconstruction).
Step 6: Prompt Compiler + Model Adapter
Phase 1: Assemble VisionSpec / EditorialSpec — read prompts/compiler.md
Phase 2: Route to adapter by target.model:
User: model: flux or "用 Flux 生成" → set adapter, do not re-analyze.
Same direction, different model: reuse VisionSpec / EditorialSpec, swap adapter only.
Step 7: Prompt Reviewer
Before generation, run conflict detection. Read prompts/reviewer.md.
Examples:
- Swiss grid + Kinfolk organic → reject or resolve
- MUJI + heavy typography → reject
- Brutalist + soft pastoral palette → warn
Auto-correct incompatible pairings.
Step 7.5: Prompt Optimizer
The Reviewer answers is this prompt legal. It cannot fix a prompt that is legal but weakly worded — the ground clause buried in sentence four, a spec number softened to "a small accent", the same instruction stated twice.
Half a step because it decides nothing. Every clause it emits must trace to a field the spec already holds; a clause with no spec field behind it is a rejection back to the Compiler, not an invention. It never sets a ground, style, subject, layout, palette, or type scale.
Seven ordered ops: provenance_strip · ground_first · concretize · bind_numbers · dedupe · route_negatives · compress. Most gain comes from concretize and bind_numbers — a model renders nouns and numbers, never an adjective about taste.
This does not break the "mutate the spec, never the prompt string" rule in prompts/iteration.md. That rule forbids hand patches after generation. The Optimizer is deterministic, pre-generation, and recorded: same spec + same adapter + same ruleset_version produces the same string, so the manifest still replays.
No op firing is the normal outcome. The same op firing twice means the adapter emits that defect every time — fix adapters/{model}.md, not the string.
Variant mode emits 2–3 prompts from one spec, differing only on clause_density, specificity, and emphasis_order. A variant that differs on any spec field is a second direction, not a variant. Offer it only when the user asks to compare wordings, when the model is generic, or when a prompt-attributed dimension failed last run.
Read prompts/optimizer.md.
Step 8: Quality Evaluator
After generation, score a quality vector — not one number. Ten weighted dimensions: subject, composition, focal_point, palette, typography, texture, style_coherence, photo_fidelity, intent_fit, platform_fit. Each 0.00–1.00; inapplicable dimensions are null, never 0.
overall is the weighted mean and maps to grade A–D. Any single dimension below 0.60 fails the run regardless of overall.
The evaluator must name lowest_failing and responsible_layer — without them the iteration loop has nothing to act on.
Read prompts/evaluator.md.
Step 9: Iteration Engine
A low score points at one layer, not at the whole image. Fix that layer, recompile, regenerate.
lowest failing dimension → responsible layer → one minimal spec mutation → recompile → re-score
- One mutation per iteration. Two at once makes the next score unattributable.
- Mutate the spec, never the prompt string. A hand-patched prompt cannot be replayed from the manifest.
- Escalate when the same dimension fails twice: Compiler → Recovery → Planner → Art Direction → Intent.
- Re-enter at the Compiler after every mutation. "Responsible layer" names whose decision changed, not an entry point.
- Stop at
overall ≥ 0.85 with no dimension below 0.60, at 3 iterations, on score regression, or when escalation reaches Intent — then report, don't guess.
Taste disagreement ("I don't like the blue") is a direction change, not a QC failure — route it to prompts/art-direction.md.
Read prompts/iteration.md.
Ground & Render Mode
Two axes, required, with no default:
| Axis | Values | Question it answers |
|---|
design_tokens.ground | paper-light · neutral-gray · dark · saturated · full-bleed-photo · duotone | What is the canvas field itself? |
direction.render_mode | photographic · photo-plus-graphic · graphic · painterly · mixed | What medium is the image made of? |
Enum values are contract tokens, never prompt words — every adapter resolves them to prose through assets/ground.md.
An unset value is a rejection, not a fallback. This matters more than it looks: before these fields existed, an undecided ground fell through to warm ivory paper on every run, because that was the most-repeated value in the repo. A field with no default cannot be skipped.
render_mode is not abstraction_level. Medium and distance-from-source are independent — a photographic image can still be full-abstract.
Presets
A preset is a shipped VisualMemory with source: preset — a locked partial spec authored in the repo instead of established from a run. It reuses the entire memory mechanism; nothing new enforces it.
The Style Gate offers these as a numbered menu at the start of a run; preset: ivory-postcard is the shortcut that skips the menu, for automation and series work. Three different grounds is deliberate — it is what gives the candidate-diversity rule something to draw on. Registry and authoring guide: presets/registry.md.
Unlike a series memory, a preset may lock composition, aspect_ratio, and layout: a preset is avowedly a template, which is what it is for.
Visual Memory & Series
The engine keeps a visual system across images, not just within one.
Visual Memory locks the identity fields — style, visual language, palette, typography, texture tier — and leaves layout, composition, abstraction level, and aspect ratio free per image. Locking composition produces a template, not a system. On a memory run, Art Direction still runs but auto-commits to the locked DNA rather than offering candidates. A lock that no longer fits gets forked, never silently mutated. Contract: spec/visual-memory.schema.md.
