thanhnguyxnorg/blendops

environment-shot

Plan a wide environment / establishing-shot scene (locale, time of day, weather, focal subject, depth, atmosphere) before any modeling or render attempt.

查看源码

使用方法

  1. 复制安装命令,在自己的智能体环境安装前先检查源码包。
  2. 从原始文档核对所需工具、凭据和支持的宿主。
  3. 说明输入材料与期望交付物,并对照来源检查输出。
npx skills add https://github.com/thanhnguyxnorg/blendops --skill environment-shot

开始前需了解

  • 安装不包含模型订阅、API 额度或访问文件的权限。
  • 仓库星数和安装量反映活跃情况,不等于独立质量或安全测试。
仓库原始内容

按源仓库内容呈现,保留标题、案例、代码、表格、链接以及原文引用的演示图片。

environment-shot

Purpose

Translate a wide place-establishing intent (locale, time of day, weather, atmosphere, foreground/midground/background depth layers, and focal subject or explicit "none") into a structured scene plan with an explicit assumptions register. Distinct from character-portrait (single subject), product-hero-scene-planner (manufactured products), and interior-scene (interior rooms/furniture/surfaces — an establishing shot of an interior room belongs there when furniture or surfaces dominate). Hands off asset rows to asset-discovery and validation to composition-quality-checker, lighting-quality-checker, and scene-readiness.

Quick start

  • confirm upstream brief from intent-to-3d-brief-writer and verify a realizable (non-invented) locale scope
  • pin locale, time of day, weather, and atmosphere
  • pin foreground (FG), midground (MG), and background (BG) depth layers
  • pin focal subject (or explicitly declare none)
  • record an explicit assumptions register
  • flag asset rows for asset-discovery and emit validation handoffs with Runtime status: Not Run and Artifact status: Not Produced

When to use

  • intent is establishing a wide place (landscape, cityscape, outdoor locale, or wide interior space where furniture/surfaces do not dominate)
  • before any modeling, Geometry Nodes, or scene building
  • when consumer is editorial, cinematic, or brand

When not to use

<!-- blendops:generated:start exclusions -->

  • character portrait → use ../character-portrait/SKILL.md (use character-portrait)
  • product hero → use ../product-hero-scene-planner/SKILL.md (use product-hero-scene-planner)
  • interior arch viz with focus on furniture / surfaces → use ../interior-scene/SKILL.md (use interior-scene)

<!-- blendops:generated:end exclusions -->

Trigger phrases

<!-- blendops:generated:start triggers -->

  • "establishing shot of"
  • "wide shot of"
  • "city / forest / desert / ocean scene"
  • "set the scene"

<!-- blendops:generated:end triggers -->

Prerequisites / readiness

  • brief from intent-to-3d-brief-writer is locked
  • locale scope is realizable and bounded (never invented)
  • target aspect ratio and use case are known

Input schema

Required inputs

InputWhy it is required
Realizable localeDrives bounded asset and depth-layer scope without inventing unrequested worldbuilding
Time of dayDrives sunlight direction, sky tone, and lighting mood
Weather + atmosphereDrives visibility, haze/fog, and surface condition
Focal subject intentIdentifies the visual anchor (or explicitly confirms none)

Optional inputs

InputUse
Reference imagesCalibrate place scope, depth layers, and mood
Aspect ratioGuides framing across FG / MG / BG layers
Era / periodAligns asset discovery rows

Assumptions to confirm

  • Single time-of-day snapshot (unless animation is explicitly requested upstream).
  • Locale scope is realizable and bounded, not invented.
  • Interior rooms dominated by furniture or surfaces are routed to interior-scene.
  • Planning-only output: Runtime status stays Not Run and Artifact status stays Not Produced.

Output schema

Primary output

A structured place-establishing plan covering: Locale (bounded scope), Time of day, Weather, Atmosphere, Depth layers (FG / MG / BG), Focal subject (or explicit none), Mood / style, and an Assumptions register.

