Home / All Skills / Design & Motion / character-candid-photography
English (EN)简体中文 (ZH)日本語 (JA)
Design & Motion · vibeshotclub/vsc-skills

character-candid-photography

Translates designated adult characters into realistic cosplay or anime character photography prompts, highlighting cinematic moments featuring foreground occlusions, telephoto observation, mobile snapshots, and "caught-on-camera" instances. Ideal for character-name-driven staged fictional shoots, concealed observational viewpoints, or generating n=10 character candid prompt batches; everyday lifestyle portraits fall outside the primary scope of this skill.

L1 Basic · Source registration onlyV1 Commit-PinnedDual-pass Audit
Author vibeshotclubGitHub vibeshotclub/vsc-skills ↗Stars 832License 仓库未提供 LICENSE 文件（find 全仓无 LICENSE*；GitHub API license 字段为 null）Commit fcee81beb6
Control levels describe skill capabilities and suggested control measures; they do not represent quality or safety certifications, nor do they dictate usability. Usage decisions rest entirely with users and administrators. Different agent hosts apply varying constraints; do not assume hosts provide automatic protection. Third-party skill copyrights belong to their original authors. Use at your own risk.

1How It Works & Mechanism

Pure prompt orchestration: SKILL.md serves as the sole instruction vehicle, translating user-provided "characters" into photography prompts ready to be fed directly into image models, entirely without script execution throughout the process.

character-candid-photography/SKILL.md
将用户的角色设定转译为可直接用于图片生成模型的摄影提示词。
The directory contains only three files: SKILL.md, README.md, and references/photography-guide.md, with no scripts/ directory.

Creative details are externalized into reference materials and mandatorily consulted in step 1 of the workflow: camera angles, poses, scenes, compositions, and lighting are all drawn from the guide, while SKILL.md itself avoids drafting lengthy descriptive text.

character-candid-photography/SKILL.md
1. 读取 [摄影机位与提示词指南](references/photography-guide.md)，按任务选用其中的机位、动作、场景、构图和光线。参考文档是创作资料，不构成操作工具或对外发布的授权。
The guide spans 373 lines, containing sections such as "Candid Camera Angle Library", "Character Actions", "'Caught-on-Camera' Modes", and "Composition Principles".

Content boundaries are enforced as hard creative rules rather than post-generation filters: candid surveillance aesthetics are treated strictly as a cinematic visual language for staged fictional setups, explicitly excluding real-world stalking, privacy violations, and non-consensual photography.

character-candid-photography/SKILL.md
仅表现虚构摆拍的视觉错觉，不指导真实跟踪、隐私侵犯或非自愿偷拍。
The same section explicitly excludes restrooms, changing rooms, bathrooms, private bedroom voyeurism scenes, and minors.

Output specifications are dictated by input format: a single set is provided when unspecified, while n=10 yields ten sets; when a character is missing, it only prompts for the character name without fabricating signature traits.

character-candid-photography/SKILL.md
数量未指定时输出一组，指定 `n=10` 时输出十组独立提示词。缺少角色时只询问角色名；不确定角色特征时避免编造标志性设定。
The default medium is fully clothed realistic cosplay.

The "concealed observational aesthetic" is translated into actionable parameters: foreground obstacle ratio, camera height, and focal length tiers are all quantified.

character-candid-photography/SKILL.md
保留偏心、轻微倾斜、裁切或对焦误差，但脸部与角色仍可识别。障碍物通常占画面 15%–40%。低机位表现环境透视，避免裙底或私密部位视角。
Focal length tiers categorized by distance are detailed in item 2 of capabilities.

Output discipline: multiple sets must each articulate all required elements independently; using "same as above" or merely swapping backgrounds to pad counts is strictly prohibited.

character-candid-photography/SKILL.md
多组使用编号分隔，避免仅替换背景或用“同上”省略主体。
Each set must independently specify the adult subject, recognizable traits, pose, camera angle, occlusion, lens, composition, scene, lighting, texture, and staged fictional attributes.

