Image Analysis Router

Image Analysis Router

Use this skill when the main challenge is not "describe the image" but "pick the right way to read the image first."

Read in order

• Read references/route-matrix.md first.
• Read references/output-contract.md before drafting the final answer.
• Read only the route file(s) you actually need:
• references/method-graphic-design.md
• references/method-photography.md
• references/method-painting-illustration.md
• references/method-interior-design.md
• references/method-architecture-urban.md
• references/method-infographic-diagram.md
• references/method-product-industrial-design.md
• references/method-comics-sequential.md
• references/method-fashion-styling.md
• references/method-sculpture-installation-craft.md
• references/method-game-visual-design.md
• references/method-scientific-medical-imaging.md
• references/method-typography-lettering.md
• references/method-presentation-document.md
• references/method-film-frame.md
• references/method-generic-mixed.md
• references/method-universal-fallback.md

Workflow

1. Gather the user's goal, image count, file names, and any supplied context. 2. If prompt text or file names are available, run:

• py .\scripts\route_image_request.py --prompt "<user request>" --file "<name-or-path>" ...

3. Treat the script output as a routing prior, not final truth. 4. Visually inspect at least one representative image before locking the route. 5. Pick one primary route per image or subgroup. If the batch is mixed, split it instead of forcing one lens onto everything. 6. State route confidence:

• high: the dominant route clearly matches both the visual evidence and the user's goal
• medium: one route leads, but a second route is plausible
• low: the image is hybrid, under-specified, or the file is missing enough context

7. Load only the method file(s) needed for the chosen route. 8. Produce both required artifacts:

• analysis report
• study report

Routing rules

• Prefer graphic-design when typography, layout, branding, CTA, hierarchy, packaging, poster logic, or image-text composition is central.
• Prefer photography when camera capture, timing, lens behavior, exposure, depth, realism, or documentary truth claims are central.
• Prefer painting-illustration when the image reads as painted, drawn, stylized, symbolic, art-historical, or cross-cultural in a way that needs iconography or style analysis.
• Prefer interior-design when the image is fundamentally about space, zoning, furniture, materials, atmosphere, lighting, or circulation, regardless of whether it is a photo or render.
• Prefer architecture-urban when the image is fundamentally about building massing, facade language, site relationship, public realm, skyline, streetscape, landscape integration, or urban spatial order.
• Prefer infographic-diagram when the image is mainly a chart, process diagram, map, flow graphic, system schematic, or information visualization where correctness and readability matter more than visual mood.
• Prefer product-industrial-design when the image is mainly about an object, device, furniture piece, tool, packaging structure, prototype, or product render and the core question is form, usability, manufacturability, or material-finish logic.
• Prefer comics-sequential when the image is a comic page, manga page, strip, webtoon panel set, picture-book spread, or other sequential storytelling image where panel order and narrative pacing matter.
• Prefer fashion-styling when the image is mainly about clothing, silhouette, layering, grooming, accessories, runway/editorial styling, or personal styling rather than camera craft alone.
• Prefer sculpture-installation-craft when the image is mainly about a three-dimensional artwork, object ensemble, craft piece, material assembly, or site-specific installation and the core question is volume, material, presence, or viewing path.
• Prefer game-visual-design when the image is mainly a game screenshot, HUD/UI overlay, character sheet, environment concept tied to gameplay, or scene where readability and play experience matter alongside style.
• Prefer scientific-medical-imaging when the image is mainly diagnostic, scientific, microscopic, radiologic, technical, or evidence-bearing and the core question is interpretability, annotation, or visual validity rather than aesthetic mood.
• Prefer typography-lettering when the image is mainly type-driven: lettering, logotypes, calligraphy, type posters, wordmarks, or text-as-form where letter construction and reading texture matter more than general layout.
• Prefer presentation-document when the image is mainly a slide, report page, proposal page, dashboard report, or document spread where argument flow and page-to-page communication matter more than poster impact.
• Prefer film-frame when the image is a movie still, animation frame, storyboard frame, or a cinematic composition where shot language matters more than poster/layout logic.
• Prefer generic-mixed when the image does not cleanly fit the above routes, or when the batch mixes multiple domains that need separate treatment.
• Prefer universal-fallback when the image does not match any route strongly enough but still deserves a structured visual reading instead of a generic summary.

If one image straddles multiple routes, choose based on what the user is actually trying to learn. Example: a movie poster usually goes to graphic-design; the raw frame inside that poster may go to film-frame only if the user wants cinematic analysis.

If a building image is mainly being judged as a photograph, photography can still win. If the same image is being judged for facade rhythm, site fit, or city effect, prefer architecture-urban.

If a product image is mainly an ad or ecommerce layout, graphic-design can still win. If the core question is form, ergonomics, or detail logic, prefer product-industrial-design.

If none of the named routes fits well, use universal-fallback rather than forcing a bad match. The fallback route should still produce a layered analysis and study report.

Until a dedicated UI product route exists, send standalone interface screenshots to graphic-design and emphasize hierarchy, information density, state clarity, and interaction cues.

Guardrails

• Do not pretend certainty when the route is ambiguous.
• Do not judge paintings with poster or ad-performance standards unless the user explicitly asks for commercial translation.
• Do not treat documentary or journalistic photography as pure aesthetics when ethics, context, or truthfulness are part of the request.
• Do not infer hidden generation workflows for AI art unless visible evidence supports the claim.
• Do not over-read narrative meaning from a single still frame; state the missing motion, sound, or sequence context.
• Do not collapse mixed batches into a single verdict when the images clearly serve different roles.

Batch handling

• For a same-project batch, analyze both individual quality and set-level consistency.
• For before/after batches, isolate what improved, what regressed, and what still blocks the target result.
• For mood boards or reference packs, look for repeated signals instead of over-critiquing one image in isolation.

Recommended response pattern

Follow references/output-contract.md.

At minimum, the final answer should include:

1. route decision and confidence 2. why that route fits 3. the route-specific analysis 4. a concrete study report with drills or next-step practice

Image Analysis Router

中文文档 | English

<div align="center">

🖼️ OpenAI + Claude Skill | Route first, analyze second

  

</div>

🎯 This skill does not treat every image like the same homework. It decides how the image should be read first, then returns a targeted critique and a practical study plan.

✨ Features

🚀 Quick Start

Prerequisites

# 1. Python for the local routing script
py --version

# 2. An AI agent environment that can read local skills
# For example: an environment that can load SKILL.md, references/, scripts/, and image input

# 3. Image input
# Local images, screenshots, filenames, OCR text, or image-heavy requests with context

Usage

🧩 This is a local skill for OpenAI- and Claude-based agent environments. Put the whole folder into your skill directory:

mkdir -p "$CODEX_HOME/skills"
cp -r ./image-analysis-router "$CODEX_HOME/skills/image-analysis-router"

📁 If your agent uses a different skill folder, such as .agents/skills/, replace the target path and keep the same folder structure.

💬 Then call the skill with a real image request, for example:

"Use image-analysis-router to review this poster. Focus on hierarchy and typography."

"Break down this batch of interior renders and tell me what to practice next."

"Figure out which route fits this slide first, then give me a study report."

🔎 If you only have text clues, filenames, or OCR, run the routing script first:

py .\scripts\route_image_request.py --prompt "Analyze this tower facade and check how it meets the street" --file "tower-facade-render.jpg"

🧭 How Routing Works

🛣️ The skill answers one question before anything else: what is the right way to read this image?

Image request
    ↓
[1️⃣ Text and filename pass] ──→ route_image_request.py generates a routing prior
    ↓
[2️⃣ Visual confirmation] ──→ confirm the main route or split the batch
    ↓
[3️⃣ Method selection] ──→ choose 1 of 17 specialized routes
    ↓
[4️⃣ Final output] ──→ Analysis Report + Study Report

Route Confidence

🧠 The skill marks how sure it is about the route before moving on:

🗂️ Route Coverage

🖍️ The skill currently ships with 17 routes for common image-reading jobs.

Design and communication

• graphic-design: posters, brand visuals, packaging, UI screenshots, ad creatives
• infographic-diagram: charts, maps, process diagrams, information graphics
• typography-lettering: type posters, lettering, logotypes, calligraphy, letterform study
• presentation-document: slides, report pages, proposal pages, document spreads

Image and narrative

• photography: documentary, portrait, street, editorial, commercial photography
• film-frame: movie stills, animation frames, storyboard shots, cinematic compositions
• comics-sequential: comic pages, manga, strips, webtoon panels, sequence storytelling
• game-visual-design: game UI, HUD, level screenshots, character panels

Art and space

• painting-illustration: paintings, illustrations, concept art, stylized image work
• interior-design: interior renders, room photos, material and furniture studies
• architecture-urban: facades, street views, public space, urban and site relationships
• sculpture-installation-craft: sculpture, installation, ceramics, craft-based 3D work

Objects and specialist imagery

• product-industrial-design: products, prototypes, object form, packaging structure
• fashion-styling: outfits, silhouettes, accessories, lookbooks, styling visuals
• scientific-medical-imaging: medical scans, microscopy, technical and research imagery

Fallback routes

• generic-mixed: mixed batches that need grouping before critique
• universal-fallback: images that do not fit cleanly anywhere else but still need a structured read

🧠 What You Get

📌 Every run starts with a route decision: which route won, how confident the skill is, and why that route fits better than the alternatives.

📝 The Analysis Report gives a quick read, a route-specific breakdown, and a final judgment about what the image is trying to do and where it actually lands.

📚 The Study Report turns that into next steps: what to learn now, what drills to run, what to watch next time, and what comparisons are worth making.

📁 Project Structure

image-analysis-router/
├── SKILL.md
├── README.md
├── README.zh-CN.md
├── agents/
│   └── openai.yaml
├── scripts/
│   └── route_image_request.py
└── references/
    ├── route-matrix.md
    ├── output-contract.md
    └── method-*.md

⚙️ Configuration

🛠️ The skill does not require a separate config file for the default workflow.

🔤 If you only want a local routing guess before the full image review, you can call the script directly:

py .\scripts\route_image_request.py --prompt "<user goal>" --file "<file name or path>" --hint "<OCR or extra clue>"

📎 Parameters:

• --prompt: the user's goal or question
• --file: image file name or path, repeatable
• --hint: OCR text, title, note, or any extra clue, repeatable

📄 Output

📦 A normal run gives you two core outputs:

🧪 For batches, the skill can also comment on set-level consistency, before/after changes, and whether the images should be split into subgroups first.

⚠️ Ground Rules

🧱 The script output is a starting point, not the final answer.

👀 Important judgments have to go back to visible evidence.

📐 Different image types should not be judged with the same standard.

🤝 When the route is unclear, the skill should say so instead of bluffing.

✅ Supported Environments

💻 This skill works best in AI agent environments that can load local skill folders and accept image input.

