unbundle

A clean, ergonomic Rust library for extracting video frames, audio tracks, and subtitles from media files using FFmpeg.

use unbundle::MediaUnbundler;

let mut unbundler = MediaUnbundler::open("video.mp4")?;

// Extract a frame at 30 seconds
let frame = unbundler.video().frame_at(Duration::from_secs(30))?;
frame.save("frame_30s.png")?;

// Extract complete audio track
unbundler.audio().save("audio.wav", AudioFormat::Wav)?;

Why unbundle?

Type-safe API — frames as image::DynamicImage, audio as bytes or files, subtitles as structured events
Flexible extraction — by frame number, timestamp, range, interval, or custom frame lists
Streaming support — lazy iterators and async streams avoid buffering entire frame sets
Rich metadata — dimensions, frame rates, codecs, chapters, per-frame decode info
Production-ready — progress callbacks, cancellation tokens, hardware acceleration, parallel processing

Use Cases

Video thumbnails — contact sheets, smart frame selection, chapter previews
Media processing — format conversion, audio extraction, subtitle manipulation
Analysis tools — scene detection, keyframe analysis, GOP structure inspection
Content indexing — frame extraction for search, waveform visualization
Transcoding pipelines — lossless remuxing, audio re-encoding

Installation

Add to your Cargo.toml:

[dependencies]
unbundle = "3.0"

Or with additional features:

[dependencies]
unbundle = { version = "3.0", features = ["async-tokio", "parallel", "hw-accel"] }

System Requirements

unbundle requires FFmpeg libraries (4.0+) installed on your system.

Linux (Debian/Ubuntu):

sudo apt-get install libavcodec-dev libavformat-dev libavutil-dev \
    libswscale-dev libswresample-dev libavdevice-dev pkg-config

macOS:

brew install ffmpeg pkg-config

Windows:

Download FFmpeg development builds from https://ffmpeg.org/download.html or use vcpkg:

vcpkg install ffmpeg:x64-windows

Set FFMPEG_DIR environment variable if FFmpeg is not in your PATH.

Quick Start

Extract Video Frames

use std::time::Duration;
use unbundle::MediaUnbundler;

let mut unbundler = MediaUnbundler::open("input.mp4")?;

// Extract the first frame
let frame = unbundler.video().frame(0)?;
frame.save("first_frame.png")?;

// Extract a frame at 30 seconds
let frame = unbundler.video().frame_at(Duration::from_secs(30))?;
frame.save("frame_30s.png")?;

Extract Multiple Frames

use std::time::Duration;
use unbundle::{FrameRange, MediaUnbundler};

let mut unbundler = MediaUnbundler::open("input.mp4")?;

// Every 30th frame
let frames = unbundler.video().frames(FrameRange::Interval(30))?;

// Frames between two timestamps
let frames = unbundler.video().frames(
    FrameRange::TimeRange(Duration::from_secs(10), Duration::from_secs(20))
)?;

// Specific frame numbers
let frames = unbundler.video().frames(
    FrameRange::Specific(vec![0, 50, 100, 150])
)?;

Streaming Frame Iteration

use unbundle::{FrameRange, MediaUnbundler};

let mut unbundler = MediaUnbundler::open("input.mp4")?;

// Push-based: process each frame without buffering
unbundler.video().for_each_frame(
    FrameRange::Range(0, 99),
    |frame_number, image| {
        image.save(format!("frame_{frame_number}.png"))?;
        Ok(())
    },
)?;

// Pull-based: lazy iterator with early exit
let iter = unbundler.video().frame_iter(FrameRange::Range(0, 99))?;
for result in iter {
    let (frame_number, image) = result?;
    image.save(format!("frame_{frame_number}.png"))?;
}

Extract Audio

use std::time::Duration;
use unbundle::{AudioFormat, MediaUnbundler};

let mut unbundler = MediaUnbundler::open("input.mp4")?;

// Save complete audio track to WAV
unbundler.audio().save("output.wav", AudioFormat::Wav)?;

// Extract a 30-second segment as MP3
unbundler.audio().save_range(
    "segment.mp3",
    Duration::from_secs(30),
    Duration::from_secs(60),
    AudioFormat::Mp3,
)?;

// Extract audio to memory
let audio_bytes = unbundler.audio().extract(AudioFormat::Wav)?;

// Multi-track: extract the second audio track
let audio_bytes = unbundler.audio_track(1)?.extract(AudioFormat::Wav)?;

Extract Subtitles

use unbundle::{MediaUnbundler, SubtitleFormat};

let mut unbundler = MediaUnbundler::open("input.mkv")?;

// Extract subtitle events with timing
let events = unbundler.subtitle().extract()?;
for event in &events {
    println!("[{:?} → {:?}] {}", event.start_time, event.end_time, event.text);
}

