YouTube 缩略图
五款 1280×720 版式。下方展示的每张图片均由其下方的脚本直接渲染生成,脚本文件位于代码仓库的 examples/ 目录中。
要运行其中任意一个,请安装 uv,克隆代码仓库,并在仓库根目录下执行脚本。脚本会从 assets/ 加载素材并将 PNG 图片写入 examples/,同时覆盖同名旧文件。
git clone https://github.com/sjquant/quickthumb
cd quickthumb
uv run python examples/youtube_thumbnail_01.py
前后对比卡片
![]()
照片背景,左侧为主标题,右侧是两张倾斜卡片,将核心卖点具象化呈现。
- 水平
LinearGradient从接近纯黑过渡到完全透明,使左半部分足够暗以保证文字清晰可读,同时右侧仍能透出背景照片。 - 三行主标题合并为一个
.text()图层,由各自具备独立字号与颜色的TextPart构成。 - 两张卡片是带有
Stroke描边的旋转rectangle形状;前景卡片还投射了Shadow投影。每张卡片上的文字均采用相同的rotation旋转角度。
examples/youtube_thumbnail_01.py
"""
YouTube thumbnail: before / after cards.
- Photo background, darkened on the left by a horizontal gradient
- Three-line headline built from TextParts with different sizes and colors
- Two rotated cards (rectangle + stroke, plus a shadow on the front one)
"""
import os
from quickthumb import Canvas, Filter, LinearGradient, Shadow, Stroke, TextPart
FILE_DIR = os.path.dirname(__file__)
ASSETS_DIR = os.path.join(FILE_DIR, "..", "assets")
os.environ["QUICKTHUMB_FONT_DIR"] = os.path.join(ASSETS_DIR, "fonts")
os.environ["QUICKTHUMB_DEFAULT_FONT"] = "Roboto"
# Create 16:9 YouTube thumbnail (1280x720) with method chaining
(
Canvas.from_aspect_ratio("16:9", 1280)
.background(
image=os.path.join(ASSETS_DIR, "images", "c-g-JgDUVGAXsso-unsplash.jpg"),
effects=[Filter(brightness=0.76, contrast=1.08, saturation=0.5)],
)
# Darken the left side so the headline reads over the photo
.background(
gradient=LinearGradient(
angle=90,
stops=[("#000000FA", 0.0), ("#000000D6", 0.48), ("#00000014", 1.0)],
)
)
.shape(
shape="rectangle",
position=("8%", "13%"),
width=12,
height=12,
color="#0A84FF",
)
.text(
content="BEFORE / AFTER",
size=18,
color="#D2D2D7",
weight=700,
letter_spacing=3,
position=("10.5%", "12.5%"),
)
# Add headline and subtitle as rich text with different sizes and colors
.text(
content=[
TextPart(text="BETTER\n", color="#F5F5F7", size=96),
TextPart(text="THUMBNAILS.\n", color="#F5F5F7", size=78),
TextPart(text="FASTER.", color="#0A84FF", size=96),
],
size=96,
line_height=0.98,
position=("7%", "54%"),
align=("left", "middle"),
bold=True,
)
.shape(
shape="rectangle",
position=(832, 370),
width=260,
height=300,
color="#242426",
border_radius=16,
rotation=-7,
align=("center", "middle"),
effects=[Stroke(width=2, color="#FFFFFF55")],
)
.text(
content="BEFORE\nTOO MUCH\nNO FOCUS",
size=22,
color="#A1A1A6",
weight=700,
line_height=1.7,
letter_spacing=1,
position=(750, 270),
rotation=-7,
)
.shape(
shape="rectangle",
position=(1054, 366),
width=300,
height=340,
color="#0A84FF",
border_radius=18,
rotation=5,
align=("center", "middle"),
effects=[
Stroke(width=2, color="#FFFFFFAA"),
Shadow(offset_x=0, offset_y=18, color="#00000088", blur_radius=24),
],
)
.text(
content="AFTER\nONE IDEA.\nONE FOCUS.",
size=28,
color="#FFFFFF",
weight=800,
line_height=1.55,
position=(958, 248),
rotation=5,
)
# Render the thumbnail
.render(os.path.join(FILE_DIR, "youtube_thumbnail_01.png"))
)
print(f"✓ YouTube thumbnail created: {os.path.join(FILE_DIR, 'youtube_thumbnail_01.png')}")
print(" Replace the background and text content with your own.")