🔗 Related Files

📚 Key files in this repo:

• README.zh-CN.md
• SKILL.md
• references/route-matrix.md
• references/output-contract.md
• scripts/route_image_request.py

图像分析路由器

中文文档 | English

<div align="center">

🖼️ OpenAI + Claude Skill | 先选读图方法，再给分析结论

  

</div>

🎯 这个 Skill 不会把所有图片都按同一套模板硬拆一遍。它会先判断这张图该怎么读，再给出对路的分析和能落地的练习建议。

✨ 功能特性

🚀 快速开始

环境要求

# 1. Python（用于本地分流脚本）
py --version

# 2. 一个支持本地 Skill 的 AI 代理环境
# 例如能读取 SKILL.md、references/、scripts/，并接受图片输入的环境

# 3. 图片输入
# 本地图片、截图、文件名、OCR 文本，或带上下文的图像请求都可以

使用方式

🧩 这是一个本地 Skill，适用于 OpenAI 和 Claude 生态里的代理环境。把整个目录放进你的 Skill 目录里：

mkdir -p "$CODEX_HOME/skills"
cp -r ./image-analysis-router "$CODEX_HOME/skills/image-analysis-router"

📁 如果你的代理用的是别的 Skill 目录，比如 .agents/skills/，把目标路径换掉就行，目录结构保持不变。

💬 安装好以后，直接向代理发图像分析请求，比如：

"用 image-analysis-router 分析这张海报，重点看层级和字体。"

"帮我拆一下这组室内效果图，顺便告诉我下一步怎么练。"

"先判断这页 PPT 该走哪条路线，再给我一份学习报告。"

🔎 如果你手里只有文件名、提示词或 OCR 文本，也可以先跑分流脚本：

py .\scripts\route_image_request.py --prompt "分析这张建筑立面图，看看它和街道关系处理得怎么样" --file "tower-facade-render.jpg"

🧭 分流怎么工作

🛣️ 这个 Skill 先回答一个更关键的问题：这张图到底该用什么方法看。

图像请求
    ↓
[1️⃣ 文本与文件名初筛] ──→ route_image_request.py 生成路由 prior
    ↓
[2️⃣ 实际看图复核] ──→ 确认主路线，或先把混合批次拆开
    ↓
[3️⃣ 选择分析方法] ──→ 从 17 条专用路线里选最合适的一条
    ↓
[4️⃣ 最终输出] ──→ Analysis Report + Study Report

路线信心

🧠 在正式展开分析前，Skill 会先标出自己对路线判断有多确定：

🗂️ 路线覆盖

🖍️ 现在内置了 17 条路线，够覆盖大多数常见的读图任务。

设计与传播

• graphic-design：海报、品牌图、包装、UI 截图、广告创意
• infographic-diagram：图表、地图、流程图、信息图
• typography-lettering：字体海报、字标、书法、字形研究
• presentation-document：PPT 页面、报告页、提案页、文档页

影像与叙事

• photography：纪实、人像、街拍、商业摄影
• film-frame：电影截图、动画帧、分镜、剧照式画面
• comics-sequential：漫画页、条漫、分镜页、顺序叙事
• game-visual-design：游戏界面、HUD、关卡截图、角色面板

艺术与空间

• painting-illustration：绘画、插画、概念图、风格化图像
• interior-design：室内效果图、空间照片、材质和家具研究
• architecture-urban：建筑立面、街景、公共空间、场地关系
• sculpture-installation-craft：雕塑、装置、陶艺、工艺类立体作品

对象与专业图像

• product-industrial-design：产品外观、原型、物件形态、包装结构
• fashion-styling：穿搭、造型、配饰、lookbook、时尚视觉
• scientific-medical-imaging：医学图像、显微图、科研图、技术图像

兜底路线

• generic-mixed：图像组太杂，得先分组再分析
• universal-fallback：没有明显归类，但仍然值得结构化阅读

🧠 你会拿到什么

📌 每次分析都会先告诉你：最后选了哪条路线、把握有多大、为什么这条比别的路线更合适。

📝 Analysis Report 会给你一个很快能抓住重点的判断，再展开按路线细拆，最后落到整体评价：这张图想做什么，它到底做到了多少。

📚 Study Report 会把结论翻成下一步动作，包括现在该补什么、适合练什么、下次最容易犯什么错，以及必要时该对照哪些参考。

📁 项目结构

image-analysis-router/
├── SKILL.md
├── README.md
├── README.zh-CN.md
├── agents/
│   └── openai.yaml
├── scripts/
│   └── route_image_request.py
└── references/
    ├── route-matrix.md
    ├── output-contract.md
    └── method-*.md

⚙️ 配置说明

🛠️ 默认流程不需要额外配置文件，拿来就能用。

🔤 如果你只是想先做一次本地分流预判，也可以直接跑脚本：

py .\scripts\route_image_request.py --prompt "<用户目标>" --file "<文件名或路径>" --hint "<OCR 或补充线索>"

📎 参数说明：

• --prompt：用户问题或目标
• --file：图片文件名或路径，可重复传入
• --hint：OCR 文本、标题、备注等补充线索，可重复传入

📄 输出内容

📦 正常一次分析，至少会有两份核心结果：

🧪 如果是成组图片，它还会补充整组一致性、前后变化，以及要不要先拆成几个子组。

⚠️ 使用原则

🧱 脚本结果只是起点，不是最后答案。

👀 重要判断必须回到能看见的证据。

📐 不同类型的图，不能用同一套标准硬套。

🤝 路线拿不准时，就老实说不确定，不硬编。

✅ 支持环境

💻 这个 Skill 最适合用在能读取本地 Skill 目录、也能接收图片输入的 AI 代理环境里。

🔗 相关文件

📚 这个仓库里最关键的文件在这里：

• README.md
• SKILL.md
• references/route-matrix.md
• references/output-contract.md
• scripts/route_image_request.py

Architecture And Urban Method

Use this route for building exteriors, facades, campus views, landscape-architecture views, streetscapes, plazas, urban blocks, masterplan visuals, and other images where built form and public spatial order matter more than interior furnishing.

This route should read like a built-environment critique: what the object or place is, how it meets the site, how it organizes form, and what kind of experience it creates at street and city scale.

Evidence ladder

Keep claims separated by certainty:

• FACT: directly visible massing, facade elements, openings, materials, landscape, people, and street conditions
• INFERENCE: strong read about program, circulation, climate response, or urban intent backed by visible evidence
• HYPOTHESIS: weaker reads about unseen plan logic, access routes, structure, or performance that need caveats

Do not confuse a dramatic angle with proof that the project works in reality.

Core lenses

1. Boundary and type 2. Site and context slicing 3. Massing and silhouette 4. Facade and opening logic 5. Ground plane, access, and public realm 6. Material, light, and environmental response 7. Human experience and urban effect 8. Cross-view consistency when the batch supports it

Analysis order

1. Boundary and type

• Name the likely object: tower, low-rise, campus building, pavilion, cultural building, commercial block, housing, public space, landscape intervention, and so on.
• State the likely scale and viewpoint: street view, bird's-eye, oblique render, elevation-like crop, nighttime hero shot.
• Note what is missing or uncertain:
• no plan or section
• no surrounding streets shown
• only one facade visible
• render only
• no human scale cues

2. Site and context slicing

• Break the image into:
• primary built object
• ground plane and access zone
• adjacent buildings or landscape
• skyline or background context
• Explain whether the project feels embedded in its site or dropped onto it.
• Identify visible context clues:
• setback
• street edge
• corner condition
• terrain shift
• water edge
• tree canopy
• plaza or forecourt

3. Massing and silhouette

• Read overall form: monolithic, fragmented, terraced, layered, bridged, carved, iconic, background-neutral, and so on.
• Judge proportion, rhythm, void-to-solid balance, and skyline behavior.
• State whether the massing feels coherent from the shown angle or only graphically striking.
• If the image implies a big move, say what that move is actually doing.

4. Facade and opening logic

• Inspect window rhythm, panel logic, structural bays, shading depth, balcony language, and articulation.
• Ask whether the facade expresses:
• order
• depth
• climate response
• institutional identity
• luxury signal
• anonymity
• Distinguish genuine facade intelligence from surface patterning.

5. Ground plane, access, and public realm

• Evaluate entries, thresholds, canopies, stairs, ramps, corners, storefronts, podium behavior, and public-facing edges.
• Ask whether someone approaching the building would understand how to enter and inhabit the place.
• For urban scenes, inspect block permeability, frontage continuity, pedestrian scale, and whether the space invites staying or only passing through.

6. Material, light, and environmental response

• Name primary and secondary material families.
• Judge tactile logic, weathering plausibility, transparency/opacity balance, and how the project handles light.
• If visible, comment on shading devices, overhangs, fins, screens, vegetation, and likely climate response.
• For renders, say when the atmosphere is persuasive but the material realism or environmental logic is weak.

7. Human experience and urban effect

• Evaluate scale feel, dignity, warmth, legibility, street presence, civic generosity, and whether the project supports everyday use.
• Distinguish:
• iconic object quality
• neighborhood fit
• public-space quality
• image-first spectacle
• If people are present, use them to judge scale and comfort rather than as decoration alone.

8. Cross-view synthesis

When the user gives multiple views or a project set, add:

• whether the core formal idea survives across views
• whether the ground condition matches the hero image promise
• whether facade, massing, and site logic agree with each other

Do not infer full urban performance from one cropped exterior image.

9. Overall judgment

• Explain whether the work wins through massing, facade intelligence, site fit, public-space quality, atmosphere, or some mix.
• Name the biggest strength and the real limit.
• Distinguish "photogenic architecture" from "convincing built environment."

Common failure patterns

• strong hero silhouette with weak ground condition
• decorative facade pattern without deeper order
• oversized object that ignores pedestrian scale
• render glow hiding unresolved entry, shade, or material logic
• public realm treated as leftover space

Study report emphasis

• recommend one massing-comparison drill
• recommend one facade rhythm or opening-study drill
• recommend one site-and-ground-plane walkthrough exercise
• recommend one climate or material-response study

Comics And Sequential Method

Use this route for comic pages, manga pages, webtoon episodes, strips, picture-book spreads, motion-comic boards, and other sequential visual storytelling pages where panel order and pacing matter.

This route should read like a sequence breakdown: what happens in each beat, how the page moves the eye, and whether image, panel order, and text are working together.

Evidence ladder

Keep claims separated by certainty:

• FACT: directly visible panels, gutters, balloons, captions, sound effects, poses, backgrounds, and page structure
• INFERENCE: strong read about pacing, emotional emphasis, point of view, or intended reading order backed by visible page design
• HYPOTHESIS: weaker story or tonal reads that depend on missing previous/next pages and need caveats

Do not judge a comic page as a single illustration when the main job is sequential storytelling.

