movielite 0.2.1

A performance-oriented Python video editing library focused on speed and simplicity.

MovieLite

A fast and intuitive Python video editing library, designed as a lightweight alternative to MoviePy.

The animation above was fully generated using MovieLite itself!
Check out the code here.

Overview

MovieLite is a Python video editing library built with a focus on speed and simplicity for common workflows. It offers a clean, chainable API that will feel familiar to MoviePy users, but delivers significant performance gains for CPU-based rendering.

The library achieves this speed by leveraging Numba's JIT compilation for critical operations like alpha blending and transformations. While it doesn't aim for feature-for-feature parity with MoviePy and currently lacks GPU acceleration, it excels at providing a much faster experience for everyday tasks, such as:

• Applying transformations (zoom, position, scale).
• Compositing overlays (text, images, other videos).
• Rendering projects with multiple effects.

If you love MoviePy's approach but need a significant speed boost for your CPU-bound projects, MovieLite is for you.

Key Features

• Frame-by-frame processing: Full control over every frame and audio sample
• Performance optimized: Uses Numba JIT compilation for critical rendering operations
• Clean API: Chainable methods for intuitive video editing workflows
• Comprehensive effects library: Built-in visual effects (vfx), audio effects (afx), and transitions (vtx)
• Multiprocessing support: Parallel rendering for faster video generation
• Audio mixing: Mix multiple audio tracks with per-sample control
• Memory efficient: Streaming architecture for large video files

Performance Comparison

MovieLite is designed to be faster than moviepy for common video editing tasks. Performance improvements come from:

1. Numba JIT compilation: Critical rendering loops are compiled to native code
2. Optimized compositing: Efficient alpha blending and frame composition
3. Memory management: Streaming architecture reduces memory footprint
4. Multiprocessing: Parallel frame rendering for multi-core systems

Benchmark Results

To run benchmarks comparing movielite with moviepy 2.2.1:

python benchmarks/compare_moviepy.py --input /path/to/input

Real benchmark results (1280x720 video, 30fps):

Task	movielite	moviepy	Speedup
No processing	6.34s	6.71s	1.06x 🚀
Video zoom	9.52s	31.81s	3.34x 🚀
Fade in/out	8.53s	9.03s	1.06x 🚀
Text overlay	7.82s	35.35s	4.52x 🚀
Video overlay	18.22s	75.47s	3.14x 🚀
Alpha video overlay	10.75s	42.11s	3.92x 🚀
Complex mix*	38.07s	175.31s	4.61x 🚀
Total	99.24s	375.79s	3.79x 🚀

*Complex mix includes: video with zoom + fade, image clips with fade, text overlay, video overlay - all composed together.

movielite excels at:

• Transform operations (zoom, scale, resize) - up to 3.17x faster
• Text overlays and compositing - up to 3.75x faster
• Video overlays and layering - up to 3.83x faster
• Alpha channel compositing - up to 3.36x faster
• Complex multi-effect compositions - up to 4.21x faster

Results may vary based on hardware, video codec, and complexity.

Architecture

Frame-by-Frame Processing

Like moviepy, movielite operates on a frame-by-frame basis:

# Every frame is individually processed
for frame_idx in range(total_frames):
    t = frame_idx / fps
    frame = clip.get_frame(t)  # Get raw frame
    frame = apply_transforms(frame, t)  # Apply effects
    frame = blend_with_other_clips(frame, t)  # Composite
    write_frame_to_output(frame)

This approach provides:

• Complete control over every pixel
• Ability to apply time-based effects
• Support for complex compositing operations
• Memory efficiency through streaming

Numba Optimization

Critical operations are JIT-compiled with Numba:

@numba.jit(nopython=True, cache=True)
def blend_foreground_with_bgr_background_inplace(
    background, foreground, x, y, opacity, mask, ...
):
    # Pixel-perfect alpha blending at native speed
    ...