醒目的提问
![]()
在大幅降低饱和度的照片背景上,醒目地放置一个超大字号的问题。
Filter(saturation=0.32)降低了照片的色彩饱和度,使橙色强调色成为全图唯一的鲜艳色彩。- 主标题的每一行均为固定像素位置的独立文本图层,从而能精细控制行间距。
- 引导文案上方的细分割线是一个 1px 高的
#FFFFFF33矩形(20% 不透明度的白色)。
examples/youtube_thumbnail_02.py
"""
YouTube thumbnail: one large question.
- Desaturated photo background with an angled gradient for contrast
- Each headline line is its own text layer at a fixed position
- A 1px translucent rule above the hook line
"""
import os
from quickthumb import Canvas, Filter, FitMode, LinearGradient
FILE_DIR = os.path.dirname(__file__)
ASSETS_DIR = os.path.join(FILE_DIR, "..", "assets")
os.environ["QUICKTHUMB_FONT_DIR"] = os.path.join(ASSETS_DIR, "fonts")
os.environ["QUICKTHUMB_DEFAULT_FONT"] = "Roboto"
# Canvas dimensions
WIDTH = 1280
HEIGHT = 720
(
Canvas(WIDTH, HEIGHT)
# 1. Background Image with darkening
.background(
image=os.path.join(ASSETS_DIR, "images", "denise-jans-WIRvXd1PYlg-unsplash.jpg"),
fit=FitMode.COVER,
effects=[Filter(brightness=0.78, contrast=1.12, saturation=0.32)],
)
# 2. Gradient Overlay (Dark bottom-left to transparent top-right)
# This ensures text readability on the left side
.background(
gradient=LinearGradient(
angle=110,
stops=[("#000000F7", 0.0), ("#000000B8", 0.44), ("#00000012", 1.0)],
),
)
# 3. Typography
# Subtitle: "ARE YOU"
.text(
content="ARE YOU",
size=20,
color="#FF9F0A",
position=(54, 74),
weight=700,
letter_spacing=3,
)
# Headline Line 1: "BURNING"
.text(
content="BURNING",
size=148,
color="#F5F5F7",
position=(46, 108),
weight=800,
)
# Headline Line 2: "OUT?"
.text(
content="OUT?",
size=148,
color="#FF9F0A",
position=(46, 250),
weight=800,
)
# 4. Footer / Hook
.shape(
shape="rectangle",
position=(52, 554),
width=620,
height=1,
color="#FFFFFF33",
)
.text(
content="5 SIGNS YOU'RE IGNORING",
size=32,
color="#F5F5F7",
position=(52, 592),
weight=600,
)
# Render
.render(os.path.join(FILE_DIR, "youtube_thumbnail_02.png"))
)
print(f"✓ Thumbnail created: {os.path.join(FILE_DIR, 'youtube_thumbnail_02.png')}")
出镜人像

左侧为主标题,右侧是演讲者人像。
- 人像通过 URL 下载,并使用
remove_background=True完成自动抠图。这需要安装rembg额外依赖(支持 Python 3.11+)和网络连接:uv run --extra rembg python examples/youtube_talking_head.py。 position=(1280, 720)配合align=("right", "bottom"),无论人像素材实际尺寸多大,都能将其固定在画布右下角。max_width="55%"确保主标题在触及人像之前自动折行。
examples/youtube_talking_head.py
"""
YouTube Talking-Head Thumbnail
Classic split layout with subject on the right and headline on the left:
- Black background with a blue panel on the right
- Portrait from a URL, cut out with remove_background=True (needs quickthumb[rembg])
and pinned to the bottom-right corner
- Two-tone headline built from TextParts, wrapped with max_width
"""
import os
from quickthumb import Canvas, FitMode, LinearGradient, Shadow, Stroke, TextPart
FILE_DIR = os.path.dirname(__file__)
ASSETS_DIR = os.path.join(FILE_DIR, "..", "assets")
OUTPUT_PATH = os.path.join(FILE_DIR, "youtube_talking_head.png")
os.environ["QUICKTHUMB_FONT_DIR"] = os.path.join(ASSETS_DIR, "fonts")
os.environ["QUICKTHUMB_DEFAULT_FONT"] = "Roboto"
# Replace with your subject photo (local path or URL)
PORTRAIT_URL = (
"https://images.unsplash.com/photo-1560250097-0b93528c311a?auto=format&fit=crop&w=900&q=80"
)
(
Canvas(1280, 720)
.background(color="#000000")
.background(
gradient=LinearGradient(
angle=90,
stops=[("#000000", 0.0), ("#000000E8", 0.48), ("#00000011", 1.0)],
