create-2d-composition
关于
This skill enables programmatic 2D graphics creation using SVG generation, layout algorithms, and image compositing. It's ideal for developers building custom diagrams, reproducible figures, or automating visual asset production. Use it for batch processing workflows and creating chart types beyond standard libraries.
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-2d-composition在 Claude Code 中复制并粘贴此命令以安装该技能
技能文档
造二維構圖
以程式生二維圖形:SVG 構建、圖表佈局演算法、影像合成、批次處理流。涵向量圖形生成、點陣圖像處理、字體排印,與圖表、示意圖、資訊圖之自動化產出。
適用時機
- 以程式生示意圖、流程圖、資訊圖
- 造可重現之科學圖或出版圖
- 自動化產出徽章、圖示、視覺資產
- 合成多圖像或資料可視化
- 建標準庫所無之自訂圖表型
- 以參數變動批次生圖
- 為網頁或印刷造 SVG 範本
輸入
| 輸入 | 類型 | 描述 | 範例 |
|---|---|---|---|
| 佈局規格 | 配置 | 尺寸、邊距、網格佈局 | 畫布 800x600px、20px 邊距 |
| 視覺元素 | 資料/資產 | 形狀、文字、圖像、資料點 | 矩形座標、標籤、圖示 |
| 樣式參數 | CSS/屬性 | 色、字體、筆劃寬、透明度 | fill="#3366cc"、stroke-width="2" |
| 資料源 | 檔案/陣列 | 待可視化或標註之值 | CSV 資料、JSON 配置 |
| 輸出格式 | 字串 | SVG、PNG、PDF、合成格式 | output.svg、300 DPI PNG |
步驟
1. 立 Python 環境
裝二維構圖所需之庫:
# Core libraries
pip install svgwrite pillow cairosvg
# Optional: advanced features
pip install drawsvg reportlab pycairo
# For data-driven graphics
pip install matplotlib numpy pandas
預期: 庫裝成 失敗時: 察 Python 版本(3.7+),用虛擬環境以避衝突
2. 造基本 SVG 圖形
以 svgwrite 生 SVG:
import svgwrite
from svgwrite import cm, mm
def create_basic_svg(output_path):
"""Create a simple SVG graphic."""
# Initialize drawing (use mm for precise dimensions)
dwg = svgwrite.Drawing(output_path, size=('180mm', '120mm'), profile='full')
# Add background rectangle
dwg.add(dwg.rect(
insert=(0, 0),
size=('100%', '100%'),
fill='white'
))
# Add shapes
dwg.add(dwg.circle(
center=(90*mm, 60*mm),
r=30*mm,
fill='lightblue',
stroke='navy',
stroke_width=2
))
dwg.add(dwg.rect(
insert=(30*mm, 30*mm),
size=(60*mm, 40*mm),
fill='lightgreen',
stroke='darkgreen',
stroke_width=2,
rx=5, # Rounded corners
ry=5
))
# Add text
dwg.add(dwg.text(
'Example Graphic',
insert=(90*mm, 20*mm),
text_anchor='middle',
font_size='18pt',
font_family='Arial',
fill='black'
))
dwg.save()
print(f"Saved: {output_path}")
預期: SVG 檔已生,含形狀與文字 失敗時: 察 svgwrite 版本,驗輸出目錄可寫
3. 以佈局邏輯建示意圖
以算出之位造結構化示意圖:
def create_flowchart(steps, output_path):
"""Generate a flowchart from list of steps."""