This provides near-native performance for:

• Alpha blending operations
• Pixel-wise transformations
• Color space conversions

What It Supports

Video Capabilities

• Video clips with frame-level access
• Alpha video clips (transparency support via AlphaVideoClip)
• Image clips (static images as video frames)
• Text rendering (via pictex library)
• Mask support for advanced compositing effects
• Video effects: fade, blur, color adjustments, vignette, zoom, glitch effects
• Video transitions: crossfade, dissolve, blur dissolve
• Position, scale, opacity, and size transformations
• Custom frame transformations
• Video looping

Audio Capabilities

• Audio clips with sample-level access
• Audio effects: fade in/out
• Volume control and volume curves
• Multiple audio track mixing
• Audio from video files
• Custom audio transformations

Output Formats

• Video codec: libx264 (H.264)
• Container format: MP4
• Audio codec: AAC
• Quality presets: LOW, MIDDLE, HIGH, VERY_HIGH

Limitations

• Output is limited to MP4 format with libx264 codec
• No GPU acceleration (CPU-based rendering only)

Installation

pip install movielite

Requirements

• Python 3.10+
• FFmpeg (must be installed and available in PATH)
• NumPy
• OpenCV (opencv-python)
• Numba (for JIT compilation)
• multiprocess
• tqdm
• pictex (for TextClip)

Installing FFmpeg

Windows:

1. Download FFmpeg from ffmpeg.org
2. Extract to a permanent location (e.g., C:\ffmpeg)
3. Add the bin directory to your PATH environment variable

macOS:

brew install ffmpeg

Linux:

sudo apt-get install ffmpeg  # Debian/Ubuntu
sudo yum install ffmpeg      # CentOS/RHEL

Quick Start

Basic Video Editing

from movielite import VideoClip, VideoWriter

# Load and extract a segment
clip = VideoClip("input.mp4")
subclip = clip.subclip(10, 15)  # Extract seconds 10-15

# Create output
writer = VideoWriter("output.mp4", fps=30)
writer.add_clip(
    subclip
    .set_position((100, 100))
    .set_opacity(0.8)
)
writer.write()

clip.close()

Adding Effects

from movielite import VideoClip, VideoWriter, vfx

clip = VideoClip("video.mp4")

# Chain multiple effects
clip.add_effect(vfx.FadeIn(1.0))
clip.add_effect(vfx.FadeOut(1.5))
clip.add_effect(vfx.Blur(intensity=3.0))

writer = VideoWriter("output.mp4", fps=clip.fps, size=clip.size)
writer.add_clip(clip)
writer.write()

clip.close()

Text Overlays

from movielite import VideoClip, TextClip, VideoWriter
from pictex import Canvas, LinearGradient, Shadow

# Create styled text
canvas = (
    Canvas()
    .font_family("Arial")
    .font_size(60)
    .color("white")
    .padding(20)
    .background_color(LinearGradient(["#2C3E50", "#FD746C"]))
    .border_radius(10)
    .text_shadows(Shadow(offset=(2, 2), blur_radius=3, color="black"))
)

video = VideoClip("background.mp4")
text = TextClip("Hello World", start=2, duration=3, canvas=canvas)
text.set_position((video.size[0] // 2 - text.size[0] // 2, 100))

writer = VideoWriter("output.mp4", fps=video.fps, size=video.size)
writer.add_clip(video)
writer.add_clip(text)
writer.write()

video.close()

Concatenating Videos

from movielite import VideoClip, VideoWriter

clip1 = VideoClip("intro.mp4", start=0)
clip2 = VideoClip("main.mp4", start=clip1.end)
clip3 = VideoClip("outro.mp4", start=clip2.end)

writer = VideoWriter("final.mp4", fps=30)
writer.add_clips([clip1, clip2, clip3])
writer.write()

clip1.close()
clip2.close()
clip3.close()

Audio Mixing

from movielite import VideoClip, AudioClip, VideoWriter, afx

video = VideoClip("video.mp4")

# Background music at 50% volume
music = AudioClip("background.mp3", start=0, volume=0.5)
music.add_effect(afx.FadeIn(2)).add_effect(afx.FadeOut(2))

# Sound effect at specific time
sfx = AudioClip("ding.wav", start=5.0, volume=1.0)

writer = VideoWriter("output.mp4", fps=video.fps)
writer.add_clip(video)  # Video includes its own audio track
writer.add_clip(music)
writer.add_clip(sfx)
writer.write()

video.close()

Transitions

from movielite import VideoClip, VideoWriter, vtx

clip1 = VideoClip("scene1.mp4", start=0, duration=5)
clip2 = VideoClip("scene2.mp4", start=4, duration=5)  # 1s overlap

# Apply crossfade transition
clip1.add_transition(clip2, vtx.CrossFade(duration=1))

writer = VideoWriter("output.mp4", fps=30)
writer.add_clips([clip1, clip2])
writer.write()

clip1.close()
clip2.close()

Custom Frame Transformations

import cv2
import numpy as np
from movielite import VideoClip, VideoWriter

clip = VideoClip("video.mp4")

