下面是一组“在线 3D 预览/检查工具”生成型 AI 提示词（Prompts）。每条 Prompt 都要求 AI 输出可运行的完整项目（含文件结构、源码、命令、部署与测试/QA），用于在浏览器中上传并预览 GLTF/GLB/STL/OBJ，同时做基础体检并导出报告。
GLTF/GLB 在线检查器：场景结构与资源引用报告
适合把 GLTF/GLB 拖拽上传后快速预览，并生成“节点树/网格统计/贴图引用/潜在问题”报告。
英文 Prompt:
Build a production-ready web app: an online GLTF/GLB inspector and report generator.
Requirements:
- Tech stack: Next.js (App Router) + Type + Three.js.
- UI: single-page tool with drag-and-drop upload and file picker.
- Rendering: show a 3D viewport with orbit controls, grid, axes, background toggle.
- Parsing: use three/examples loaders to load GLTF/GLB locally in the browser (no server upload required).
- Analysis: compute and display:
- Scene graph tree (nodes, meshes, cameras, lights)
- Mesh statistics: vertex/triangle count per mesh and totals
- Materials: list materials, texture maps used, missing/failed textures
- Bounding box, dimensions (in model units), suggested real-world scale input
- Warnings: non-manifold clues if detectable, zero-area faces hints, too many draw calls
- Report export:
- Generate a downloadable report JSON and an HTML summary (client-side) including all metrics and warnings.
- Output deliverables:
1) Full project file tree
2) All source code for each file
3) Installation and run commands
4) Production build command
5) Vercel deployment steps
6) Tests: include at least 5 tests (unit tests for analyzers + UI smoke test) using Vitest and Playwright.
- Quality:
- Add error handling for invalid files and loader errors.
- Ensure the app works offline after initial load.
- Do NOT generate any images; do not mention image generation tools.
Return the answer as a complete, copy-pastable code with clear sections per file.
中文释义:
让 AI 直接产出一个可部署的 GLTF/GLB 在线检查器：既能预览，也能把场景结构、网格/三角面统计、材质贴图引用问题等整理成可下载报告。
STL 在线预览与“打印前体检”报告（不做渲染图）
适合 3D 打印前快速检查 STL：尺寸、法线/面片统计、薄壁/悬空等风险提示，并导出体检报告。
英文 Prompt:
Create a deployable online STL preview + preflight check tool.
Scope:
- Category: 3D preview / inspection (do not generate renders or images).
- Stack: Next.js + Type + Three.js.
Features:
- Upload STL (binary and ASCII), render in an interactive viewport.
- Controls: orbit, wireframe toggle, backface culling toggle, auto-center and fit-to-view.
- Preflight checks (client-side):
- Bounding box and dimensions (mm toggle with user-provided unit scale)
- Triangle count and estimated complexity level
- Normal consistency: detect flipped normals heuristically
- Degenerate triangles detection
- Simple thin-wall heuristic: highlight triangles in regions below a threshold (approximate)
- Overhang heuristic hint (angle threshold, purely informational)
- Report:
- Download JSON report + a human-readable HTML report.
Deliverables:
- Provide the entire project file tree.
- Provide full source code.
- Provide commands: install, dev, build, start.
- Provide deployment guide for Vercel.
- Testing/QA:
- At least 5 automated checks: e.g., STL parser tests, geometry stats tests, report generator tests, Playwright upload flow.
- Add a QA checklist for manual verification.
Constraints:
- No server-side file uploads required.
- Do not mention Midjourney/Stable Diffusion or any image generation.
- Accessibility: basic keyboard navigation and readable error messages.
中文释义:
让 AI 生成一个“STL 在线预览 + 打印前体检”工具：重点是检查与报告，不是出图渲染；并要求给出可运行代码、部署与测试。
OBJ/MTL 资源完整性检查器：贴图引用与路径问题定位
适合排查 OBJ + MTL + 贴图资源打包是否完整、引用路径是否能在浏览器环境正确解析。
英文 Prompt:
Build an online OBJ/MTL package validator and previewer.
Tech:
- Next.js + Type + Three.js.
Inputs:
- Support uploading a ZIP that contains .obj, .mtl, and textures (png/jpg/webp).
- Support uploading files separately (multiple files) and resolving references.
Core requirements:
- Preview the loaded model in a viewport.
- Parse the MTL and list:
- Materials found
- Texture references (map_Kd, map_Ns, bump, etc.)
- Missing files, unsupported formats, suspicious absolute paths
- Provide a "fix suggestions" panel:
- Suggest renaming and relative path normalization
- Suggest a standard folder layout for web delivery
- Generate downloadable validation reports:
- JSON report with missing assets and reference graph
- HTML summary with actionable items
Deliverables:
- Full code with file tree.
- Commands to run, build, and deploy.
- Automated tests:
- Unit tests for ZIP unpacking, MTL parsing, reference resolution.
- Playwright test for uploading a sample ZIP fixture.
- At least 5 test cases total.
Constraints:
- Everything runs in-browser; no external network fetch of textures.
- Do not generate images and do not mention image generation tools.
中文释义:
让 AI 产出一个 OBJ/MTL 在线校验工具：重点是“资源引用是否齐全、路径是否正确、缺什么怎么修”，并提供可下载的检查报告。
多格式 3D 预览工作台：统一统计面数/材质数/尺寸并导出报告
适合团队内部做素材验收：同一套指标口径对 GLTF/GLB/STL/OBJ 做基础统计与报告导出。
英文 Prompt:
Implement an online 3D preview workbench that supports GLTF/GLB, STL, and OBJ.
Stack:
- Next.js (App Router) + Type .
- Three.js for rendering.
Functional requirements:
- Unified upload area; detect file type and choose loader.