dwg = svgwrite.Drawing(output_path, size=('800px', '600px'))
# Layout parameters
box_width = 120
box_height = 60
spacing_y = 100
start_x = 340
start_y = 50
for i, step in enumerate(steps):
y_pos = start_y + i * spacing_y
# Draw box
box = dwg.add(dwg.g(id=f'step_{i}'))
box.add(dwg.rect(
insert=(start_x, y_pos),
size=(box_width, box_height),
fill='lightblue',
stroke='navy',
stroke_width=2,
rx=5,
ry=5
))
# Add text (wrapped if needed)
text_lines = wrap_text(step, max_width=16)
text_y = y_pos + box_height/2 - (len(text_lines)-1) * 7
for j, line in enumerate(text_lines):
box.add(dwg.text(
line,
insert=(start_x + box_width/2, text_y + j*14),
text_anchor='middle',
font_size='12pt',
font_family='Arial',
fill='black'
))
# Draw arrow to next step
if i < len(steps) - 1:
arrow_start_y = y_pos + box_height
arrow_end_y = y_pos + spacing_y
dwg.add(dwg.line(
start=(start_x + box_width/2, arrow_start_y),
end=(start_x + box_width/2, arrow_end_y),
stroke='black',
stroke_width=2,
marker_end=dwg.marker(
id='arrow',
viewBox='0 0 10 10',
refX=5,
refY=5,
markerWidth=6,
markerHeight=6,
orient='auto'
)
))
dwg.save()
def wrap_text(text, max_width=20):
"""Simple text wrapping."""
words = text.split()
lines = []
current_line = []
for word in words:
test_line = ' '.join(current_line + [word])
if len(test_line) <= max_width:
current_line.append(word)
else:
if current_line:
lines.append(' '.join(current_line))
current_line = [word]
if current_line:
lines.append(' '.join(current_line))
return lines
預期: 流程圖已生,含相連之方塊與箭頭 失敗時: 調佈局計算,驗箭頭標記之定義
4. 合成點陣圖像
以 Pillow 合多圖:
from PIL import Image, ImageDraw, ImageFont, ImageFilter
import os
def composite_images(image_paths, output_path, layout='grid'):
"""Composite multiple images into single output."""
# Load images
images = [Image.open(path) for path in image_paths]
if layout == 'grid':
# Calculate grid dimensions
n = len(images)
cols = int(n ** 0.5)
rows = (n + cols - 1) // cols
# Get max dimensions
max_width = max(img.width for img in images)
max_height = max(img.height for img in images)
# Create composite canvas
canvas_width = cols * max_width
canvas_height = rows * max_height
composite = Image.new('RGB', (canvas_width, canvas_height), 'white')
# Paste images
for i, img in enumerate(images):
row = i // cols
col = i % cols
x = col * max_width
y = row * max_height
composite.paste(img, (x, y))
elif layout == 'horizontal':
# Horizontal concatenation
total_width = sum(img.width for img in images)
max_height = max(img.height for img in images)
composite = Image.new('RGB', (total_width, max_height), 'white')
x_offset = 0
for img in images:
composite.paste(img, (x_offset, 0))
x_offset += img.width
elif layout == 'vertical':
# Vertical concatenation
max_width = max(img.width for img in images)
total_height = sum(img.height for img in images)
composite = Image.new('RGB', (max_width, total_height), 'white')
y_offset = 0
for img in images:
composite.paste(img, (0, y_offset))
y_offset += img.height
composite.save(output_path)
print(f"Saved composite: {output_path}")
def add_annotations(image_path, annotations, output_path):
"""Add text annotations to image."""
img = Image.open(image_path)
draw = ImageDraw.Draw(img)
# Load font
try:
font = ImageFont.truetype("Arial.ttf", 24)
except:
font = ImageFont.load_default()
for annotation in annotations:
text = annotation['text']
position = annotation['position']
color = annotation.get('color', 'black')
# Add text shadow for readability
shadow_offset = 2
draw.text(
(position[0] + shadow_offset, position[1] + shadow_offset),
text,
font=font,
fill='white'
)
draw.text(position, text, font=font, fill=color)
img.save(output_path)
預期: 合成圖已生,佈局得宜 失敗時: 察輸入圖皆在、驗圖像模式相容
5. 生資料驅動之圖形
自資料造可視化:
import numpy as np
def create_bar_chart_svg(data, labels, output_path):
"""Generate SVG bar chart from data."""