Secondary output

  • asset rows flagged for asset-discovery
  • validation handoffs to composition-quality-checker, lighting-quality-checker, and scene-readiness
  • explicit Runtime status: Not Run and Artifact status: Not Produced labels

Evidence / caveat output

txt
Runtime status: Not Run | Attempted | Produced | Verified | Failed
Artifact status: Not Produced | Produced | Verified | Failed
Evidence used: <links, paths, logs, or "none">
Limitations: <known gaps>

Required laws

  • ../../laws/evidence-before-done.md
  • ../../laws/non-blender-user-language.md
  • ../../laws/no-arbitrary-python-interface.md
  • ../../laws/official-runtime-only.md

Official runtime boundary

Planning-only skill: no modeling, no Geometry Nodes, no render, and no Blender execution. Runtime status stays Not Run and Artifact status stays Not Produced (downgrade-only, never upgrade). If the brief or locale scope is unclear, hand off to intent-to-3d-brief-writer.

Operating procedure

  1. Read the brief from intent-to-3d-brief-writer; confirm wide place-establishing scope and ensure locale scope is realizable (not invented).
  2. Route to interior-scene if the scene is an interior room where furniture or surfaces dominate.
  3. Pin time of day, weather, and atmosphere (per references/time-and-weather-rules.md).
  4. Define wide camera placement and framing intent (per references/wide-shot-camera-rules.md).
  5. Plan foreground (FG), midground (MG), and background (BG) depth layers (per references/depth-layering-rules.md).
  6. Identify the focal subject or explicitly document none.
  7. Record all open assumptions in the assumptions register.
  8. List required assets for asset-discovery and emit validation handoffs to composition-quality-checker, lighting-quality-checker, and scene-readiness with Runtime status: Not Run and Artifact status: Not Produced.

Decision tree

txt
Scene is an interior room where furniture or surfaces dominate?
  → Stop; route to interior-scene
Focus is a single character or a manufactured product?
  → Stop; route to character-portrait or product-hero-scene-planner
Locale scope is unbounded or unclear (would require inventing scope)?
  → Stop; hand off to intent-to-3d-brief-writer to bound locale scope
Locale + time-of-day combination implausible (e.g. "Antarctic noon" with deep dark mood)?
  → Warn; clarify intent
"City scene" + tight close-up framing?
  → Warn; this is not establishing — re-frame or use different planner
Weather conflicts with brief mood (sunny + noir)?
  → Fail; pick one
Focal subject not chosen but every asset has same visual weight?
  → Fail; eye has nowhere to land

Playbooks

Playbook A: City sunset establishing (16:9 cinematic)

35mm wide; camera elevated; warm sun direction; haze layer; subject = skyline silhouette; foreground rooftops; background sky gradient.

Playbook B: Forest mid-day (1:1 brand campaign)

24mm wide; camera at human eye-line; key sun + dappled light; foreground branches; midground tree trunks; background distant haze.

Playbook C: Stylized cyberpunk street (2.39:1 cinematic)

24–35mm; rain wet ground reflections; neon practicals; light haze + atmospheric fog; foreground silhouette; midground signage / cars; background distant towers.

Mode handling

Text-only mode

Produce the environment plan and assumptions register; keep Runtime status: Not Run and Artifact status: Not Produced.

Runtime-ready mode

Planning-only even when runtime is available downstream: keep Runtime status: Not Run and Artifact status: Not Produced (downgrade-only, never upgrade), and hand off to asset-discovery and quality checkers.

Blocked runtime mode

Produce the environment plan and assumptions register; keep Runtime status: Not Run and Artifact status: Not Produced, and log blockers under Limitations.