Core lenses

1. Boundary and sequence type 2. Page or strip slicing 3. Reading order and panel logic 4. Staging, acting, and environment 5. Text-image coordination 6. Pacing, emphasis, and payoff 7. Cross-page continuity when the batch supports it

Analysis order

1. Boundary and sequence type

• Name the format: single comic page, spread, webtoon segment, newspaper strip, storyboard-like sequence, picture-book spread.
• State the likely reading direction if visible.
• Note what is missing or uncertain:
• previous/next page missing
• dialogue cropped
• translated text absent
• only one panel from a sequence

2. Page or strip slicing

• Break the page into panels or beats.
• Identify:
• establishing panel
• reaction panel
• action panel
• reveal panel
• quiet transition
• payoff beat
• If the page is dense, group panels into mini-sequences first.

3. Reading order and panel logic

• Check whether the eye path is obvious.
• Inspect:
• panel size hierarchy
• gutter rhythm
• overlap
• inset logic
• bleed or border breaks
• cliffhanger placement
• Ask whether the page supports smooth reading or creates accidental confusion.

4. Staging, acting, and environment

• Judge pose clarity, body direction, gaze, silhouette, background support, and how well the page orients the reader in space.
• For dialogue-heavy pages, ask whether acting and composition keep the scene alive.
• For action pages, ask whether movement direction and impact are legible.

5. Text-image coordination

• Inspect balloons, captions, sound effects, label placement, text density, and whether text blocks choke the image.
• Ask whether dialogue order matches panel flow.
• Distinguish expressive lettering from noisy lettering.

6. Pacing, emphasis, and payoff

• Explain how the page controls time:
• compression
• pause
• repetition
• acceleration
• reveal
• State which panel carries the emotional or narrative peak.
• Distinguish a beautiful page from a page that lands its beat.

7. Cross-page synthesis

When the user gives multiple pages or a chapter segment, add:

• continuity of reading rhythm
• consistency of acting and spatial orientation
• escalation or release across pages
• whether splash moments are earned or overused

Do not infer a whole chapter arc from one isolated page.

8. Overall judgment

• Explain whether the sequence wins through clarity, pace, acting, layout, atmosphere, humor, tension, or some mix.
• Name the biggest strength and the real blocker.
• Distinguish "good drawing" from "good sequential storytelling."

Common failure patterns

• pretty panels with weak page rhythm
• confusing reading order
• balloon placement fighting panel flow
• action energy without spatial clarity
• dramatic splash panel unsupported by surrounding beats

Study report emphasis

• recommend one panel-order redraw drill
• recommend one balloon-placement or text-density drill
• recommend one acting-and-silhouette drill
• if multiple pages exist, add one pacing-comparison drill

Fashion And Styling Method

Use this route for fashion editorials, runway looks, lookbooks, styling grids, outfit photos, beauty-fashion hybrids, and other images where clothing, silhouette, grooming, and persona signal matter more than camera craft alone.

This route should read like a styling critique: what the look is saying, how the outfit is built, and whether the styling choices hold together on a body and in a context.

Evidence ladder

Keep claims separated by certainty:

• FACT: directly visible garments, layers, accessories, grooming choices, fit cues, and body proportions
• INFERENCE: strong read about taste level, subculture, brand tier, or intended persona backed by visible evidence
• HYPOTHESIS: weaker assumptions about season, styling brief, audience, or hidden garment construction that need caveats

Do not let striking photography hide weak styling logic.

Core lenses

1. Boundary and look type 2. Silhouette and proportion 3. Garment, textile, and accessory read 4. Styling system and layering logic 5. Body, grooming, and persona signal 6. Context, wearability, and image-world fit 7. Cross-look consistency when the batch supports it

Analysis order

1. Boundary and look type

• Name the likely context: runway, editorial, ecommerce, street style, personal styling, beauty campaign, lookbook, celebrity styling, costume-like fashion image.
• State the likely job:
• sell garments
• build persona
• signal trend literacy
• create editorial atmosphere
• show styling range
• Note what is missing or uncertain:
• crop hides full silhouette
• shoes or accessories absent
• motion blur hides fabric behavior
• only one look from a full collection

2. Silhouette and proportion

• Read overall shape:
• fitted
• oversized
• columnar
• A-line
• layered-volume
• sculptural
• deconstructed
• Judge balance between upper and lower body, vertical breaks, waist logic, and whether the silhouette feels intentional from head to toe.
• Distinguish dramatic proportion from accidental imbalance.

3. Garment, textile, and accessory read

• Inventory the main garments and accessory roles.
• Inspect:
• fabric weight
• drape
• sheen
• texture contrast
• hardware
• trim
• print or pattern use
• Ask whether accessories complete the idea, overstate it, or drag it off course.

4. Styling system and layering logic

• Explain how pieces are combined:
• repetition
• contrast
• tonal layering
• subversion
• uniform logic
• high-low mixing
• Judge whether the styling has one coherent thesis or too many competing ideas.
• If the look relies on one hero move, say what that move is and whether the rest supports it.

5. Body, grooming, and persona signal

• Read how hair, makeup, posture, casting, and gesture interact with the clothes.
• Ask what persona the look projects:
• severe
• romantic
• utilitarian
• luxury
• club
• avant-garde
• soft power
• anti-fashion
• Keep body-fit claims proportional to what the image actually shows.

6. Context, wearability, and image-world fit

• Ask whether the look makes sense for its implied setting and audience.
• Distinguish:
• editorial success
• real-life wearability
• brand signal strength
• trend mimicry
• For commercial images, judge whether the styling helps the product read clearly.

7. Cross-look synthesis

When the user gives multiple looks, add:

• silhouette consistency
• recurring accessory logic
• color and material policy
• how much range exists without losing identity

Do not invent a full styling language from one cropped outfit shot.

8. Overall judgment

• Explain whether the look wins through silhouette, fabric tension, styling intelligence, persona clarity, or some mix.
• Name the strongest move and the limiting factor.
• Distinguish "photographs well" from "styles well."

Common failure patterns

• interesting garment pieces with no full-look logic
• accessories and grooming fighting the clothes
• oversized or layered styling that collapses body proportion
• trend-coded styling without a clear persona
• editorial drama covering weak wearability or poor product read

Study report emphasis

• recommend one silhouette-copy drill
• recommend one fabric-and-accessory pairing drill
• recommend one persona-board exercise
• if it is a look set, add one consistency-vs-range exercise

Film Frame Method

Use this route for movie stills, animation frames, storyboard panels, and sequence screenshots where shot language matters more than marketing layout.

This route should read like a frame study: what the shot is doing, how the frame is built, and what dramatic pressure it carries.

Limits first

State early what a still frame cannot prove:

• editing rhythm
• sound design
• full scene arc
• performance timing across the shot

If multiple frames are provided, you can discuss continuity, coverage, and implied rhythm with more confidence.

Evidence ladder

Keep claims separated by certainty:

• FACT: directly visible shot size, blocking, palette, props, and frame structure
• INFERENCE: strong read about lens feel, genre mood, dramatic intent, or scene function backed by visible evidence
• HYPOTHESIS: weaker plot or sequence-level reading that needs caveats

Do not turn one evocative frame into a full plot summary.

Core lenses

1. Boundary and source type 2. Frame slicing 3. Camera and shot reverse 4. Blocking and composition 5. Light, color, and production texture 6. Narrative pressure and genre signal 7. Sequence or set-level continuity when the batch supports it

Analysis order

1. Boundary and source type

• Name the likely source type: live-action still, animation frame, storyboard image, previz-like frame, or unknown.
• Note what is missing or uncertain:
• subtitle overlay
• promo still rather than narrative frame
• crop only
• no surrounding frames
• no sound

2. Frame slicing

• Break the frame into foreground, middle, background, and off-screen pressure when useful.
• Inventory the major bodies, props, eyelines, practicals, and empty spaces.
• State what the viewer is pushed to notice first.
• If a prop or gesture carries most of the tension, call it out early.

3. Camera and shot reverse

• Identify shot scale, angle, height, and lens feel.
• Comment on distance, perspective compression or stretch, and whether the camera feels intimate, neutral, surveillant, monumental, and so on.
• If you make a reverse read, keep it proportional to the evidence.

4. Blocking and composition

• Explain how bodies, props, architecture, and negative space are arranged.
• Judge whether the frame creates:
• clarity
• tension
• distance
• imbalance
• spectacle
• intimacy
• For animation and storyboard frames, also inspect silhouette clarity and staging legibility.

5. Light, color, and production texture

• Read contrast, palette, practical light sources, and surface texture.
• Distinguish naturalistic, theatrical, symbolic, stylized, grimy, glossy, dreamlike, and other lighting modes when useful.
• Note whether the frame depends on costume, set texture, smoke, rain, haze, or depth layering to work.

6. Narrative pressure and genre signal

• Infer the frame's likely function:
• reveal
• threat
• intimacy
• transition
• isolation
• spectacle
• aftermath
• setup
• Keep inference proportional to the still.
• If genre signals are strong, say how the frame uses them instead of merely naming them.

7. Sequence or set-level synthesis

When the user gives multiple frames, add:

• continuity of blocking
• continuity of light and palette
• escalation or release of dramatic pressure
• whether the sequence feels covered intentionally or repetitively

Do not fake sequence logic from a single frame.

8. Overall judgment

• Explain whether the frame works because of clarity, tension, emotional distance, immersion, spectacle, stylization, or some mix.
• Name the strongest move and the limiting factor.
• Distinguish "pretty frame" from "dramatically useful frame."

Common failure patterns

• over-reading plot from one frame
• describing aesthetics without saying what they do dramatically
• calling a marketing still cinematic just because it is wide and moody
• reading production polish as dramatic strength when blocking is weak

Study report emphasis

• recommend one frame-grab comparison exercise
• recommend one blocking or silhouette decomposition drill
• recommend one lighting-and-palette study
• recommend one sequence-level follow-up if more frames can be collected

Game Visual Design Method

Use this route for gameplay screenshots, HUD and menu screens, character sheets, inventory layouts, level views, ability UIs, key art tightly tied to gameplay, and other game visuals where play readability matters alongside style.

This route should read like a play-facing critique: what the player needs to understand, how the world and interface cooperate, and whether the image supports actual play instead of only looking good.

Evidence ladder

Keep claims separated by certainty:

• FACT: directly visible HUD, map, prompts, avatar, enemies, environment, icons, health bars, menus, and objective cues
• INFERENCE: strong read about game genre, play state, challenge level, or intended attention path backed by visible evidence
• HYPOTHESIS: weaker assumptions about full mechanics, progression, or narrative context that need caveats

Do not judge a game screenshot like a film still if play readability is part of the job.

Core lenses

1. Boundary and gameplay context 2. World/UI layer slicing 3. Readability and navigation 4. Encounter or task clarity 5. Art direction, mood, and diegesis 6. Feedback, state, and player risk 7. Cross-screen synthesis when the batch supports it

Analysis order

1. Boundary and gameplay context

• Name the likely image type: live gameplay, HUD screen, menu, inventory, map, dialogue screen, key art tied to play, level overview.
• State the likely genre if visible: shooter, RPG, strategy, survival, action adventure, puzzle, gacha, sim, and so on.
• Note what is missing or uncertain:
• no motion
• no player input context
• cropped HUD
• no prior state

2. World/UI layer slicing

• Break the image into:
• core play space
• player avatar or controlled unit
• enemies or hazards
• navigation landmarks
• HUD layer
• prompts or objective layer
• State what the player is expected to look at first.

