create-3d-scene
关于
This Claude Skill automates Blender scene creation using Python's bpy API to programmatically generate 3D scenes with objects, materials, lighting, and cameras. It's designed for developers needing reproducible visualizations, automated rendering setups, or batch scene generation. Use it to integrate 3D visualization into data pipelines or create template scenes for workflow automation.
快速安装
Claude Code
推荐npx skills add pjt222/agent-almanac -a claude-code/plugin add https://github.com/pjt222/agent-almanacgit clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/create-3d-scene在 Claude Code 中复制并粘贴此命令以安装该技能
技能文档
造三維場景
以 Python API(bpy)程式化立一完整 Blender 場景。配置場景階層、加網格物件、以節點式著色器造 PBR 材質、置光與相機、設環境/世界之設。
適用時機
- 由頭造可重現之三維可視化場景
- 自動化產品視覺化或建築渲染之設
- 以程式生多場景變體
- 為批次渲染流建範本場景
- 手動精修前之場景佈局原型
- 整合三維可視化於資料管線或報告系統
輸入
| 輸入 | 類型 | 描述 | 範例 |
|---|---|---|---|
| 場景規格 | 配置 | 物件、材質、光之需 | 產品尺寸、材色、光設 |
| 輸出需求 | 參數 | 解析度、渲染引擎、品質設 | 1920x1080、Cycles、128 取樣 |
| 資產路 | 檔路 | 外部模型、貼圖、HDRI | /path/to/hdri.exr、product_model.obj |
| 相機設 | 參數 | 位、旋、焦距、景深 | location=(7,-7,5)、lens=50mm |
| 環境 | 配置 | 世界著色器、背景、環境光設 | HDRI 光、純色、漸層 |
步驟
1. 立腳本結構
造一 Python 腳本,含當之匯入與結構:
#!/usr/bin/env python3
"""
Scene setup script for Blender.
Usage: blender --background --python setup_scene.py
"""
import bpy
import math
import os
from pathlib import Path
def clear_scene():
"""Remove all objects from the scene."""
bpy.ops.object.select_all(action='SELECT')
bpy.ops.object.delete(use_global=False)
# Clear orphaned data
for block in bpy.data.meshes:
if block.users == 0:
bpy.data.meshes.remove(block)
for block in bpy.data.materials:
if block.users == 0:
bpy.data.materials.remove(block)
def main():
clear_scene()
# Scene setup steps follow
if __name__ == "__main__":
main()
預期: 腳本結構含 clear_scene() 與 main() 函式 失敗時: 察 Python 語法,驗 bpy 匯入於 Blender Python 環境中可行
2. 加網格物件
造原始或匯入之網格物件:
def add_objects():
"""Add mesh objects to scene."""
# Add cube
bpy.ops.mesh.primitive_cube_add(
size=2.0,
location=(0, 0, 1)
)
cube = bpy.context.active_object
cube.name = "Product_Base"
# Add sphere
bpy.ops.mesh.primitive_uv_sphere_add(
radius=1.0,
segments=32,
ring_count=16,
location=(3, 0, 1)
)
sphere = bpy.context.active_object
sphere.name = "Detail_Sphere"
# Import external model (optional)
# bpy.ops.import_scene.obj(filepath="model.obj")
return cube, sphere
預期: 物件顯於場景,名與位皆合 失敗時: 察運算子語法,驗座標,確保無命名衝突
3. 以節點式著色器造材質
以著色器節點立 PBR 材質:
def create_material(name, base_color, metallic=0.0, roughness=0.5):
"""Create a PBR material with node setup."""
# Create material
mat = bpy.data.materials.new(name=name)
mat.use_nodes = True
nodes = mat.node_tree.nodes
links = mat.node_tree.links
# Clear default nodes
nodes.clear()
# Add Principled BSDF
node_bsdf = nodes.new(type='ShaderNodeBsdfPrincipled')
node_bsdf.location = (0, 0)
node_bsdf.inputs['Base Color'].default_value = base_color + (1.0,) # Add alpha
node_bsdf.inputs['Metallic'].default_value = metallic
node_bsdf.inputs['Roughness'].default_value = roughness
# Add Material Output
node_output = nodes.new(type='ShaderNodeOutputMaterial')
node_output.location = (300, 0)
# Link nodes
links.new(node_bsdf.outputs['BSDF'], node_output.inputs['Surface'])
return mat
def apply_materials(cube, sphere):
"""Apply materials to objects."""