)
)
.shape(
shape="rectangle",
position=(840, 0),
width=440,
height=720,
color="#0A84FF",
)
.shape(
shape="rectangle",
position=(52, 132),
width=56,
height=5,
color="#0A84FF",
border_radius=3,
)
.text(
content="30-DAY REVIEW",
size=18,
color="#A1A1A6",
weight=900,
letter_spacing=1,
position=(52, 64),
)
# Main headline: stacked rich text with accent highlight
.text(
content=[
TextPart(text="AI TOOLS.\n", color="#FFFFFF", weight=800),
TextPart(text="WHAT ACTUALLY\nWORKS.", color="#0A84FF", weight=900),
],
size=92,
line_height=0.92,
position=(48, 142),
max_width="55%",
effects=[
Stroke(width=1, color="#000000"),
Shadow(offset_x=0, offset_y=8, color="#00000099", blur_radius=14),
],
)
# Sub-copy anchored near the bottom
.text(
content="24 TESTED. 3 KEPT.",
size=32,
color="#D4D4D4",
weight=500,
position=(52, 620),
effects=[Shadow(offset_x=2, offset_y=2, color="#000000", blur_radius=6)],
)
# Subject portrait — bottom-anchored on the right side
.image(
path=PORTRAIT_URL,
position=(1280, 720),
width=570,
height=710,
fit=FitMode.COVER,
align=("right", "bottom"),
remove_background=True,
effects=[Shadow(offset_x=-14, offset_y=0, color="#00000088", blur_radius=22)],
)
.render(OUTPUT_PATH)
)
print(f"✓ Talking-head thumbnail created: {OUTPUT_PATH}")
print(" Replace PORTRAIT_URL with your subject photo for a personalised layout.")
反应 / 解说

醒目的关键词、巨大的数字与充满情绪的面部表情。
- 居中置于人像后方的
RadialGradient径向渐变增添了微弱的红光光晕。顶层的LinearGradient则保持文本一侧的深色背景。 - 底层作为大背景的
#1是一个 470px、不透明度为 13% 的超大文本图层。 - 人像使用了
remove_background=True(需要rembg额外依赖)与Filter(saturation=0.08),几乎呈现为黑白质感。
examples/youtube_reaction.py
"""
YouTube Reaction / Commentary Thumbnail
- Dark base with a dim red radial glow behind the subject
- Cut-out portrait (remove_background=True, needs quickthumb[rembg]),
almost fully desaturated
- A huge "#1" at low opacity behind everything
- "VIRAL" headline and a view-count line built from TextParts
"""
import os
from quickthumb import Canvas, Filter, FitMode, LinearGradient, RadialGradient, Shadow, TextPart
FILE_DIR = os.path.dirname(__file__)
ASSETS_DIR = os.path.join(FILE_DIR, "..", "assets")
OUTPUT_PATH = os.path.join(FILE_DIR, "youtube_reaction.png")
os.environ["QUICKTHUMB_FONT_DIR"] = os.path.join(ASSETS_DIR, "fonts")
os.environ["QUICKTHUMB_DEFAULT_FONT"] = "Roboto"
PORTRAIT_PATH = os.path.join(ASSETS_DIR, "images", "reaction-surprised.jpg")
(
Canvas(1280, 720)
.background(color="#000000")
.background(
gradient=RadialGradient(
center=(0.82, 0.48),
stops=[("#3D1110", 0.0), ("#000000", 0.68)],
)
)
.background(
gradient=LinearGradient(
angle=90,
stops=[("#000000", 0.0), ("#000000E8", 0.46), ("#00000011", 1.0)],
)
)
.text(
content="#1",
size=470,
color="#FF453A",
opacity=0.13,
weight=900,
letter_spacing=-18,
position=(1210, 28),
align=("right", "top"),
)
.shape(
shape="ellipse",
position=(1034, 372),
width=548,
height=548,
color="#1C1C1E",
align=("center", "middle"),
)
.image(
path=PORTRAIT_PATH,
position=(1280, 720),
width=620,
height=710,
fit=FitMode.COVER,
align=("right", "bottom"),
remove_background=True,
effects=[
Filter(contrast=1.08, saturation=0.08),
Shadow(offset_x=-12, offset_y=12, color="#00000099", blur_radius=26),
],
)
.text(
content="48H SIGNAL",
size=17,
color="#A1A1A6",
weight=700,
letter_spacing=3,
position=(64, 54),
)
.shape(
shape="rectangle",