3. Readability and navigation

• Ask whether the player can quickly identify:
• where they are
• what matters
• where to go
• what is dangerous
• what is interactive
• Judge color coding, icon hierarchy, contrast, and landmark clarity.
• Distinguish atmospheric darkness from unreadable darkness.

4. Encounter or task clarity

• Explain what task or tension the screen appears to set up.
• Judge whether enemies, objectives, loot, resources, cooldowns, or states are legible enough to support decision-making.
• For menus and loadouts, ask whether tradeoffs and selections are understandable at a glance.

5. Art direction, mood, and diegesis

• Read palette, world texture, character readability, environmental storytelling, and whether the interface feels diegetic, overlaid, or split awkwardly.
• Distinguish "beautiful frame" from "usable frame."

6. Feedback, state, and player risk

• Inspect whether the player can see current status, danger, reward, and response options.
• Ask where the player is most likely to hesitate, misread, or miss an important cue.
• For competitive or high-speed play, weight clarity over flourish.

7. Cross-screen synthesis

When the user gives multiple screens, add:

• consistency of icon meaning
• consistency of color semantics
• continuity between world cues and UI cues
• whether the game keeps its visual contract across combat, exploration, and menus

Do not infer full UX quality from one polished screenshot.

8. Overall judgment

• Explain whether the image wins through atmosphere, clarity, diegesis, encounter readability, interface discipline, or some mix.
• Name the strongest move and the blocking weakness.
• Distinguish "good art direction" from "good play-facing communication."

Common failure patterns

• cinematic atmosphere masking poor interaction readability
• overloaded HUD competing with world cues
• stylish icons with weak semantic clarity
• environment beauty with poor landmark differentiation
• menus that look premium but slow down decision-making

Study report emphasis

• recommend one HUD reduction or clarity drill
• recommend one landmark-and-navigation drill
• recommend one state-feedback comparison drill
• if multiple screens exist, add one consistency-of-visual-language exercise

Generic Or Mixed Method

Use this route when the image is ambiguous or the batch spans multiple domains.

This route is for triage first, critique second.

Evidence ladder

• FACT: what the image plainly contains
• INFERENCE: strongest route guess backed by visible evidence
• HYPOTHESIS: weaker route guesses or context assumptions that still need confirmation

If certainty is low, say that up front instead of pretending the batch is cleaner than it is.

Core job

1. cluster 2. route 3. only then critique

Analysis order

1. Cluster the inputs

• Split by dominant function:
• communication
• camera capture
• artwork
• space
• cinematic still
• other
• If only one image exists, name the top route and the runner-up route.
• If the batch is mixed, group it before giving judgments.

2. Do the universal four-step read

1. Describe: what is plainly visible 2. Structure: how the image organizes attention 3. Meaning: what it seems to communicate or evoke 4. Judgment: what works, what fails, what route would unlock deeper critique

3. Collapse uncertainty

• Explain what evidence made you choose the current route.
• Explain what evidence kept the runner-up route alive.
• Name the missing context that would make the call cleaner:
• original file type
• more frames
• wider crop
• project context
• user goal

4. Escalate if needed

• If the user needs depth, recommend the best next route and switch into that route's method.
• If the ambiguity itself matters, stop after triage plus a short first-pass critique.

Study report emphasis

• tell the user which category to study first and why
• give one sorting or comparison exercise
• say what evidence to collect next time to enable a sharper critique
• if the batch is mixed, recommend how to regroup it before deeper analysis

Graphic Design Method

Use this route for posters, brand graphics, packaging, layout-heavy images, ad creatives, key visuals, social cards, and UI screenshots.

This route should read like a communication teardown, not a vague style opinion.

Evidence ladder

Keep claims separated by certainty:

• FACT: directly visible structure, copy, spacing, color, and assets
• INFERENCE: strong read about audience, task, brand tone, or intended reading order backed by visible evidence
• HYPOTHESIS: weaker assumptions about campaign strategy, funnel stage, or missing states that need a caveat

Do not state audience or business intent as certainty if the screen or asset does not actually prove it.

Core lenses

1. Boundary and task 2. Message slicing 3. Hierarchy and attention path 4. Structure and grid discipline 5. Typography 6. Color, asset treatment, and brand fit 7. Persuasion, usability, or conversion logic 8. Set-level consistency when the batch supports it

Analysis order

1. Boundary and task

• Name the format: poster, landing hero, packaging face, carousel card, app screen, dashboard view, banner, and so on.
• State the likely job of the image: inform, persuade, signal brand, guide action, or package information.
• Note any hard limits:
• crop only
• missing surrounding screens
• low resolution
• no interaction states
• one frame from a larger system

2. Message slicing

• Inventory every communication unit:
• headline
• support copy
• CTA
• logo or brand mark
• product or hero image
• icons, labels, prices, badges, navigation, tabs, metadata
• Group them into primary, secondary, and tertiary information.
• State what the design is asking the viewer to notice first, second, and third.
• For UI, separate content, chrome, status, and actions.

3. Hierarchy and attention path

• Check whether the reading path is obvious and stable.
• Look at:
• scale contrast
• weight contrast
• placement
• grouping
• white space
• focal tension
• directional cues from images or arrows
• Name the strongest hierarchy move and the main conflict.
• If the design is intentionally noisy, say whether the noise feels controlled or accidental.

4. Structure and grid discipline

• Read alignment, margins, spacing rhythm, columns, module logic, and balance.
• Use grouping language only when it helps:
• proximity
• similarity
• continuity
• closure
• figure/ground
• For packaging, inspect panel logic and how information wraps or stacks.
• For UI, inspect density, alignment consistency, component hierarchy, and whether controls feel system-based or one-off.

5. Typography

• Judge font pairing, scale ladder, weight distribution, line length, line spacing, and legibility at likely viewing size.
• Check whether typography carries the tone cleanly or is doing too many jobs at once.
• For interface work, inspect label clarity, placeholder overuse, button copy strength, and overflow risk if visible.

6. Color, asset treatment, and brand fit

• Name primary, secondary, and accent colors.
• Judge contrast, palette discipline, emotional tone, and whether the color decisions support the task.
• Inspect how images, icons, cutouts, gradients, shadows, and textures are treated:
• premium
• loud
• editorial
• generic
• over-processed
• Distinguish intentional tension from accidental clash.
• If brand identity is visible, say whether the execution feels on-brand, off-brand, or generic.

7. Persuasion, usability, or conversion logic

• For ads, posters, and packaging:
• ask whether the design signals value fast enough
• check if the CTA or selling point is clear
• inspect whether decorative moves hurt persuasion
• For UI:
• ask whether the screen tells the user what to do next
• inspect action priority, state clarity, error risk, and whether the most important control is visually obvious
• Distinguish "visually stylish" from "functionally convincing."

8. Cross-image or system synthesis

When the user gives a campaign set, component family, or before/after versions, add a set-level read:

• what stays consistent
• what drifts
• whether hierarchy and tone survive across formats
• whether the system feels designed or merely similar

Do not fake system logic from one isolated image.

9. Overall judgment

• Explain whether the design wins through clarity, tone, memorability, conversion focus, system discipline, or some mix.
• Name the biggest strength and the limiting factor.
• Distinguish "looks designed" from "communicates well."

Common failure patterns

• weak or unstable first focal point
• too many competing accents
• spacing that feels accidental instead of systemic
• typography that performs as decoration but not as communication
• brand cues that are too generic to carry identity
• polished surface hiding weak task clarity
• UI screens that look clean but do not guide action

Study report emphasis

• recommend one hierarchy drill
• recommend one spacing or grid cleanup drill
• recommend one typography simplification drill
• recommend one color or brand-consistency drill
• if it is a UI set or campaign set, add one system-consistency drill

Infographic And Diagram Method

Use this route for charts, graphs, maps, dashboards used as information displays, flowcharts, process diagrams, system schematics, comparison tables, and other information visuals where correctness and readability matter more than mood.

This route should read like an information audit: what the visual is trying to explain, whether the structure is truthful, and where readers are likely to get lost.

Evidence ladder

Keep claims separated by certainty:

• FACT: directly visible labels, chart forms, icons, arrows, legends, axes, ordering, and grouping
• INFERENCE: strong read about audience, intended message, or reading sequence backed by visible structure
• HYPOTHESIS: weaker assumptions about hidden dataset quality, stakeholder intent, or omitted context that need caveats

Do not judge a data graphic like a poster. A beautiful chart that misleads is still weak.

Core lenses

1. Boundary and information task 2. Message and claim slicing 3. Structural readability 4. Data or logic integrity cues 5. Labeling, notation, and annotation 6. Visual encoding and emphasis 7. User interpretation risk 8. Cross-page or system consistency when the batch supports it

Analysis order

1. Boundary and information task

• Name the format: bar chart, line chart, map, process flow, decision tree, system diagram, comparison table, infographic panel, KPI board, and so on.
• State the likely job:
• explain a process
• compare values
• summarize a system
• persuade with evidence
• orient a reader
• Note what is missing or uncertain:
• source data absent
• axes cropped
• legend missing
• page is only one panel from a series

2. Message and claim slicing

• Identify the main claim or takeaway.
• Break the visual into message units:
• title or headline
• chart body
• annotation
• legend
• source note
• icon layer
• callout or conclusion
• State what the reader is likely to understand first and what gets delayed or buried.

3. Structural readability

• Inspect reading order, grouping, alignment, spacing, scan path, and cognitive load.
• Ask whether the structure helps the reader understand step by step or forces backtracking.
• For flows and diagrams, inspect arrow logic, node ordering, branch clarity, and loop readability.
• For maps, inspect orientation, scale cues, labeling density, and route clarity.

4. Data or logic integrity cues

• Look for visible signs of distortion or ambiguity:
• truncated axes
• misleading scale jumps
• inconsistent units
• decorative shapes distorting quantity
• ambiguous arrow direction
• category mixing
• If actual correctness cannot be verified from the image alone, say that and limit the claim to visible integrity cues.

5. Labeling, notation, and annotation

• Judge title precision, label clarity, legend quality, unit visibility, and annotation usefulness.
• Ask whether the reader can decode the system without external narration.
• Distinguish concise labeling from under-explained labeling.

6. Visual encoding and emphasis

• Explain how color, size, position, icon style, and emphasis are used to show meaning.
• Judge whether the encoding supports the claim or adds noise.
• Distinguish persuasive emphasis from manipulative emphasis.
• If the visual uses illustration or branding heavily, ask whether it helps memory or blocks understanding.