- Show common metrics:
- Bounding box and dimensions
- Mesh count, material count
- Vertex/triangle totals
- Texture count (where applicable)
- Camera/lights presence (for GLTF)
- Export:
- One-click download JSON report with a consistent schema across formats.
- Also export a simple HTML report.
Non-functional:
- Robust error handling per format.
- Performance safeguards for huge meshes (warn and allow cancel).
Deliverables:
- Full project tree and full source.
- Setup + run + build + start commands.
- Vercel deployment instructions.
- Testing:
- Minimum 5 automated tests across: format detection, metrics computation, report schema validation, UI flow.
- Provide a QA checklist.
Do not include any image generation prompts or rendering output generation. Focus on preview and inspection.
中文释义:
让 AI 做一个“多格式统一口径”的 3D 预览与统计工作台：上传即可得到通用指标，并导出 JSON/HTML 报告，便于验收与归档。
GLTF 动画与骨骼检查器：播放、轨道统计与异常提示
适合检查带动画的 GLTF：能否播放、动画轨道是否缺失、骨骼/蒙皮是否异常，并输出检查报告。
英文 Prompt:
Build a GLTF animation & rig inspector web tool.
Tech:
- Next.js + Type + Three.js.
Features:
- Upload GLTF/GLB locally.
- Preview: 3D viewport + animation controls (play/pause, speed, clip selector).
- Inspector panels:
- List animation clips and track counts
- List skinned meshes and skeleton bone counts
- Detect common issues (heuristics):
- Missing animations
- Very large/small scale
- Zero-length clips
- NaN transforms in keyframes if detectable
- Report export:
- Download JSON/HTML report with animations summary and warnings.
Deliverables:
- Full project code , file tree, commands, deployment steps.
- Tests:
- At least 5 tests: animation data extraction, report generator, UI upload and clip switch (Playwright).
- Constraints:
- No server upload; process in browser.
- Do not generate images and do not mention any image generation tools.
中文释义:
让 AI 生成一个“GLTF 动画/骨骼体检”在线工具：能预览并播放动画，同时把轨道/骨骼信息与潜在异常导出为报告。
3D 模型性能风险评估器：Draw Call / 材质拆分建议（预览+报告）
适合做 Web 端上线前评估：通过统计网格、材质、贴图等，给出性能风险提示与优化建议，并留档。
英文 Prompt:
Create an online 3D model performance risk assessor.
Supported formats:
- GLTF/GLB primarily; optionally STL/OBJ with reduced metrics.
Stack:
- Next.js + Type + Three.js.
Metrics & heuristics:
- Estimate draw calls (meshes * materials)
- Count unique materials and textures
- Identify large textures if available (image dimensions after loading)
- Provide warnings thresholds (configurable in UI)
- Provide optimization suggestions (text only): merge meshes, reduce materials, atlas textures, limit texture sizes, LOD hints
Export:
- Download JSON + HTML report.
Deliverables:
- Complete code and file tree.
- Commands and deployment instructions.
- Tests: at least 5 automated tests for metric extraction and report schema.
Constraints:
- No image generation; no poster/renders.
- Keep everything local in browser.
中文释义:
让 AI 输出一个“性能风险评估器”：上传模型后给出 draw call/材质/贴图等风险提示与优化建议，并导出可下载报告用于团队评审。
3D 资产验收清单生成器：自定义规则 + 批量报告打包
适合素材验收流程：按团队自定义规则（阈值/必填字段）做检查，并把报告打包下载归档。
英文 Prompt:
Build a web- d 3D asset acceptance tool with customizable rules and batch reporting.
Tech:
- Next.js + Type + Three.js.
Features:
- Allow uploading multiple GLTF/GLB files in a batch.
- For each file, compute:
- Dimensions, vertex/triangle totals
- Mesh/material counts
- Basic warnings
- Rules engine:
- Users can define thresholds (max triangles, max materials, max textures, etc.)
- Rules are stored in localStorage and can be exported/imported as JSON.
- Batch export:
- Generate per-asset JSON reports and one combined index.json
- Provide a "download all" as a ZIP (client-side) containing reports
Deliverables:
- Provide full code with file tree.
- Provide run/build/deploy instructions.
- Tests:
- At least 5 automated tests: rules evaluation, report aggregation, ZIP packaging, UI batch upload flow.
- Constraints:
- Local-only processing; no server upload.
- Do not generate any images and do not mention image generation tools.
中文释义:
让 AI 做一个“批量验收 + 自定义规则”的在线工具：同一套阈值规则检查多份模型，并把报告打包下载，方便归档与追溯。
3D 预览可访问性与错误体验增强模板：统一空态/错误态/加载态
适合把 3D 工具做得更可用：完善错误提示、加载进度、键盘可达性，并提供可复用的工程模板与测试。
英文 Prompt:
Generate a reusable starter kit for 3D preview/inspection web tools with strong UX.
Stack:
- Next.js + Type + Three.js.
Must include:
- Upload component with drag-drop + keyboard support.
- Loading state with progress (where possible) and cancellation.
- Error boundary and friendly error messages for common loader failures.
- Basic accessibility: focus management, skip , ARIA labels.
- A minimal inspector panel structure (placeholders) and report export scaffolding (JSON/HTML).
Deliverables:
- Full project file tree and complete code.
- Commands and deployment steps.
- Tests:
- Minimum 5 tests: component tests (Vitest), Playwright smoke tests for upload and error flows.
Constraints:
- Keep scope strictly about 3D preview/inspection.
- Do not generate images or mention image generation tooling.
中文释义:
让 AI 生成一个可复用的 3D 工具模板：把上传、加载、错误体验与可访问性都做扎实，并附带测试，后续你可以在此基础上扩展任何 3D 检查功能。
