概述

Particle Logo Convergence

Generate a 1080p particle animation video from any logo or image. Pure local processing — Python + FFmpeg, no API key needed.

When to Use

Trigger this skill when the user wants to:

Create a particle animation or logo reveal video
Turn a logo / image into a motion graphics clip
Generate a short video with synthesized music from a static image

Requirements

FFmpeg must be installed (brew install ffmpeg)
Python deps: numpy, Pillow (installed automatically via uv run --with)

Workflow

Step 1: Confirm inputs

Ask the user for (if not already provided):

Input image — any PNG/JPG/WEBP; transparent-background PNGs work best
Output path — e.g. output.mp4
Size (optional) — defaults to auto-detect from image aspect ratio
Duration (optional) — defaults to 6 seconds
Particles (optional) — defaults to 10,000
Background color (optional) — defaults to auto-detect

Step 2: Run the script

uv run --with numpy --with Pillow python3 <skill-path>/scripts/particle_logo.py <input_image> <output.mp4> [--size WxH] [--duration N] [--particles N] [--fps 30|60] [--bg COLOR] [--no-audio]

The script path is scripts/particle_logo.py relative to this SKILL.md.

Parameters:

Parameter	Default	Description
-----------	---------	-------------
`--size WxH`	auto	Video size (e.g. `1920x1080`, `1080x1920`, `1080`)
`--duration N`	6.0	Duration in seconds (range 3-30)
`--particles N`	10000	Particle count (range 1000-50000). More = finer detail but slower
`--fps 30\	60`	60	Frame rate. 30 = smaller file, 60 = smoother
`--bg COLOR`	auto	Background color: `auto` / `white` / `black` / `#RRGGBB`
`--no-audio`	off	Skip audio synthesis, output video only

Examples:

# Auto size (preserves image aspect ratio)
uv run --with numpy --with Pillow python3 <skill-path>/scripts/particle_logo.py logo.png output.mp4

# Force specific size
uv run --with numpy --with Pillow python3 <skill-path>/scripts/particle_logo.py logo.png output.mp4 --size 1920x1080

# Custom duration and particles
uv run --with numpy --with Pillow python3 <skill-path>/scripts/particle_logo.py logo.png output.mp4 --duration 10 --particles 20000

# Fast preview (fewer particles, lower fps, no audio)
uv run --with numpy --with Pillow python3 <skill-path>/scripts/particle_logo.py logo.png output.mp4 --particles 3000 --fps 30 --no-audio

# Black background, portrait video
uv run --with numpy --with Pillow python3 <skill-path>/scripts/particle_logo.py logo.png output.mp4 --bg black --size 1080x1920

Step 3: Verify output

Check that:

The output file exists and is > 1 MB (a near-zero file means all-black video)
The script printed a sensible Auto BG color and a non-zero targets count

If targets is 0 or the file is too small, see Troubleshooting below.

Auto-Detection Behavior

Image type	Background	Foreground mask
------------	------------	-----------------
Light background (white/gray)	Sampled from 4 corners	Otsu threshold on color distance
Dark background	Sampled from 4 corners	Otsu threshold on color distance
Transparent PNG (RGBA)	Forced to white	Alpha channel directly

Size auto-detection (when --size is not given):

Longest side ≤ 1920, shortest side ≤ 1080, original aspect ratio preserved
1:1 image → ~1080×1080, 16:9 → 1920×1080, 4:3 → 1440×1080

Output Specs

Property	Value
----------	-------
Duration	6 seconds (configurable 3-30s via --duration)
FPS	60 (or 30 via --fps)
Codec	H.264, CRF 18
Audio	AAC 192k, synthesized build-up music (disable via --no-audio)
Particles	10,000 (configurable 1k-50k via --particles)
Convergence	~58% duration particle build → ding → logo reveal

Troubleshooting

Symptom	Cause	Fix
---------	-------	-----
File ~0.2 MB, all black	Particle color == background (e.g., black icon on transparent → black bg)	Should be auto-fixed; if not, check that the image has an alpha channel
`targets: 0` error	No foreground pixels detected	Image may have very low contrast; not currently supported
`ffmpeg: command not found`	FFmpeg not installed	`brew install ffmpeg`
Very slow (>5 min)	Too many target pixels (>500k)	Use a simpler/smaller logo

版本历史

共 1 个版本

v1.0.0 Initial release 当前

2026-05-25 13:55 安全安全

安全检测

腾讯云安全 (Keen)

安全，无风险

查看报告

腾讯云安全 (Sanbu)