7. User interpretation risk

• State where a reader is most likely to misread, stall, or draw the wrong conclusion.
• Name the most important fix if the visual is intended for public or high-stakes communication.
• Distinguish "pretty" from "trustworthy and usable."

8. Cross-page or system synthesis

When the user gives a deck, report, or diagram family, add:

• consistency of notation
• consistency of color meaning
• consistency of scale or legend treatment
• whether pages feel like one information system or separate styles

Do not invent data quality claims from styling alone.

9. Overall judgment

• Explain whether the visual wins through clarity, trust, compression of complexity, memorability, or some mix.
• Name the biggest strength and the real blocker.
• Distinguish "looks smart" from "helps readers think correctly."

Common failure patterns

• headline claims outrunning what the visual actually proves
• overloaded annotation that hides the main pattern
• decorative illustration weakening quantitative reading
• inconsistent notation across one visual or across pages
• process diagrams with arrows that do not establish clear direction

Study report emphasis

• recommend one redraw or simplification drill
• recommend one label-and-annotation clarity drill
• recommend one truthfulness or scale-check drill
• if it is a multi-page set, add one notation-consistency drill

Interior Design Method

Use this route for room photos, interior renders, fit-out proposals, mood boards anchored on spaces, and other images where spatial experience is central.

This route should read like a space critique: what the room is, how it works, how it feels, and where the image may be flattering the design.

Evidence ladder

Keep claims separated by certainty:

• FACT: directly visible layout, objects, materials, light, and circulation clues
• INFERENCE: strong read about user type, program, atmosphere, or unresolved function backed by visible evidence
• HYPOTHESIS: weaker assumptions about hidden plan logic or unseen adjacent spaces that need a caveat

Do not confuse a beautiful rendering angle with proof that the space works well.

Core lenses

1. Boundary and program 2. Space slicing and zoning 3. Circulation and planning 4. Envelope, form, and volume 5. Lighting reverse 6. Material, color, and furniture logic 7. Usability, atmosphere, and lived reality 8. Cross-image consistency when the batch supports it

Analysis order

1. Boundary and program

• Name the likely room type and primary user.
• State the probable use case: living, dining, hospitality, retail, office, bedroom, studio, and so on.
• Note what is missing or uncertain:
• single render angle
• no plan view
• no section or lighting plan
• staging-only photo
• no scale reference

2. Space slicing and zoning

• Break the image into spatial zones:
• arrival zone
• primary activity zone
• support zone
• background or depth zone
• State whether the room reads clearly or feels spatially muddy.
• Identify the anchor point: sofa wall, dining core, bed, island, desk, fireplace, window, feature wall, and so on.

3. Circulation and planning

• Evaluate movement path, furniture spacing, access clearances, and how someone would actually use the room.
• Flag:
• blocked flows
• dead corners
• awkward furniture relationships
• focal conflict
• a missing support surface, storage, or transition zone
• Distinguish "photogenic in one angle" from "convincing in repeated use."

4. Envelope, form, and volume

• Inspect walls, ceiling, openings, vertical lines, built-ins, and overall massing.
• Judge proportion, scale layering, and whether the room has enough compression and release.
• Read whether lines and volumes feel calm, dramatic, heavy, light, cluttered, or unresolved.

5. Lighting reverse

• Separate daylight, ambient, task, accent, decorative, and practical light when visible.
• Judge direction, softness, contrast, and whether the lighting supports the room's intended function.
• For renders, say when the image sells glow and atmosphere but hides what the night scene would actually feel like.
• For real photos, note if the lighting is underpowered, too flat, too contrasty, or mismatched in color temperature.

6. Material, color, and furniture logic

• Name primary, secondary, and accent materials.
• Judge palette restraint, tactile contrast, finish consistency, and realism.
• Evaluate whether furniture scale, weight, and style are coherent with the shell.
• Distinguish "styled surface richness" from a real material logic that could survive use, wear, and maintenance.

7. Usability, atmosphere, and lived reality

• Ask whether the room supports comfort, storage, maintenance, privacy, acoustics, and the behavior it claims to support.
• Evaluate atmosphere:
• warm
• cool
• ceremonial
• relaxed
• hospitality-driven
• sterile
• over-staged
• If the user asks for critique, separate "beautiful in image" from "workable in life."

8. Cross-image synthesis

When the user gives multiple views or a same-project set, add a set-level read:

• zoning consistency
• material consistency
• lighting consistency
• whether focal points agree across rooms
• whether the project has a clear spatial voice or only isolated pretty angles

Do not invent a whole project logic from one hero shot.

9. Overall judgment

• Explain whether the space wins through planning, atmosphere, restraint, richness, practicality, or some mix.
• Name the biggest strength and the real blocker.
• Distinguish "styled well" from "designed well."

Common failure patterns

• nice styling but weak circulation
• one attractive focal wall with no total-room hierarchy
• lighting that looks dramatic in render but fails real use
• material palette that lacks tactile or maintenance logic
• furniture scale mismatches
• atmospheric image masking unresolved storage, task, or movement problems

Study report emphasis

• recommend one zoning or furniture-planning drill
• recommend one lighting study
• recommend one material-combination or finish-discipline drill
• recommend one lived-use walkthrough exercise
• if it is a project set, add one consistency drill across rooms

Painting And Illustration Method

Use this route for paintings, drawings, concept art, editorial illustration, AI art, cross-cultural visual works, and stylized image-making where medium logic matters more than camera capture.

This route should read like an artwork breakdown: first what is truly there, then how it is built, then what it may mean.

Evidence ladder

Keep claims separated by certainty:

• FACT: directly visible forms, marks, palette, composition, symbols, and material clues
• INFERENCE: strong reading about style family, iconography, process, or cultural cues backed by visible evidence
• HYPOTHESIS: weaker symbolic, cultural, or process reads that need alternatives or caveats

Do not turn symbolic interpretation into fact just because it sounds intelligent.

Choose the smallest useful lens set

Do not force every framework into every image. Pick only the lenses that help:

• Feldman: description -> analysis -> interpretation -> judgment
• Panofsky: use when symbolism or iconography is central
• Wolfflin: use when formal style comparison matters
• Xie He / East Asian lenses: use when brush logic, spirit, void, or literati aesthetics matter

Core lenses

1. Boundary and medium read 2. Descriptive inventory 3. Formal language 4. Spatial and compositional logic 5. Symbol, narrative, and cultural reading 6. Process or medium control 7. Cross-image style habits when the batch supports it

Analysis order

1. Boundary and medium read

• Name the likely image type: painting, drawing, print-like illustration, concept art, collage-like work, AI art, mixed digital piece, or unknown.
• Note what is missing or uncertain:
• crop only
• low resolution
• reproduction photograph instead of direct artwork file
• no scale context
• For AI art or digitally composited work, describe visible outcomes instead of pretending you know the hidden workflow.

2. Descriptive inventory

• Inventory figures, motifs, setting cues, palette blocks, mark types, edges, textures, and focal zones.
• Break the image into useful zones:
• foreground
• middle
• background
• symbolic accent zones
• void or negative-space zones
• Keep this phase factual before interpreting.

3. Formal language

• Look at line, mass, color, contrast, rhythm, edge behavior, depth logic, and mark-making.
• If the image benefits from comparison, use a small number of formal pairs such as:
• linear vs painterly
• planar vs recessional
• closed vs open
• multiplicity vs unity
• For East Asian or brush-led work, inspect brush energy, void handling, breath, and whether the image depends on controlled restraint or dense accumulation.

4. Spatial and compositional logic

• Explain how the image organizes attention.
• Read:
• focal hierarchy
• balance or imbalance
• figure/ground relation
• movement path
• cropping pressure
• pattern and repetition
• If the piece is highly polished but structurally weak, say so directly.

5. Symbol, narrative, and cultural reading

• Use iconography only when the image contains genuine evidence for it.
• Give one main reading, then alternatives if the image supports them.
• For cross-cultural work, name the lens carefully and avoid flattening unlike traditions into one blended cliché.
• Distinguish:
• visible cultural markers
• plausible lineage
• speculative meaning

6. Process or medium control

• Ask how the image seems to be made:
• brush-led
• ink wash
• layered digital painting
• flat-shape illustration
• collage or composite logic
• Judge whether the medium handling supports the image's meaning.
• For AI art, stay on visible evidence:
• structural coherence
• surface repetition
• hand or text issues
• object fusion
• local polish vs whole-image logic

7. Cross-image style synthesis

When the user gives a same-artist batch, series, or motif set, add a set-level read:

• repeated motifs
• recurring palette policy
• edge and mark habits
• spatial habits
• symbolic habits
• what is stable vs what drifts

Do not invent an artist DNA from unrelated images.

8. Overall judgment

• Explain whether the work wins through structure, atmosphere, symbolism, surface handling, cultural layering, or some mix.
• Name the biggest strength and the real limit.
• Distinguish "looks sophisticated" from "holds together."

Common failure patterns

• rich symbolism claimed without visible support
• high polish but weak pictorial structure
• borrowed style signals without internal coherence
• cross-cultural motifs used as surface decoration without deeper integration
• local detail energy without whole-image control
• AI-art polish masking unresolved anatomy, text, or object logic

Study report emphasis

• recommend one descriptive seeing drill before interpretation
• recommend one master-study or reconstruction drill
• recommend one formal reduction drill
• recommend one symbolism, iconography, or reference-building drill
• if it is a batch, add one consistency drill and one anti-imitation warning

Photography Method

Use this route for portraits, street, documentary, landscape, product, fashion, architecture, still life, commercial, editorial, and other camera-led images.

This route should read more like a shoot reverse-engineering pass than a generic image summary.

Evidence ladder

Keep claims separated by certainty:

• FACT: directly visible in the image
• INFERENCE: strong technical or contextual read backed by visible evidence
• HYPOTHESIS: plausible but weaker read that needs a caveat or an alternate explanation

Do not flatten all three into one voice. This matters most when judging lens choice, light setup, authenticity, intent, or documentary truth.

Core lenses

1. Boundary and intent 2. Forensics and medium 3. Scene slicing 4. Geometry and optics reverse 5. Lighting reverse 6. Color, material, and processing 7. Narrative, emotion, and culture 8. Cross-image habits when the batch supports it

Analysis order

1. Boundary and intent

• State what the photo is about on the surface.
• Name the likely subtype if helpful: portrait, street, documentary, landscape, architectural, product, fashion, editorial, still life, and so on.
• Note what is missing or uncertain: low resolution, crop, absent EXIF, one-frame limitation, or no series context.
• If the user clearly wants one lens, honor that focus. Example: "light critique" does not need a long genre essay.

2. Forensics and medium

• Give a first-pass authenticity read: likely authentic capture, likely AI-heavy, likely composite, likely mixed, or unknown.
• Name the likely medium signature when evidence exists: phone, digital camera, film, scan, CGI-like render, or unknown.
• Estimate post-processing intensity: none, light, moderate, heavy.
• Back each non-trivial claim with visible evidence:
• edge behavior
• skin or texture realism
• repeated patterns
• inconsistent reflections
• unnatural text or hands
• sharpening halos
• grain/noise character

Do not turn this into a witch hunt. If the evidence is thin, say unknown.

