使用方法
- 复制安装命令,在自己的智能体环境安装前先检查源码包。
- 从原始文档核对所需工具、凭据和支持的宿主。
- 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 camera-composition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。
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。
camera-composition
Purpose
Design high-quality composition/camera strategy for product hero scenes in plain, actionable language.
Quick start
- confirm this skill fits your goal
- provide required inputs first
- keep runtime claims evidence-bound
- follow suggested next-skill handoff
When to use
- expanding an upstream scene brief from
product-hero-scene-plannerinto detailed framing and camera direction - defining focal hierarchy, viewer attention path, camera angle, focal length mood, and depth layering before lookdev
- aligning hero framing and negative space with web/UI copy overlays
When not to use
<!-- blendops:generated:start exclusions -->
- as proof of rendered output quality or to validate an existing framing → use
../composition-quality-checker/SKILL.md(use composition-quality-checker) - to design lighting rigs or surface materials/lookdev → use
../lighting-material-plan/SKILL.md(use lighting-material-plan) - to lock final pixel resolution or output aspect specs → use
../resolution-aspect-decision/SKILL.md(use resolution-aspect-decision) - to choose delivery file formats → use
../output-format-decision/SKILL.md(use output-format-decision) - when subject intent is undefined → use
../product-hero-scene-planner/SKILL.md(Delegates out-of-scope requests to product-hero-scene-planner)
<!-- blendops:generated:end exclusions -->
Trigger phrases
<!-- blendops:generated:start triggers -->
- "improve composition and camera"
- "choose hero framing"
- "set camera strategy for product hero"
<!-- blendops:generated:end triggers -->
Prerequisites / readiness
- upstream scene concept from
product-hero-scene-planner(or clear subject and destination context) available - planning-only boundary accepted: runtime status remains
Not Runand artifacts remainNot Produced
Input schema
Required inputs
- subject intent (from
product-hero-scene-planneror user brief) - target output context (web/app/game/render)
Optional inputs
- target aspect ratio constraints (from
resolution-aspect-decisionif already locked) - UI/copy overlay constraints
- cinematic vs neutral tone preference
Assumptions to confirm
- need for negative space
- focal detail priority
- camera motion vs static hero shot intent
Output schema
Primary output
- composition-camera plan with plain-language rationale
Secondary output
- fallback framing variant for uncertain constraints
Evidence / caveat output
- unresolved visual assumptions
- explicit truth states: Runtime
Not Run, ArtifactsNot Produced(downgrade-only, never upgrade)
Required laws
- ../../laws/non-blender-user-language.md
- ../../laws/evidence-before-done.md
- ../../laws/official-runtime-only.md
- ../../laws/no-arbitrary-python-interface.md
Official runtime boundary
Planning-only skill: no Blender execution, no Python scripts, and no artifact generation. Runtime status stays Not Run and artifacts stay Not Produced (truth states are downgrade-only, never upgraded without verified evidence).
Operating procedure
- Define focal hierarchy (primary/secondary/background).
- Define hero-section objective and viewer attention path.
- Select aspect ratio framing strategy (deferring final resolution/aspect locks to
resolution-aspect-decision). - Reserve negative space for UI/text overlays if needed.
- Choose camera angle preset family.
- Set focal length guidance by mood (dramatic vs neutral).
- Define depth layering (foreground/midground/background).
- Define target silhouette/readability checks (for later validation by
composition-quality-checker). - Produce fallback framing variant for uncertainty.
- Translate plan into plain-language rationale.
Decision tree
- If copy overlay required → prioritize negative space and silhouette.
- If product detail priority high → tighter framing and controlled depth.
- If uncertainty high → provide two options and ask preference.
Playbooks
- Playbook A: Full brief path — expand upstream
product-hero-scene-planneroutput into primary framing + fallback variant (Runtime: Not Run,Artifacts: Not Produced). - Playbook B: Overlay-heavy path — prioritize negative space and silhouette separation for UI/copy constraints (
Runtime: Not Run,Artifacts: Not Produced). - Playbook C: High-uncertainty path — provide two distinct framing variants with trade-off rationale and explicit caveats (
Runtime: Not Run,Artifacts: Not Produced).