# Create materials
mat_red = create_material("RedPlastic", (0.8, 0.1, 0.1), metallic=0.0, roughness=0.4)
mat_metal = create_material("Metal", (0.8, 0.8, 0.8), metallic=1.0, roughness=0.2)
# Assign to objects
if cube.data.materials:
cube.data.materials[0] = mat_red
else:
cube.data.materials.append(mat_red)
if sphere.data.materials:
sphere.data.materials[0] = mat_metal
else:
sphere.data.materials.append(mat_metal)
預期: 材質於著色器編輯器中顯,節點連接無誤 失敗時: 察節點型存否,驗連結語法,確保色值於 [0,1] 區間
4. 立光
配光以照場景:
def setup_lighting():
"""Add lights to scene."""
# Sun light
bpy.ops.object.light_add(
type='SUN',
location=(5, 5, 10)
)
sun = bpy.context.active_object
sun.name = "KeyLight"
sun.data.energy = 3.0
sun.rotation_euler = (math.radians(45), 0, math.radians(45))
# Area light (fill light)
bpy.ops.object.light_add(
type='AREA',
location=(-4, -4, 6)
)
area = bpy.context.active_object
area.name = "FillLight"
area.data.energy = 200.0
area.data.size = 5.0
area.rotation_euler = (math.radians(60), 0, math.radians(-135))
# Point light (rim light)
bpy.ops.object.light_add(
type='POINT',
location=(2, -5, 3)
)
point = bpy.context.active_object
point.name = "RimLight"
point.data.energy = 500.0
預期: 三光,強度與位皆當 失敗時: 依渲染引擎(Cycles 與 EEVEE 之別)調能量值,察旋轉格式
5. 置相機
立相機,框取得宜:
def setup_camera():
"""Add and configure camera."""
bpy.ops.object.camera_add(
location=(7, -7, 5)
)
camera = bpy.context.active_object
camera.name = "MainCamera"
# Point camera at origin
direction = (0, 0, 1) - camera.location
rot_quat = direction.to_track_quat('-Z', 'Y')
camera.rotation_euler = rot_quat.to_euler()
# Camera settings
camera.data.lens = 50 # Focal length in mm
camera.data.dof.use_dof = True
camera.data.dof.focus_distance = 10.0
camera.data.dof.aperture_fstop = 2.8
# Set as active camera
bpy.context.scene.camera = camera
預期: 相機已置,焦距與景深設皆當 失敗時: track_to 敗則用簡旋轉法,驗鏡頭單位
6. 配世界環境
立世界著色器與背景:
def setup_world():
"""Configure world environment."""
world = bpy.data.worlds['World']
world.use_nodes = True
nodes = world.node_tree.nodes
links = world.node_tree.links
# Clear default nodes
nodes.clear()
# Add Environment Texture (for HDRI)
node_env = nodes.new(type='ShaderNodeTexEnvironment')
node_env.location = (-300, 0)
# Load HDRI if available
hdri_path = "/path/to/hdri.exr"
if os.path.exists(hdri_path):
node_env.image = bpy.data.images.load(hdri_path)
# Add Background shader
node_bg = nodes.new(type='ShaderNodeBackground')
node_bg.location = (0, 0)
node_bg.inputs['Strength'].default_value = 1.0
# Add World Output
node_output = nodes.new(type='ShaderNodeOutputWorld')
node_output.location = (300, 0)
# Link nodes
links.new(node_env.outputs['Color'], node_bg.inputs['Color'])
links.new(node_bg.outputs['Background'], node_output.inputs['Surface'])
預期: 世界著色器已配 HDRI 或純背景 失敗時: 檔缺則跳 HDRI 載入,單用 Background 節點加色
7. 配渲染設
設基本渲染參數:
def setup_render_settings():
"""Configure render settings."""
scene = bpy.context.scene
# Render engine
scene.render.engine = 'CYCLES' # or 'BLENDER_EEVEE'
scene.cycles.samples = 128
scene.cycles.use_denoising = True
# Output settings
scene.render.resolution_x = 1920
scene.render.resolution_y = 1080
scene.render.resolution_percentage = 100
# File format
scene.render.image_settings.file_format = 'PNG'
scene.render.image_settings.color_mode = 'RGBA'
scene.render.image_settings.color_depth = '16'
scene.render.filepath = "/tmp/render_"
預期: 渲染設已配,可備渲染 失敗時: 察引擎名拼法,驗解析度為正整數
8. 組織場景階層
造集合以便組織:
def organize_collections():
"""Organize objects into collections."""