3. Scene slicing

• Break the image into foreground, midground, and background when that structure exists.
• For each important element, inspect:
• focus status
• object identity
• physical state or wear
• relationship to nearby elements
• color and texture
• light on that element
• possible cultural or time cues
• If useful, mention off-frame pressure: what the crop implies, where space continues, or what seems just outside the shot.
• For reflections, shadows, mirrors, and screens, treat them as separate evidence zones.

This step is especially useful when a photo feels strong but the reason is buried in small details.

4. Geometry and optics reverse

• Estimate camera height, angle, subject distance, and lens feel.
• Comment on depth-of-field behavior, motion treatment, perspective stretch or compression, and distortion.
• If you make a technical reverse read, prefer exclusion logic:
• why this feels closer to a wider lens than a crop from far away
• why this blur feels optical instead of added later
• why the motion looks frozen, dragged, or intentionally blurred
• If the evidence is insufficient, mark the reverse read as hypothesis instead of dressing it up as fact.

5. Lighting reverse

• Identify the likely key light, fill behavior, rim or edge separation, practicals, and negative fill if visible.
• Judge direction, softness, contrast ratio feel, highlight shape, and shadow edge quality.
• Distinguish natural light, window light, flash, continuous artificial light, mixed light, and available light when possible.
• For black-and-white images, read tonal separation and zone placement instead of hue.
• If the light is doing the heavy lifting emotionally, say that explicitly.

6. Color, material, and processing

• Name primary, secondary, and accent color behavior when color matters.
• Judge saturation policy, color temperature bias, tonal range, and whether the grade is neutral, stylized, nostalgic, cinematic, commercial-clean, and so on.
• Inspect material rendering:
• skin
• fabric
• metal
• glass
• stone or wood
• water, haze, or atmosphere
• Comment on grain, noise, sharpening, micro-contrast, retouching, and texture preservation.
• Separate "technically imperfect" from "usefully expressive."

7. Narrative, emotion, and culture

• Name 3-6 emotions the image pushes toward and tie each to visible evidence.
• Give one main reading, then alternatives when the image plausibly supports them.
• Call out the image's emotional hook: the one detail, gesture, timing, or contradiction that makes it stay in memory.
• Use context carefully:
• subject + medium + form + context -> content
• For documentary, journalistic, or social images, discuss truth claim, staging risk, ethics, and social context if relevant.
• For commercial images, also inspect attention path, desire trigger, and brand fit.
• When useful, mention artistic lineage or nearby references, but only with evidence.

8. Cross-image synthesis

When the user provides a same-shooter batch, series, or before/after set, add a set-level read:

• subject preference
• composition habits
• distance and intimacy pattern
• lighting bias
• color bias
• processing habit
• repeated narrative motifs
• what is consistent vs what drifts

Do not fake a style DNA from two unrelated images.

9. Overall judgment

• Explain whether the image wins through control, immediacy, mood, tension, narrative, surface polish, or some mix.
• Name the biggest strength and the real limiting factor.
• Distinguish "good photo" from "good fit for the user's goal." A strong documentary frame and a strong product ad obey different standards.

Common failure patterns

• technically clean but emotionally flat
• strong subject buried by distracting edges, background, or timing
• good mood with weak structural discipline
• heavy edit or retouch that damages the truth claim
• impressive light with no clear subject priority
• dramatic grading used to cover a weak frame
• expressive intent claimed after the fact instead of visible in the capture

Study report emphasis

• recommend one capture drill tied to the image's real weakness
• recommend one reverse-engineering drill: lens, light, or timing
• recommend one editing or culling drill
• recommend one observation habit for future shoots
• if the user studies a series, add one consistency drill and one anti-drift rule

Presentation And Document Method

Use this route for slide decks, report pages, proposal pages, executive-summary layouts, research pages, investor decks, and other document-like visuals where argument flow and decision support matter more than poster impact.

This route should read like a communication-system critique: what this page is supposed to help a reader understand, how the page fits into a larger sequence, and whether the design supports quick decisions and sustained reading.

Evidence ladder

Keep claims separated by certainty:

• FACT: directly visible headings, bullets, charts, tables, callouts, captions, and page layout
• INFERENCE: strong read about audience, presentation mode, or page role backed by visible structure
• HYPOTHESIS: weaker assumptions about meeting context, speaker script, or hidden sequence that need caveats

Do not judge a slide like a poster if the page's job is to explain, compare, or support decisions.

Core lenses

1. Boundary and page role 2. Argument and message slicing 3. Hierarchy and scan path 4. Evidence support quality 5. Density, pacing, and audience mode 6. Sequence or document-system coherence 7. Decision support and actionability

Analysis order

1. Boundary and page role

• Name the likely page type: title slide, agenda, section opener, argument slide, data slide, recommendation page, appendix page, report spread.
• State the likely audience mode:
• presenter-led
• self-read
• mixed
• Note what is missing or uncertain:
• only one page shown
• no prior/following pages
• no speaker notes
• low-resolution crop

2. Argument and message slicing

• Identify the page's main takeaway.
• Break the page into message units:
• title
• supporting points
• chart/table
• quote
• annotation
• recommendation
• footnote/source
• State what the reader should understand first and what requires slower reading.

3. Hierarchy and scan path

• Inspect title strength, grouping, reading path, whitespace, emphasis, and whether key claims survive fast scanning.
• Ask whether the page can be skimmed correctly in a few seconds.
• Distinguish neatness from actual comprehension support.

4. Evidence support quality

• Judge how well charts, tables, screenshots, citations, and callouts support the main point.
• Ask whether supporting evidence is:
• relevant
• readable
• proportionate
• over-dense
• under-explained
• If a chart exists, say whether it is carrying the right burden or doing too much.

5. Density, pacing, and audience mode

• Explain whether the page is optimized for live presentation, async reading, or neither.
• Judge text density, annotation amount, and whether the reader is forced into wall-of-text reading.
• Distinguish "serious-looking" from "decision-friendly."

6. Sequence or document-system coherence

• If multiple pages exist, inspect:
• consistency of page logic
• consistent chart treatment
• stable hierarchy
• pacing from broad point to proof to action
• If only one page exists, state what role it appears to play in a larger argument.

7. Decision support and actionability

• Ask whether the page helps a reader decide, remember, compare, or act.
• Name what is missing if the page looks polished but fails to move the decision.

8. Overall judgment

• Explain whether the page wins through clarity, evidence support, sequence discipline, executive readability, or some mix.
• Name the biggest strength and the real blocker.
• Distinguish "clean slide" from "useful slide."

Common failure patterns

• attractive deck styling with no sharp takeaway
• dense slide that only works if the presenter explains everything verbally
• chart present but not actually tied to the headline
• report page that reads like a slide, or slide that reads like a report page
• sequence inconsistency across pages causing cognitive reset

Study report emphasis

• recommend one headline-to-evidence drill
• recommend one density reduction drill
• recommend one sequence-role mapping exercise
• if several pages exist, add one system-consistency exercise

Product And Industrial Design Method

Use this route for product renders, object photos, industrial design concepts, furniture-object studies, device interfaces tied to hardware form, prototype shots, packaging structures, and other visuals where the object itself is the design subject.

This route should read like an object critique: what problem the object seems to solve, how its form is organized, how it might feel in use, and whether the material and production logic hold up.

Evidence ladder

Keep claims separated by certainty:

• FACT: directly visible form, joints, controls, materials, interfaces, proportions, and packaging structure
• INFERENCE: strong read about ergonomics, use scenario, brand position, or manufacturing logic backed by visible evidence
• HYPOTHESIS: weaker assumptions about internal mechanism, production process, or durability that need caveats

Do not judge a product render like a poster or a product photo like pure photography if the real question is object quality.

Core lenses

1. Boundary and object type 2. Form slicing 3. Function and affordance 4. Ergonomics and use sequence 5. Detail logic and manufacturability 6. Material, color, and finish 7. Brand signal and market position 8. Cross-view consistency when the batch supports it

Analysis order

1. Boundary and object type

• Name the object type: consumer device, tool, furniture object, appliance, package structure, wearable, toy, and so on.
• State the likely primary use scenario.
• Note what is missing or uncertain:
• only one hero render
• no hand or body scale reference
• no exploded or rear view
• no interface state shown

2. Form slicing

• Break the object into major parts:
• core volume
• handle or grip zone
• control zone
• interface or display zone
• support base or hinge zone
• accessory or attachment zone
• Explain the form language:
• soft
• sharp
• technical
• friendly
• rugged
• luxury
• toy-like
• Judge overall proportion and silhouette strength.

3. Function and affordance

• Ask whether the form clearly suggests what the object does and how it is used.
• Inspect:
• buttons
• grips
• handles
• lids
• seams
• ports
• slots
• feet
• packaging open/close logic
• Distinguish clean minimalism from unclear affordance.

4. Ergonomics and use sequence

• Evaluate how a person would likely pick it up, orient it, hold it, open it, operate it, and put it away.
• If body scale is visible, use it.
• If body scale is absent, keep ergonomic claims as inference.
• Explain whether the object feels comfortable, precise, awkward, overbuilt, fragile, intimidating, or inviting.

5. Detail logic and manufacturability

• Inspect seams, split lines, joinery, edge radii, fastener concealment, venting, tolerance feel, and assembly logic.
• Ask whether details look:
• manufacturable
• over-styled
• under-resolved
• prototype-like
• production-ready
• For packaging, inspect folding logic, stacking logic, label placement, and unboxing cues.

6. Material, color, and finish

• Name primary, secondary, and accent material families.
• Judge material-finish logic:
• tactile intent
• durability signal
• cleanliness
• premium feel
• playfulness
• technical seriousness
• Distinguish surface styling from material honesty.

7. Brand signal and market position

• Explain what kind of market tier or brand voice the object suggests.
• Ask whether the form and finish align with that signal.
• Distinguish original character from trend-following.

8. Cross-view synthesis

When the user gives multiple views, add:

• whether the main form idea survives from every angle
• whether interface, grip, and structure agree
• whether the object is coherent as a product rather than only as a hero render

Do not fake production confidence from one beauty shot.

9. Overall judgment

• Explain whether the product wins through clarity of function, ergonomics, silhouette, detail intelligence, material discipline, or some mix.
• Name the biggest strength and the real blocker.
• Distinguish "rendered attractively" from "designed convincingly."

Common failure patterns

• dramatic silhouette with weak real-world handling
• minimal surface hiding unclear controls or seams
• premium styling with cheap-feeling detail logic
• interesting form language that never resolves into usable affordance
• packaging that looks graphic but opens awkwardly

Study report emphasis

• recommend one use-sequence sketch drill
• recommend one silhouette and proportion comparison drill
• recommend one seam/joint/detail study
• recommend one material-finish discipline drill

Scientific And Medical Imaging Method

Use this route for medical scans, x-rays, CT or MRI screenshots, microscope views, pathology plates, lab figures, satellite or remote-sensing imagery, technical evidence images, and other visuals where the image itself serves as evidence rather than illustration.