dwg = svgwrite.Drawing(output_path, size=('600px', '400px'))
# Chart area
margin = 50
chart_width = 500
chart_height = 300
bar_spacing = 10
# Calculate bar dimensions
n_bars = len(data)
bar_width = (chart_width - (n_bars - 1) * bar_spacing) / n_bars
# Scale data to fit chart
max_value = max(data)
scale = chart_height / max_value
# Draw axes
dwg.add(dwg.line(
start=(margin, margin),
end=(margin, margin + chart_height),
stroke='black',
stroke_width=2
))
dwg.add(dwg.line(
start=(margin, margin + chart_height),
end=(margin + chart_width, margin + chart_height),
stroke='black',
stroke_width=2
))
# Draw bars
for i, (value, label) in enumerate(zip(data, labels)):
x = margin + i * (bar_width + bar_spacing)
bar_height = value * scale
y = margin + chart_height - bar_height
# Bar
dwg.add(dwg.rect(
insert=(x, y),
size=(bar_width, bar_height),
fill='steelblue',
stroke='navy',
stroke_width=1
))
# Value label
dwg.add(dwg.text(
f'{value:.1f}',
insert=(x + bar_width/2, y - 5),
text_anchor='middle',
font_size='10pt',
fill='black'
))
# X-axis label
dwg.add(dwg.text(
label,
insert=(x + bar_width/2, margin + chart_height + 20),
text_anchor='middle',
font_size='10pt',
fill='black'
))
dwg.save()
預期: SVG 長條圖已生,資料已縮放 失敗時: 處邊界案(空資料、負值),加驗證
6. 批次生圖
自動化多圖之造:
def batch_generate_badges(users, template_path, output_dir):
"""Generate badge for each user."""
os.makedirs(output_dir, exist_ok=True)
for user in users:
output_path = os.path.join(output_dir, f"{user['id']}_badge.svg")
dwg = svgwrite.Drawing(output_path, size=('300px', '100px'))
# Background
dwg.add(dwg.rect(
insert=(0, 0),
size=('100%', '100%'),
fill='#3366cc',
rx=10,
ry=10
))
# User name
dwg.add(dwg.text(
user['name'],
insert=(150, 40),
text_anchor='middle',
font_size='20pt',
font_weight='bold',
fill='white'
))
# User role
dwg.add(dwg.text(
user['role'],
insert=(150, 70),
text_anchor='middle',
font_size='14pt',
fill='lightblue'
))
dwg.save()
print(f"Generated badge: {output_path}")
預期: 各資料項皆生獨立圖形 失敗時: 察資料結構,缺欄以預設值處之
7. SVG 轉點陣
匯 SVG 為 PNG/PDF 以供多用:
import cairosvg
def svg_to_png(svg_path, png_path, dpi=300):
"""Convert SVG to PNG with specified DPI."""
# Calculate pixel dimensions from DPI
# Assuming A4 size as example
width_inches = 8.27
height_inches = 11.69
width_px = int(width_inches * dpi)
height_px = int(height_inches * dpi)
cairosvg.svg2png(
url=svg_path,
write_to=png_path,
output_width=width_px,
output_height=height_px
)
print(f"Converted to PNG: {png_path}")
def svg_to_pdf(svg_path, pdf_path):
"""Convert SVG to PDF."""
cairosvg.svg2pdf(url=svg_path, write_to=pdf_path)
print(f"Converted to PDF: {pdf_path}")
預期: 點陣輸出已生於指定解析度 失敗時: 若缺則裝 cairo 系統庫,察 SVG 之有效性
驗證清單
- 圖形於目標應用中渲染無誤
- 文字可讀且位置得宜
- 色合規格(察十六進制碼)
- 尺寸合用例
- SVG 合標準(若需)
- 點陣匯出 DPI 正確
- 佈局適應資料變動
- 批次處理完而無誤
- 輸出檔分組得宜
- 程式含誤處理
常見陷阱
- 單位之惑:SVG 單位(px、mm、cm)與螢幕像素、印刷 DPI 之別
- 文字溢:文字超形狀邊界;宜施換行
- 字體可得:系統字體或異,宜內嵌或用網安之字體
- 座標計:網格佈局中之差一錯
- 色格式:SVG 用十六進制字串(
#rrggbb),非元組 - SVG 有效性:察 XML 結構,閉所有標籤
- 檔路:處特殊字元與檔名中之空
- 記憶體用:大批次操作或需分塊
- 比例:改尺寸時保比
- 透明:PNG 支援 alpha,JPEG 則否
相關技能
- render-publication-graphic:出版專屬之輸出需求
- create-3d-scene:三維之類似程式化法
- generate-quarto-report:整合圖形於報告中
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。