Mode handling
Text-only mode
- provide primary composition/camera plan and fallback variant in plain language
- report
Runtime: Not RunandArtifacts: Not Produced
Runtime-ready mode
- provide the composition/camera plan plus explicit target criteria for
composition-quality-checkerto validate once a scene exists - keep
Runtime: Not RunandArtifacts: Not Producedduring planning
Blocked runtime mode
- deliver full composition/camera plan and fallback variant with explicit caveats
- keep
Runtime: Not RunandArtifacts: Not Produced; make no visual success claims
Validation checklist
- [ ] focal hierarchy explicit
- [ ] hero framing objective and attention path explicit
- [ ] aspect ratio framing strategy documented (without overriding
resolution-aspect-decision) - [ ] negative space policy documented
- [ ] camera angle preset rationale documented
- [ ] focal length guidance by mood documented
- [ ] depth layering (foreground/midground/background) documented
- [ ] target silhouette/readability checks documented for
composition-quality-checker - [ ] fallback framing variant documented
- [ ] plain-language rationale included throughout
- [ ] truth states explicitly set to
Runtime: Not RunandArtifacts: Not Produced
Pass / Warn / Fail rubric
| Category | Pass | Warn | Fail |
|---|---|---|---|
| Focal hierarchy | Clear primary/secondary/background priority | Partial hierarchy | Missing hierarchy |
| Framing strategy | Matches output context and includes fallback variant | Generic framing or weak fallback | Mismatched framing or missing fallback |
| Camera rationale | Specific plain-language angle, focal-length mood, and depth layering | Vague rationale or unexplained jargon | No rationale or includes Python/execution commands |
| Role & truth boundary | Stays in planning; reports Runtime: Not Run and Artifacts: Not Produced | Mild boundary overlap with lookdev/validation | Claims verified framing quality or upgrades truth states |
Failure handling
- Conflicting goals: provide trade-off options.
- Missing constraints: ask targeted follow-up.
- Unknown destination: use conservative web-hero defaults + caveat.
Troubleshooting
- Subject lost in frame: tighten framing focus, adjust camera angle preset, or increase spatial separation between foreground, midground, and background layers.
- No room for copy: widen negative space allowance and shift primary focal placement off-center.
- Flat spatial staging: adjust camera angle and focal-length perspective or increase depth-layer spacing (route lighting/material contrast issues to
lighting-material-plan). - Existing scene framing needs pass/fail grading: hand off to
composition-quality-checker.
Best practices
- optimize for clarity before stylization
- tie each camera choice to user outcome
- preserve fallback path for uncertain constraints
Good examples
“Primary subject occupies center-right with left negative space for hero copy; camera 3/4 angle chosen for silhouette and depth readability.”
Bad examples
“Use cinematic camera.” (no context, no criteria)
User-facing response template
- Composition summary (focal hierarchy, attention path, negative space, depth layering, silhouette checks)
- Camera strategy summary (aspect strategy, angle preset, focal length by mood)
- Fallback framing variant
- Why this helps your goal (plain-language rationale)
- Truth states (
Runtime: Not Run,Artifacts: Not Produced) and unverified assumptions - Recommended next skill handoff
Anti-patterns
- skipping required laws or runtime boundary statements
- claiming runtime/artifact success without evidence
- claiming runtime or artifact states above
Not Run/Not Produced(violating downgrade-only truth rules) - writing Python code or Blender camera execution commands
- designing lighting rigs or material lookdev instead of handing off to
lighting-material-plan - validating rendered/viewport composition instead of handing off to
composition-quality-checker - locking final export resolution/aspect (
resolution-aspect-decision) or file formats (output-format-decision) - producing jargon-heavy camera specs without plain-language rationale
Cross-skill handoff
<!-- blendops:generated:start handoffs -->
- [upstream_reference] Upstream input from →
../product-hero-scene-planner/SKILL.md - [conditional] Lateral output specs (if needed) (output-format-decision) →
../output-format-decision/SKILL.md - [conditional] Lateral output specs (if needed) (resolution-aspect-decision) →
../resolution-aspect-decision/SKILL.md - [progression] Next (lookdev & lighting) →
../lighting-material-plan/SKILL.md - [progression] Then (validation) (scene-readiness) →
../scene-readiness/SKILL.md - [progression] Then (validation) (composition-quality-checker) →
../composition-quality-checker/SKILL.md
<!-- blendops:generated:end handoffs -->
Non-goals
- executing Blender commands, writing Python scripts, or generating scene/render artifacts
- designing lighting setups or materials (owned by
lighting-material-plan) - validating or scoring completed scene framing (owned by
composition-quality-checker) - locking output resolution/aspect standards (owned by
resolution-aspect-decision) or file formats (owned byoutput-format-decision)
References
- https://www.blender.org/lab/mcp-server/
- https://claude.com/resources/tutorials/using-the-blender-connector-in-claude
- https://docs.blender.org/manual/en/latest/advanced/command_line/index.html