This route should read like an interpretability audit: what the image appears to show, how reliably it shows it, what annotation or display choices help or hurt reading, and where caution is required.

Evidence ladder

Keep claims separated by certainty:

• FACT: directly visible structures, labels, overlays, scales, false-color choices, crops, and artifacts
• INFERENCE: strong read about modality, region of interest, or interpretive challenge backed by visible evidence
• HYPOTHESIS: weaker assumptions about diagnosis, measurement, or scientific conclusion that need explicit caution

Do not act like a clinician or domain specialist unless the user separately provided that context. Stay anchored to image legibility, annotation, and visible evidence.

Core lenses

1. Boundary and evidence type 2. Signal and region slicing 3. Annotation, scale, and orientation 4. Artifact and clarity risk 5. Display choices and comparability 6. Audience fit and interpretation limits 7. Cross-image synthesis when the batch supports it

Analysis order

1. Boundary and evidence type

• Name the likely evidence type: radiology-like scan, microscope image, pathology tile, lab figure, remote-sensing image, technical inspection image, and so on.
• State what is missing or uncertain:
• no legend
• no modality label
• no scale bar
• crop only
• no before/after or comparison view

2. Signal and region slicing

• Identify the primary region of interest and secondary regions if visible.
• Explain where the signal seems strongest, weakest, cluttered, or ambiguous.
• If multiple panels exist, state how they differ and what comparison the viewer is being asked to make.

3. Annotation, scale, and orientation

• Inspect arrows, labels, captions, panel tags, orientation markers, scale bars, units, and color legends.
• Ask whether a non-expert or expert reader could locate the key area without extra narration.
• Distinguish adequate annotation from over-annotation and under-annotation.

4. Artifact and clarity risk

• Look for visible issues:
• blur
• noise
• compression
• clipping
• low contrast
• aliasing
• false-color ambiguity
• overlaid text blocking the signal
• Explain where interpretation risk is highest.

5. Display choices and comparability

• Judge whether brightness, contrast, crop, orientation, panel alignment, and color mapping help fair comparison.
• If the image is part of a scientific figure, ask whether the display choices support or potentially distort interpretation.
• Keep claims on visible display integrity unless the underlying data is actually shown.

6. Audience fit and interpretation limits

• State who the image currently seems legible for: specialist, trained reader, or general audience.
• Explain what the image can support and what it cannot support on its own.
• If the user wants a public-facing explanation, recommend where annotation or simplification is needed.

7. Cross-image synthesis

When the user gives multiple panels, timepoints, or modalities, add:

• comparability of scale and contrast
• consistency of annotation
• whether the comparison is visually fair
• where side-by-side reading is most likely to fail

Do not turn a visual inspection into a medical or scientific conclusion.

8. Overall judgment

• Explain whether the image succeeds as evidence through clarity, annotation, comparability, and interpretability, or whether those are breaking down.
• Name the strongest support and the main risk.
• Distinguish "scientifically serious-looking" from "actually readable and careful."

Common failure patterns

• missing scale or orientation
• color maps that look dramatic but mislead
• labels covering the key region
• before/after panels shown with inconsistent display conditions
• specialist image presented to a general audience with no bridge explanation

Study report emphasis

• recommend one annotation and labeling drill
• recommend one display-normalization or comparison drill
• recommend one audience-translation exercise

Sculpture, Installation, And Craft Method

Use this route for sculpture, installation, ceramics, metalwork, glass, assemblage, textile craft, site-specific work, gallery objects, and other three-dimensional or materially constructed works where presence and making matter more than product use.

This route should read like a spatial-material critique: how the work occupies volume, how it is assembled, how it asks to be viewed, and what kind of presence it generates.

Evidence ladder

Keep claims separated by certainty:

• FACT: directly visible mass, parts, supports, joins, textures, plinths, hanging systems, surrounding space, and viewer relation
• INFERENCE: strong read about process, ritual use, fragility, site relation, or symbolic stance backed by visible evidence
• HYPOTHESIS: weaker assumptions about internal armature, hidden mechanics, or artist intention that need caveats

Do not flatten a three-dimensional work into a single front-view picture unless the image forces that limitation.

Core lenses

1. Boundary and work type 2. Volume and part slicing 3. Material and assembly logic 4. Site, support, and viewing path 5. Light, shadow, and bodily presence 6. Meaning, ritual, or symbolic charge 7. Craft resolution and durability cues 8. Cross-view synthesis when the batch supports it

Analysis order

1. Boundary and work type

• Name the likely work type: freestanding sculpture, wall relief, suspended installation, ceramic vessel, craft object, assemblage, environmental installation, wearable craft object, and so on.
• State the likely scale if visible.
• Note what is missing or uncertain:
• only one angle
• no scale figure
• no site context
• no close-up of joinery or surface

2. Volume and part slicing

• Break the work into major masses, arms, voids, hanging elements, support structures, or repeated modules.
• Explain how the eye moves around or through the work.
• Distinguish solid mass from negative space and whether voids are doing real compositional work.

3. Material and assembly logic

• Name primary and secondary materials.
• Inspect:
• joinery
• seams
• weave or coil logic
• welds
• glaze or finish
• carving traces
• fastening systems
• Judge whether the assembly feels integral, provisional, delicate, rugged, ceremonial, or unresolved.

4. Site, support, and viewing path

• Ask how the work meets the floor, wall, ceiling, plinth, case, or landscape.
• Explain whether the support disappears, becomes part of the work, or undermines it.
• If the work is site-specific, read how it alters movement, pause, sightline, or crowd behavior.

5. Light, shadow, and bodily presence

• Judge how the work uses reflection, translucency, matte absorption, cast shadow, or silhouette.
• Ask what a viewer's body feels in relation to it:
• invited close
• held back
• dwarfed
• protected
• threatened
• slowed down
• Distinguish surface beauty from full spatial presence.

6. Meaning, ritual, or symbolic charge

• Give one main reading and, if needed, alternatives.
• Tie symbolism to visible evidence:
• form
• material
• repetition
• placement
• scale
• damage or fragility
• Avoid inventing biography or intent unless the work itself strongly supports the reading.

7. Craft resolution and durability cues

• Inspect whether the finish, edges, alignment, and material behavior feel precise, rough by choice, rough by limitation, fragile, archival, or temporary.
• For craft work, distinguish "labor-intensive" from "resolved."

8. Cross-view synthesis

When the user gives multiple views, add:

• whether the work holds from every angle
• whether material and support logic stay convincing
• whether the emotional read changes when the viewpoint changes

Do not fake whole-body knowledge from one cropped catalog shot.

9. Overall judgment

• Explain whether the work wins through mass, material, tension, presence, ritual charge, craft intelligence, or some mix.
• Name the strongest move and the weakest unresolved point.
• Distinguish "interesting object" from "compelling spatial work."

Common failure patterns

• front-view drama with weak side or rear logic
• rich materiality with no strong volumetric idea
• concept declared by text but not carried by form
• craft labor visible, but structure still feels arbitrary
• installation scale doing the work that the forms themselves do not

Study report emphasis

• recommend one multi-angle volume sketch drill
• recommend one material/join detail study
• recommend one site-and-viewer-path exercise
• if it is a series, add one consistency-of-presence drill

Typography And Lettering Method

Use this route for type posters, lettering studies, wordmarks, font proofs, calligraphy, kinetic-type stills, type specimens, and other images where the letterforms themselves are the main subject.

This route should read like a type critique: how the letters are built, how they relate to each other, what voice they carry, and whether they still perform as readable language.

Evidence ladder

Keep claims separated by certainty:

• FACT: directly visible letter shapes, spacing, stroke contrast, alignment, scale changes, and texture
• INFERENCE: strong read about era, register, intended tone, or use case backed by visible letterform behavior
• HYPOTHESIS: weaker assumptions about font family lineage, production process, or brand strategy that need caveats

Do not treat typography-led work like generic layout decoration.

Core lenses

1. Boundary and type artifact 2. Character construction 3. Spacing, rhythm, and texture 4. Hierarchy and reading behavior 5. Voice, register, and use fit 6. Production or system consistency 7. Cross-sample synthesis when the batch supports it

Analysis order

1. Boundary and type artifact

• Name the likely artifact: wordmark, type poster, lettering piece, calligraphic composition, font proof, specimen, title treatment, motion-type still.
• State the likely job:
• brand name
• expressive title
• readable text block
• display texture
• formal exploration
• Note what is missing or uncertain:
• only one word shown
• no lowercase/uppercase coverage
• no small-size test
• cropped baseline or margin context

2. Character construction

• Inspect stroke contrast, terminals, counters, joins, curves, corners, x-height feel, ascender/descender behavior, and structural consistency.
• Ask whether the letterforms feel designed, borrowed, distorted, improvised, or unfinished.
• Distinguish expressive deviation from broken construction.

3. Spacing, rhythm, and texture

• Judge kerning, tracking, word spacing, line spacing, and black-white balance.
• Explain whether the text creates an even texture or lurches from tight to loose.
• If the piece is display-first, say whether the rhythm still supports reading.

4. Hierarchy and reading behavior

• Inspect scale shifts, weight shifts, alignment, and how the eye enters and exits the text.
• Ask whether the composition asks to be read, scanned, decoded, or merely looked at.
• Distinguish deliberate friction from accidental reading resistance.

5. Voice, register, and use fit

• Explain the voice:
• institutional
• literary
• luxury
• brutal
• playful
• technical
• ceremonial
• handmade
• Ask whether that voice fits the likely use case.
• Distinguish authentic character from trend styling.

6. Production or system consistency

• Inspect whether multiple sizes, weights, or glyph behaviors feel like one system.
• For lettering, inspect whether alternate letters feel related or patched together.
• For calligraphy, inspect whether pressure, slant, and rhythm remain coherent.

7. Cross-sample synthesis

When the user gives several words, weights, or poster variants, add:

• structural consistency
• spacing consistency
• whether the expressive range still feels like one family or one hand

Do not infer full type-system quality from one heroic wordmark crop.

8. Overall judgment

• Explain whether the work wins through form quality, spacing control, texture, voice, readability, or some mix.
• Name the strongest move and the limiting flaw.
• Distinguish "looks typographic" from "is typographically convincing."

Common failure patterns

• expressive shapes with weak spacing
• attractive texture that collapses at reading speed
• one memorable character carrying an otherwise weak alphabet
• trend-coded distress or distortion hiding basic construction problems
• brand voice claimed by styling without functional fit

Study report emphasis

• recommend one letter reconstruction drill
• recommend one spacing and text-color drill
• recommend one use-case fit exercise
• if multiple samples exist, add one family-consistency drill

Universal Fallback Method

Use this route when the image deserves a serious reading but does not fit any named route strongly enough.