position=(64, 92),
width=610,
height=1,
color="#FFFFFF33",
)
.text(
content="VIRAL",
size=212,
color="#F5F5F7",
weight=900,
letter_spacing=-5,
position=(52, 104),
)
.text(
content=[
TextPart(text="12.4M", color="#F5F5F7", weight=700),
TextPart(text=" VIEWS", color="#A1A1A6", size=24, weight=600),
],
size=92,
color="#F5F5F7",
weight=700,
letter_spacing=-2,
position=(60, 392),
)
.text(
content="WHY IT WORKED — AND WHAT COMES NEXT",
size=19,
color="#A1A1A6",
weight=600,
letter_spacing=2,
position=(64, 650),
)
.shape(
shape="rectangle",
position=(64, 620),
width=610,
height=1,
color="#FFFFFF33",
)
.render(OUTPUT_PATH)
)
print(f"✓ Reaction / commentary thumbnail created: {OUTPUT_PATH}")
教程 / 解说

无需任何摄影照片:主标题、蓝色面板和几行“代码”。
- 面板上的代码为一个统一的文本图层。每个
TextPart设置各自的颜色与字重,轻松模拟出代码语法高亮的效果。 - 配色方案抽离为模块级常量(
BLUE、WHITE、GRAY),更换缩略图风格只需修改三行定义。
examples/youtube_tutorial_explainer.py
"""
YouTube Tutorial / Explainer Thumbnail
- Black background with a blue panel on the right; no photos
- Two-tone headline built from TextParts
- A code snippet on the panel, "highlighted" by giving each TextPart
its own color and weight
"""
import os
from quickthumb import Canvas, TextPart
FILE_DIR = os.path.dirname(__file__)
ASSETS_DIR = os.path.join(FILE_DIR, "..", "assets")
OUTPUT_PATH = os.path.join(FILE_DIR, "youtube_tutorial_explainer.png")
os.environ["QUICKTHUMB_FONT_DIR"] = os.path.join(ASSETS_DIR, "fonts")
os.environ["QUICKTHUMB_DEFAULT_FONT"] = "Roboto"
BLUE = "#0A84FF"
WHITE = "#F5F5F7"
GRAY = "#86868B"
def main():
"""Render the tutorial thumbnail."""
canvas = (
Canvas(1280, 720)
.background(color="#000000")
.shape(
shape="rectangle",
position=(704, 0),
width=576,
height=720,
color=BLUE,
)
.shape(
shape="rectangle",
position=(56, 58),
width=10,
height=10,
color=BLUE,
)
.text(
content="HOW TO / PYTHON",
size=17,
color="#A1A1A6",
weight=700,
letter_spacing=3,
position=(88, 52),
)
.text(
content="PYTHON / RUN",
size=16,
color="#FFFFFFB8",
weight=700,
letter_spacing=2,
position=(1224, 52),
align=("right", "top"),
)
.shape(
shape="rectangle",
position=(56, 96),
width=1168,
height=1,
color="#FFFFFF33",
)
.text(
content=[
TextPart(text="MASTER\n", color=WHITE, weight=700),
TextPart(text="PYTHON.", color=BLUE, weight=800),
],
size=148,
line_height=0.88,
letter_spacing=-4,
position=(50, 136),
)
.text(
content="in 30 days — from scratch",
size=27,
color=GRAY,
weight=500,
position=(58, 474),
)
.text(
content="BUILD / RUN / SHIP",
size=15,
color="#A1A1A6",
weight=700,
letter_spacing=3,
position=(60, 646),
)
.text(
content=[
TextPart(text="def ", color="#FFFFFFB8", weight=500),
TextPart(text="ship", color="#1D1D1F", weight=800),
TextPart(text="(idea):\n", color="#FFFFFF", weight=500),
TextPart(text=" build(idea)\n", color="#FFFFFF", weight=500),
TextPart(text=" return ", color="#FFFFFFB8", weight=500),
TextPart(text='"LIVE"', color="#1D1D1F", weight=800),
],
size=42,
line_height=1.55,
position=(744, 168),
)
.shape(
shape="rectangle",
position=(744, 474),
width=480,
height=1,
color="#FFFFFF66",
)
.text(
content=">>> SHIPPED / LIVE",
size=36,
color="#FFFFFF",
weight=700,
letter_spacing=1,
position=(744, 510),
)
)
canvas.render(OUTPUT_PATH)
print(f"✓ Tutorial / explainer thumbnail created: {OUTPUT_PATH}")
print(" Swap the step labels and headline text for any how-to topic.")
if __name__ == "__main__":
main()