# Apply custom sepia effect
def sepia_transform(frame: np.ndarray, t: float) -> np.ndarray:
    kernel = np.array([
        [0.131, 0.534, 0.272],
        [0.168, 0.686, 0.349],
        [0.189, 0.769, 0.393]
    ])
    sepia = cv2.transform(frame, kernel)
    return np.clip(sepia, 0, 255).astype(np.uint8)

clip.add_transform(sepia_transform)

writer = VideoWriter("output.mp4", fps=clip.fps)
writer.add_clip(clip)
writer.write()

clip.close()

Masking for Advanced Compositing

import numpy as np
from movielite import VideoClip, TextClip, VideoWriter
from pictex import Canvas

# Video to be masked
video = VideoClip("waves.mp4", start=0, duration=10)

# Create text as mask
canvas = Canvas().font_size(200).color("white").background_color("transparent")
text = TextClip("Hello World!", start=0, duration=10, canvas=canvas)

# Animate mask position and scale
text.set_position(lambda t: (
    960 - text.size[0] // 2,
    500 + int(20 * np.sin(2 * np.pi * (t / text.duration)))
))
text.set_scale(lambda t: 1.0 + 0.4 * (t / text.duration))

# Apply mask - video only visible through text
video.set_mask(text)
video.set_size(1920, 1080)

writer = VideoWriter("output.mp4", fps=30, size=(1920, 1080))
writer.add_clip(video)
writer.write()

video.close()

Multiprocessing for Faster Rendering

from movielite import VideoClip, VideoWriter, VideoQuality

clip = VideoClip("input.mp4")

writer = VideoWriter("output.mp4", fps=clip.fps, size=clip.size)
writer.add_clip(clip)

# Use 8 parallel processes for rendering
writer.write(processes=8, video_quality=VideoQuality.HIGH)

clip.close()

API Reference

For detailed API documentation, see docs/api.md.

Core Classes

• VideoClip - Load and manipulate video files
• AlphaVideoClip - Video with alpha channel support
• ImageClip - Static images as video frames
• TextClip - Rendered text overlays
• AudioClip - Audio track handling
• VideoWriter - Render composition to file

Effect Modules

• vfx - Visual effects (fade, blur, color, zoom, glitch)
• afx - Audio effects (fade in/out)
• vtx - Video transitions (crossfade, dissolve)

Advanced Usage

For advanced topics, see docs/advanced.md:

• Custom effect development
• Performance optimization techniques
• Memory management strategies
• Advanced audio processing
• Masking and compositing techniques
• Integration with other libraries

Examples

Additional examples are available in the examples/ directory:

• basic_editing.py - Simple cuts and concatenation
• effects_showcase.py - All built-in effects
• text_animations.py - Animated text overlays
• audio_mixing.py - Complex audio compositions
• transitions.py - Transition examples
• masking_effects.py - Advanced masking techniques

Roadmap and Future Directions

Our focus will be on three key areas: performance, developer experience, and flexibility.

Contributions are highly welcome! If any of these ideas resonate with you, feel free to open an issue or a pull request to discuss it.

1. Performance Enhancements

• GPU Acceleration: While MovieLite is well optimized for CPU, adding optional GPU support (e.g., using CuPy or PyTorch for transformations and blending) is a major goal. This would unlock another order-of-magnitude performance boost for users with compatible hardware.
• Core Rendering Optimizations: We plan to introduce more intelligent caching. For instance, an ImageClip with a static scale shouldn't be re-rendered on every frame.
• Expanded Numba JIT Usage: Some of the current visual effects (vfx) can be rewritten using Numba to run at near-native speed, further improving rendering times for complex compositions.

2. Developer Experience (DevEx)

• A More Intuitive API: We want to make the library even more pleasant to work with. This includes:
  • Adding convenience features like string-based positioning (.set_position("center")).
  • Implementing a deep .copy() method for clips to make duplication easier.
  • Providing ways to manage the effect stack, such as removing or replacing transformations after they've been added.
• Better Type Hinting and Documentation: Improving inline documentation and type hints to enhance autocompletion and code analysis in modern editors.

3. Flexibility and Advanced Control

• Flexible Output Options: We aim to give users more control over the final render by:
  • Officially supporting more video/audio codecs and container formats (e.g., WebM, MOV).
  • Exposing FFmpeg parameters directly in the VideoWriter, similar to MoviePy's ffmpeg_params, for advanced users who need fine-grained control over encoding quality, presets, and other options.
• Expanded Effects Library: Continuously adding new, performant effects to the vfx, afx, and vtx modules based on community feedback and common use cases.

License

MIT License - see LICENSE for details.

Changelog

See CHANGELOG.md for version history and changes.