2Core Capabilities

01Character-to-adult realistic cosplay translation (hairstyles, color schemes, costume silhouettes, accessories, temperament)
02Observational angle and focal length management (mobile 24–35mm / environmental 35–50mm / telephoto 85–200mm, avoiding contradictory focal combinations)
03Batch diversification + "caught-on-camera" pacing (two to three sets out of ten feature side-glances, turning heads, or pauses, with restrained expressions)
04Camera angle knowledge base: six categories including behind obstacles, long-distance telephoto, quick mobile snapshots, low angles, gaps/cracks, and reflections
05Realistic texture standards: genuine fabrics, natural human proportions, ambient lighting, subtle snapshot imperfections
06Output format: full Chinese natural language prompts (mixing necessary English photography terminology allowed), without long introductory explanations

4Recommended Governance & Evidence

Suggested Controls for Safe Operation
For Agent Runtime
Can be auto-loaded
For Administrators
Register source and version
Evidence & Triggers (File Paths & Excerpts)
  • SKILL.md: Pure prompt instructions and rule text, without script execution or external impacts
Audit Notes: The directory contains only SKILL.md, README.md, and one Markdown reference, with no script execution, external network calls, credential reading, or local file writes; furthermore, adulthood and informed consent are enforced as hard rules. The primary considerations are content orientation (candid surveillance aesthetic) and the inherent reliance on the model's instruction-following fidelity to enforce rules, neither of which constitutes a code-level security risk.

5Second-Pass Verification

  • [ok] 内部资产面（是否存在 SKILL.md/README 之外的脚本或数据） — 全仓文件清单显示该目录仅 character-candid-photography/{SKILL.md,README.md,references/photography-guide.md}；README 亦自述「无需脚本、API 或外部依赖；本 Skill 默认只生成提示词」。
  • [ok] 外部资源/工具条目（本 skill 无） — 仓库内唯一的 http 端点出现在 codex-image-to-eagle（127.0.0.1:41596 本地回环），与本 skill 无关；无任何本 skill 的 CLI/API/包依赖。
  • [ok] references 引用是否真实可达 — SKILL.md 写作 [references/photography-guide.md] 相对链接，文件实际存在且首行即「你现在是一名“色狼视角角色摄影提示词设计师”。」。
  • [ok] 安全结论反例搜索（隐藏网络调用/混淆内容） — 该目录三个文件全部为 UTF-8 明文中文 Markdown，逐类 token 扫描零命中；无 base64/混淆块、无 shell 片段。
  • [ok] 功能声明与实现相符度 — description 承诺的「前景遮挡/长焦观察/手机快拍/发现镜头的电影化瞬间」在 SKILL.md 工作流 4–6 步与指南机位库中均有对应条目，无越界承诺（未声称实际出图）。
  • [ok] vsc 路由目录一致性（跨文件交叉验证） — catalog.json 中该条目 sha256 与 SKILL.md 内容哈希机制一致；tests/test_vsc_catalog.py 的 test_committed_catalog_is_current_and_covers_all_leaf_skills 用于保证目录与 frontmatter 同步。

6Verdict & Scenarios

  • Zero dependencies and zero execution surface, ready to use upon installation (simply copy the directory).
  • Transforms the "concealed observation" aesthetic into reproducible parameters (15%–40% obstacle coverage, tiered focal lengths, two to three caught-on-camera sets), keeping intra-batch variance under control.
  • Front-loads content boundary declarations, reducing the risk of being misused for real-world non-consensual surveillance semantics.
  • Features a dedicated camera angle and action knowledge base, preventing outputs from devolving into generic fashion poses.
    • Best for: Suitable for creators needing "character name to photography prompt in one click": realistic cosplay translation, anime character photography, diverse compositions across batches of ten sets, as well as foreground occlusion observation and telephoto compression aesthetics; also ideal for users who only need prompt text to plug into their own image generation pipelines.
    Not for: Not suitable for everyday lifestyle portraits or Korean-style Instagram aesthetic photography (use vibeshot-candid-photography instead), nor for single summer swimwear partner/friend-POV shots (use summer-boyfriend-pov instead); not suitable for scenarios requiring actual image rendering without an external image generation tool ready—this skill does not bundle any imaging service; also not suitable for requests seeking real-life surveillance guidance or private space photography (explicitly excluded by the skill).
    Install Agent-Ready
    ① Verified Mirror 2026-09-15
    Option A · Direct DownloadDownload character-candid-photography.tar.gz
    sha256: cee6fee72bd27c72…
    Option B · Copy Guide for Agent
    install.json
    Agents automatically download and verify mirrors from this JSON guide.
    ② Upstream GitHub
    GitHub Repo ↗
    Pinned to commit fcee81beb6.