This route is not "shrug and summarize." It is a first-principles visual analysis for novel, mixed, technical, or unfamiliar artifacts.

Evidence ladder

Keep claims separated by certainty:

• FACT: directly visible layers, marks, zones, text, objects, interactions, and boundaries
• INFERENCE: strongest plausible read about function, audience, or use backed by visible evidence
• HYPOTHESIS: weaker route guesses or contextual reads that need alternatives or caveats

Core job

1. identify what kind of artifact this seems to be 2. describe how it organizes attention 3. explain what it appears to do, communicate, or evidence 4. state uncertainty clearly 5. suggest which named route would become useful if more context arrived

Analysis order

1. Boundary and artifact guess

• State what the image most likely is in plain language.
• List the top two or three plausible categories without forcing certainty.
• Note what key context is missing:
• source file type
• surrounding pages or frames
• scale
• motion
• captions
• user goal

2. Layer and zone slicing

• Break the visual into major zones or layers:
• primary content
• secondary content
• text or metadata layer
• interface/controls
• empty or buffer space
• background or environmental context
• State which zone dominates attention and why.

3. Attention structure

• Explain how the eye moves.
• Judge whether the artifact uses:
• focal point
• sequence
• comparison
• pattern
• density
• contrast
• repetition
• Distinguish deliberate complexity from accidental clutter.

4. Functional reading

• Ask what the image seems to help a viewer do:
• recognize
• compare
• feel
• navigate
• diagnose
• decide
• remember
• decode
• If the function is unclear, say so directly and explain what conflicts are causing that.

5. Meaning and evidence

• Give one main interpretation of what the artifact is doing.
• Add alternatives when the evidence supports them.
• Tie every interpretation to visible evidence rather than defaulting to theory names.

6. Uncertainty map

• State where confidence is high and where it collapses.
• Name what extra context would most improve the read.
• If one named route is becoming likely, say which one and why.

7. Overall judgment

• Explain whether the image succeeds on its apparent terms.
• Name the strongest visible quality and the most serious blind spot.
• Distinguish "interesting" from "legible" and "legible" from "effective."

Study report emphasis

• recommend one context-gathering step
• recommend one redraw, regrouping, or re-annotation exercise
• recommend one comparison exercise against a better-understood artifact type

Output Contract

Always return two artifacts:

1. Analysis Report 2. Study Report

1. Route Decision

Open with:

• chosen route
• confidence
• one short reason tied to visible evidence and the user's goal

If the image is a single frame, render, or cropped detail, note that limitation early.

2. Analysis Report

Start with a short digest, then expand.

Recommended structure:

• Quick Read: 3-5 concise findings
• Detailed Analysis: use the route-specific dimensions from the chosen method file
• Overall Judgment: what works, what weakens the image, and what the image is trying to do

Rules:

• Tie every important claim to visible evidence.
• Prefer plain language over theory names unless the theory genuinely improves clarity.
• If the user asked for comparison, explicitly separate shared traits from differences.
• If the route confidence is not high, keep the judgment proportional.

Photography minimums

When the chosen route is photography, the detailed section should cover these blocks when the evidence allows:

• forensics and medium
• scene slicing or foreground/midground/background read
• geometry or optics read
• lighting read
• color/material/processing read
• emotion/context read

Separate direct facts from stronger and weaker inferences instead of blending them into one certainty tone.

Graphic design minimums

When the chosen route is graphic-design, the detailed section should cover these blocks when the evidence allows:

• task and format
• message slicing
• hierarchy
• structure or grid
• typography
• color/asset treatment/brand fit
• persuasion or usability logic

Painting and illustration minimums

When the chosen route is painting-illustration, the detailed section should cover these blocks when the evidence allows:

• medium read
• descriptive inventory
• formal language
• spatial or compositional logic
• symbol/culture/meaning read
• medium or process control

Interior design minimums

When the chosen route is interior-design, the detailed section should cover these blocks when the evidence allows:

• room type and program
• zoning or circulation
• form or envelope read
• lighting read
• material/furniture/palette read
• usability or lived-reality read

Architecture and urban minimums

When the chosen route is architecture-urban, the detailed section should cover these blocks when the evidence allows:

• building or place type
• site/context read
• massing or silhouette
• facade/opening logic
• ground plane/public realm
• material/light/environmental response

Infographic and diagram minimums

When the chosen route is infographic-diagram, the detailed section should cover these blocks when the evidence allows:

• format and information task
• message or claim slicing
• structural readability
• data/logic integrity cues
• labeling/annotation
• interpretation risk

Product and industrial design minimums

When the chosen route is product-industrial-design, the detailed section should cover these blocks when the evidence allows:

• object type and use scenario
• form slicing
• affordance or function logic
• ergonomics or use sequence
• detail/manufacturing logic
• material/color/finish read

Comics and sequential minimums

When the chosen route is comics-sequential, the detailed section should cover these blocks when the evidence allows:

• page or sequence type
• panel or beat slicing
• reading order
• staging and acting
• text-image coordination
• pacing or payoff

Fashion and styling minimums

When the chosen route is fashion-styling, the detailed section should cover these blocks when the evidence allows:

• look type or styling context
• silhouette or layering read
• garment/material/accessory read
• body/styling balance
• persona/brand/social signal
• wearability or styling logic

Sculpture, installation, and craft minimums

When the chosen route is sculpture-installation-craft, the detailed section should cover these blocks when the evidence allows:

• work type and scale context
• volume or part slicing
• material/join/assembly read
• viewing path or site relation
• presence/meaning/ritual read
• craft-resolution read

Game visual design minimums

When the chosen route is game-visual-design, the detailed section should cover these blocks when the evidence allows:

• image type and gameplay context
• scene/ui/HUD slicing
• readability and navigation cues
• environment/character/interaction clarity
• art direction vs play function
• player interpretation risk

Scientific and medical imaging minimums

When the chosen route is scientific-medical-imaging, the detailed section should cover these blocks when the evidence allows:

• image/evidence type
• signal and annotation read
• clarity or artifact risk
• interpretability limits
• communication adequacy for the likely audience

Avoid pretending to provide professional diagnosis. Keep the analysis on visible evidence, labeling, and interpretability unless the user has separately provided domain context.

Typography and lettering minimums

When the chosen route is typography-lettering, the detailed section should cover these blocks when the evidence allows:

• type artifact or use case
• letterform construction
• spacing/rhythm/texture
• voice or stylistic register
• legibility and performance at likely size
• system consistency if multiple samples exist

Presentation and document minimums

When the chosen route is presentation-document, the detailed section should cover these blocks when the evidence allows:

• page or slide role
• argument/message slicing
• hierarchy and scan path
• chart/table/callout support quality
• page-sequence coherence if multiple pages exist
• decision-support clarity

Film frame minimums

When the chosen route is film-frame, the detailed section should cover these blocks when the evidence allows:

• frame limits
• shot or camera read
• blocking or composition
• light/color/texture
• dramatic function or genre pressure

Generic mixed minimums

When the chosen route is generic-mixed, the detailed section should cover these blocks:

• top route and runner-up route
• universal four-step read
• what extra context would sharpen the route

Universal fallback minimums

When the chosen route is universal-fallback, the detailed section should cover these blocks:

• what kind of visual artifact this seems to be
• what visible layers or zones it contains
• how attention is organized
• what it appears to communicate, do, or evidence
• where uncertainty is high
• which named route would become useful if more context arrived

3. Study Report

Convert the critique into a learning plan.

Include:

• What to Learn Now: 2-4 concepts or habits
• Practice Drills: 2-3 concrete exercises
• What to Watch For Next Time: repeated failure modes or blind spots
• Suggested Comparisons: optional; mention only when the image clearly points toward a useful reference direction

The study report should help the user do something next, not just understand the critique.

Batch rules

• For a same-series batch, include set-level consistency.
• For before/after or versioned work, explain exactly what changed.
• For mixed batches, segment the report by subgroup instead of flattening everything.

Route Matrix

Use this file to decide which analysis method fits the input best.

Selection order

1. Start from the user's learning goal. 2. Check the image's dominant evidence. 3. Break ties with the route that best explains the image's success or failure. 4. If the batch is truly mixed, split it.

Route table

Tie-breakers

• graphic-design vs film-frame: choose graphic-design if reading order, text hierarchy, or marketing effect matters most; choose film-frame if shot language or scene meaning matters most.
• photography vs painting-illustration: choose photography if realism, capture timing, lens/light behavior, or post-processing matters; choose painting-illustration if medium logic, stylization, symbolism, or art-historical reading matters.
• painting-illustration vs interior-design: choose interior-design when the user is evaluating the space itself, even if the space is rendered or stylized.
• photography vs interior-design: choose interior-design when the subject is the room experience; choose photography when the user is critiquing the photographer's capture choices more than the space.
• architecture-urban vs photography: choose photography when the user is mainly critiquing the shot; choose architecture-urban when the real question is the building, facade, site, or city effect.
• architecture-urban vs interior-design: choose interior-design for room experience; choose architecture-urban for exterior form, site, and public realm.
• graphic-design vs infographic-diagram: choose infographic-diagram when truthfulness, notation, and explanation dominate; choose graphic-design when campaign tone, brand effect, or promotion dominates.
• graphic-design vs product-industrial-design: choose product-industrial-design when the object itself is the design subject; choose graphic-design when the object is only one element in a communication layout.
• painting-illustration vs comics-sequential: choose comics-sequential when panel order, balloons, or beats are central; choose painting-illustration when the image is effectively a single artwork.
• film-frame vs comics-sequential: choose film-frame for cinematic shot language; choose comics-sequential when page-level storytelling and reading order matter.
• fashion-styling vs photography: choose photography when the real question is the shot; choose fashion-styling when the real question is silhouette, styling logic, and persona signal.
• sculpture-installation-craft vs product-industrial-design: choose product-industrial-design for use objects and manufactured affordance; choose sculpture-installation-craft for spatial presence, craft, or non-utilitarian three-dimensional work.
• game-visual-design vs graphic-design: choose game-visual-design when play readability or diegetic UI matters; choose graphic-design for promo key art or marketing layouts detached from play.
• scientific-medical-imaging vs infographic-diagram: choose scientific-medical-imaging when the image itself is evidence; choose infographic-diagram when the image is mainly explanatory packaging around evidence.
• typography-lettering vs graphic-design: choose typography-lettering when type itself is the primary subject; choose graphic-design when type serves a broader layout.
• presentation-document vs graphic-design: choose presentation-document when argument flow and page sequence matter; choose graphic-design when single-page campaign impact matters most.
• Use universal-fallback when top scores stay weak and no tie cluster points to a meaningful subgroup.

Route confidence

• high: strong route signals and no serious competitor
• medium: one route leads, but a second route could also work
• low: not enough evidence, or the image is hybrid enough that the chosen route is only provisional

When confidence is low, say so and name the runner-up route.