Series runs Intent → Analyzer → Visual Language → Art Direction once, then fans out per output. Generate the hero first and pass QC before derivatives — a weak hero multiplies into N weak frames. Cross-image QC scores palette drift, typographic identity, texture tier, anchor legibility, and compositional variety; a single outlier is fixed alone, never by re-running the set.
Run Manifest
Every run emits spec/visual-manifest.schema.md: direction taken and runner-up, visual system, quality vector, iteration history, prompt hash, provenance.
This is what makes the cheap paths cheap:
| User says | Re-run |
|---|
| "同一張，改用 Flux" | Adapter → Reviewer → Generate |
| "改用 B 那個方向" | Planner → downstream |
| "一模一樣再生一次" | Generate only |
| "同一套視覺，換主題" | Analyzer → Art Direction (auto-commit) → Planner → downstream |
Never re-run the Analyzer when a valid image_report for the same source image already exists. Keep the manifest internal unless the user asks for it or the run is part of a series.
Editorial Score & Modes
This is the Editorial Score — an input measure of the source photo, produced by the Analyzer. It is not the Quality Score from Step 8, which measures the generated image on different bands.
| Editorial Score | Mode |
|---|
| 90+ | Premium Editorial — refined extraction, minimal recovery |
| 70–89 | Standard Editorial |
| 50–69 | Compensation Mode — apply Recovery stack |
| <50 | Concept Reconstruction — abstract reinterpretation |
Output Contract
Match the requested depth. Default to a concise direction summary plus GenerationRequest.
- Include an Image Report only when a source image is analyzed.
- Include full VisionSpec / EditorialSpec when the user asks for a reusable direction, comparison, or model switch.
- Include a generated image only when an image-generation tool is available and the user asks for generation; otherwise return the model-ready prompt.
- Include Quality Grade and evaluator notes after generating, or when the user requests review.
- Include the full quality vector only when the user asks why, or when a dimension failed.
- Include the optimizer op log only when the user asks what changed, or when it reverted. In variant mode show the prompts, never the op log — the user is choosing wordings, not reviewing edits.
- Include the VisualManifest only on request, or when the run belongs to a series.
- Present Art Direction candidates as a 3-line table — name, thesis, trade-off — never as raw YAML.
Model switch without re-analysis
User: "同一份方向，改用 Ideogram" → reuse VisionSpec / EditorialSpec, run adapters/ideogram.md only.
Direction switch without re-analysis
User: "改用 B 那個方向" → reuse Intent + Image Report, set the runner-up as selected, re-run Planner onward.
Series and continuation
User: "同一套視覺，做東京街景" → load the Visual Memory, run Analyzer on the new photo, let Art Direction auto-commit to the locked DNA, and Planner sets free fields only.
Bilingual output
- Image prompt: English (model-optimized)
- Analysis/direction summary: match user's language (中文/English)
Guardrails
Never:
- Redraw, filter, or stylize the original photo region when
photo_policy.fidelity: required
- Blindly copy fixed 60/30/10 layout — adapt proportions to subject
- Mix style languages without Reviewer pass
- Overload typography or decorative elements
- Let the Optimizer add a clause no spec field backs, or lengthen a prompt to strengthen it
Always:
- Preserve the source's arrangement, overlaps, and relative scale — abstraction removes detail, not relationships
- Name one projection, one ground plane, one light direction; give every grounded object a contact shadow
- Preserve visual identity of source photo when one is provided
- Make every abstract mark traceable to a photo fact, theme fact, brand cue, or stated goal
- Keep prompts imageable and concrete
- Apply Recovery only when Image Report warrants it
- Give each direction a thesis before giving it a palette
- Fix one layer per iteration, and stop at three
- Fork a Visual Memory rather than mutating a lock to rescue one image
Style & Layout Reference
Extending the Engine
| Extend | Action | Touch Decision Engine? |
|---|
| New style (Aesop, NYT Mag) | Add styles/foo.md | No |
| New preset | Add presets/foo.md + register | No |
| New layout | Add layouts/foo.md | No |
| New recovery | Add recovery/foo.md | No |
| New image model | Add adapters/foo.md + register | No |
| New intent family | Edit prompts/intent.md | Yes (minimal) |
| New QC dimension | Edit prompts/evaluator.md + map a layer in prompts/iteration.md | Yes (minimal) |
| New prompt-wording op | Add a row to prompts/optimizer.md + bump ruleset_version | No |
| New lockable DNA field | Edit spec/visual-memory.schema.md + enforce in prompts/reviewer.md | Yes (minimal) |
See adapters/_template.md for new models.
Extending Styles
Add new magazines/brands by creating styles/your-style.md with Style DNA table + compiler clauses. No need to rewrite SKILL.md.
Engine Reference
Agent Config
Model parameters: agents/openai.yaml