# Create collections
col_geometry = bpy.data.collections.new("Geometry")
col_lights = bpy.data.collections.new("Lights")
col_cameras = bpy.data.collections.new("Cameras")
# Link to scene
bpy.context.scene.collection.children.link(col_geometry)
bpy.context.scene.collection.children.link(col_lights)
bpy.context.scene.collection.children.link(col_cameras)
# Move objects to collections
for obj in bpy.data.objects:
# Unlink from main collection
bpy.context.scene.collection.objects.unlink(obj)
# Link to appropriate collection
if obj.type == 'MESH':
col_geometry.objects.link(obj)
elif obj.type == 'LIGHT':
col_lights.objects.link(obj)
elif obj.type == 'CAMERA':
col_cameras.objects.link(obj)
預期: 物件組織於名集合中,便於管理 失敗時: 造前察集合已存否,處孤立物件
驗證清單
- 腳本於 Blender 背景模式無誤運行
- 所期物件皆顯於場景大綱
- 材質於著色器編輯器中顯正色與性
- 相機已置,物件於框中
- 光給充足之照(測渲染)
- 世界環境載入無誤(HDRI 或背景色)
- 渲染設合輸出需求
- 場景於集合中組織得當
- 無孤立資料塊(無使用者之材質、網格)
- 腳本含 clear_scene() 以便重現
常見陷阱
- 物件命名衝突:用唯一名,造前察既存物件
- 色格式誤:RGB 值須為元組 (r, g, b, a) 於 [0,1] 區間
- 缺 alpha 通道:設色時含 alpha:
(r, g, b, 1.0) - 節點連接誤:連結前驗節點型有所期之輸入/輸出
- 相機未激活:須設
bpy.context.scene.camera = camera_object - 相對與絕對路之別:用絕對路或 Path() 以跨平台相容
- 單位之惑:Blender 預設以公尺為單位,相機鏡頭以毫米
- 旋轉格式:用
math.radians()行度-弧度之轉換 - 渲染引擎之別:EEVEE 與 Cycles 有異之功能與參數
- 記憶體洩漏:清孤立資料塊以防批次操作中之記憶體累積
相關技能
- script-blender-automation:程序化建模與批次操作之進階腳本模式
- render-blender-output:配渲染管線並執渲染
- create-2d-composition:以類似腳本法行二維圖形構圖
GitHub 仓库
相关推荐技能
content-collections
元Content Collections 是一个 TypeScript 优先的构建工具,可将本地 Markdown/MDX 文件转换为类型安全的数据集合。它专为构建博客、文档站和内容密集型 Vite+React 应用而设计,提供基于 Zod 的自动模式验证。该工具涵盖从 Vite 插件配置、MDX 编译到生产环境部署的完整工作流。
polymarket
元这个Claude Skill为开发者提供完整的Polymarket预测市场开发支持,涵盖API调用、交易执行和市场数据分析。关键特性包括实时WebSocket数据流,可监控实时交易、订单和市场动态。开发者可用它构建预测市场应用、实施交易策略并集成实时市场预测功能。
creating-opencode-plugins
元该Skill帮助开发者创建OpenCode插件,用于接入命令、文件、LSP等25+种事件。它提供了插件结构、事件API规范和JavaScript/TypeScript实现模式,适合需要拦截操作、扩展功能或自定义事件处理的场景。开发者可通过它快速构建响应式模块来增强OpenCode AI助手的能力。
sglang
元SGLang是一个专为LLM设计的高性能推理框架,特别适用于需要结构化输出的场景。它通过RadixAttention前缀缓存技术,在处理JSON、正则表达式、工具调用等具有重复前缀的复杂工作流时,能实现极速生成。如果你正在构建智能体或多轮对话系统,并追求远超vLLM的推理性能,SGLang是理想选择。