Validation checklist

  • [ ] Realizable, non-invented locale scope confirmed
  • [ ] Time of day, weather, and atmosphere pinned
  • [ ] Depth layers (FG / MG / BG) explicitly defined
  • [ ] Focal subject (or explicit none) documented
  • [ ] Assumptions register included
  • [ ] Asset rows flagged for asset-discovery
  • [ ] Validation handoffs named (composition-quality-checker, lighting-quality-checker, scene-readiness)
  • [ ] Runtime status: Not Run and Artifact status: Not Produced explicitly set

Pass / Warn / Fail rubric

VerdictCriteria
PassRealizable locale scope; consistent time/weather/atmosphere; explicit FG/MG/BG depth layers; focal subject (or explicit none); assumptions register complete; truth states locked to Not Run / Not Produced; handoffs named.
WarnMinor non-blocking assumption open in the assumptions register; asset list partial.
FailInvented or unbounded locale scope; interior room dominated by furniture/surfaces; contradictory time/weather/mood; missing FG/MG/BG layers; missing focal subject decision; truth states upgraded above Not Run / Not Produced.

Failure handling

  • Locale scope too broad or unclear → hand off to intent-to-3d-brief-writer (never invent scope).
  • Mood and weather contradict → stop and ask user to pick one consistent direction.
  • Asset availability unknown → flag missing items for asset-discovery.

Troubleshooting

ProblemResponse
Sun direction conflicts with shadows in referenceVerify time-of-day + sun azimuth; clarify reference.
Atmospheric haze hides focal subjectReduce haze density or move subject closer.
Background too detailedSimplify; use atmospheric perspective.

Best practices

  • Lock time-of-day + weather + atmosphere together; they form a system.
  • Always plan FG / MG / BG depth layers.
  • Pair establishing shot with the next narrative step (often character portrait or product reveal).

Good examples

  • "Locale: cyberpunk city street, near-future. Time: night, neon lit. Weather: light rain. Atmosphere: haze 0.05 density, volumetric scattering. Sky: minimal (occluded by buildings). Camera: 24mm, eye-level, slight low-angle tilt. FG: silhouetted figure. MG: neon signage + parked cars. BG: distant towers + fog. Focal subject: figure silhouette right-third. Mood: gritty, atmospheric."

Bad examples

  • "Wide shot of a city." — no time, weather, lens, depth.

User-facing response template

txt
Locale (bounded scope): <where + realizable boundaries>
Time of day: <time / sun state>
Weather: <weather state>
Atmosphere: <haze / fog / visibility>
Camera & framing: <wide position / altitude / tilt>
Depth layers: FG: <…> | MG: <…> | BG: <…>
Focal subject: <subject + placement, or explicit "none">
Mood + style: <line>
Assumptions register: <explicit list of confirmed assumptions>

Assets to discover: <rows for asset-discovery>
Runtime status: Not Run
Artifact status: Not Produced
Evidence used: <links, paths, logs, or "none">
Limitations: <known gaps>
Next: asset-discovery | composition-quality-checker | lighting-quality-checker | scene-readiness

Anti-patterns

  • Mixing single-character portraits (character-portrait), manufactured product shots (product-hero-scene-planner), or furniture/surface-dominant interior rooms (interior-scene) into wide place planning.
  • Inventing unbounded locale scope instead of handing off to intent-to-3d-brief-writer.
  • Skipping FG / MG / BG depth layers or failing to declare a focal subject (or explicit none).
  • Prescribing Geometry Nodes, modeling steps, or renders, or upgrading truth states above Not Run / Not Produced.

Cross-skill handoff

<!-- blendops:generated:start handoffs -->

  • [upstream_reference] Brief writer → ../intent-to-3d-brief-writer/SKILL.md
  • [upstream_reference] Asset discovery → ../asset-discovery/SKILL.md
  • [progression] Composition validation → ../composition-quality-checker/SKILL.md
  • [progression] Lighting validation → ../lighting-quality-checker/SKILL.md
  • [progression] Quality readiness → ../scene-readiness/SKILL.md

<!-- blendops:generated:end handoffs -->

Non-goals

  • Model the environment.
  • Generate Geometry Node setups.
  • Render.
  • Run Blender.

References

  • references/time-and-weather-rules.md
  • references/wide-shot-camera-rules.md
  • references/depth-layering-rules.md
  • ../../docs/skill-system.md
来自同一仓库

更多 Skills

全部 Skills