// Save as SRT file
unbundler.subtitle().save("output.srt", SubtitleFormat::Srt)?;

// Multi-track: extract the second subtitle track
unbundler.subtitle_track(1)?.save("track2.vtt", SubtitleFormat::WebVtt)?;

Progress & Cancellation

use std::sync::Arc;
use unbundle::{
    CancellationToken, ExtractionConfig, FrameRange,
    MediaUnbundler, ProgressCallback, ProgressInfo,
};

struct PrintProgress;
impl ProgressCallback for PrintProgress {
    fn on_progress(&self, info: &ProgressInfo) {
        println!("Frame {}/{}", info.current, info.total.unwrap_or(0));
    }
}

let token = CancellationToken::new();
let config = ExtractionConfig::new()
    .with_progress(Arc::new(PrintProgress))
    .with_cancellation(token.clone());

let mut unbundler = MediaUnbundler::open("input.mp4")?;
let frames = unbundler.video().frames_with_config(
    FrameRange::Range(0, 99),
    &config,
)?;

Features

Core Capabilities

Frame extraction — by frame number, timestamp, range, interval, or specific frame list
Audio extraction — to WAV, MP3, FLAC, or AAC (file or in-memory)
Subtitle extraction — decode text-based subtitles to SRT, WebVTT, or raw text
Container remuxing — lossless format conversion (e.g. MKV → MP4) without re-encoding
Rich metadata — video dimensions, frame rate, frame count, audio sample rate, channels, codec info, multi-track audio/subtitle metadata
Configurable output — pixel format (RGB8, RGBA8, GRAY8), target resolution with aspect ratio preservation
Progress & cancellation — cooperative progress callbacks and CancellationToken for long-running operations
Streaming iteration — lazy FrameIterator (pull-based) and for_each_frame (push-based) without buffering entire frame sets
Audio sample iteration — lazy AudioIterator yields mono f32 chunks for incremental audio processing
Validation — inspect media files for structural issues before extraction
Chapter support — extract chapter metadata (titles, timestamps) from containers
Frame metadata — per-frame decode info (PTS, keyframe flag, picture type) via frame_with_info / frames_with_info
Segmented extraction — extract frames from multiple disjoint time ranges in a single call with FrameRange::Segments
Stream probing — lightweight MediaProbe for quick metadata inspection without keeping the demuxer open
Thumbnail helpers — single-frame thumbnails, contact-sheet grids, and variance-based "smart" thumbnail selection
Keyframe & GOP analysis — scan video packets for keyframe positions and GOP structure without decoding
VFR detection — detect variable frame rate streams and analyze PTS distributions
Packet iteration — raw packet-level demuxer iteration for advanced inspection
Efficient seeking — seeks to the nearest keyframe, then decodes forward
Zero-copy in-memory audio — uses FFmpeg's dynamic buffer I/O

Optional Features

Enable additional functionality through Cargo features:

Feature	Description
`async-tokio`	`FrameStream` (async frame iteration) and `AudioFuture` via Tokio
`parallel`	`frames_parallel()` distributes decoding across rayon threads
`hw-accel`	Hardware-accelerated decoding (CUDA, VAAPI, DXVA2, D3D11VA, VideoToolbox, QSV)
`scene-detection`	Scene change detection via FFmpeg's `scdet` filter
`gif`	Animated GIF export from video frames
`waveform`	Audio waveform visualization data (min/max/RMS per bin)
`loudness`	Peak/RMS loudness analysis with dBFS conversion
`transcode`	Audio re-encoding between formats (e.g. AAC → MP3)
`video-writer`	Encode `DynamicImage` sequences into video files (H.264, H.265, MPEG-4)
`full`	Enables all of the above

[dependencies]
unbundle = { version = "3.0", features = ["full"] }

Feature Usage Guide

Use async-tokio when integrating with async web servers or when processing multiple videos concurrently
Use parallel for CPU-intensive batch frame extraction (e.g., generating thousands of thumbnails)
Use hw-accel when processing high-resolution video (4K+) or when CPU is a bottleneck
Use scene-detection for video analysis, automatic chapter detection, or intelligent thumbnail selection
Use gif for creating preview animations or social media content
Use waveform and loudness** for audio visualization or normalization workflows
Use transcode for audio format conversion in media pipelines
Use video-writer for creating time-lapses, slideshows, or re-encoding frame sequences

Examples

The examples/ directory contains complete, runnable examples:

Example	Description
`extract_frames`	Extract frames by number, timestamp, range, interval
`extract_audio`	Extract the complete audio track
`extract_audio_segment`	Extract a specific time range as MP3
`thumbnail_grid`	Create a thumbnail grid from evenly-spaced frames
`metadata`	Display all media metadata
`frame_iterator`	Lazy frame iteration with early exit
`pixel_formats`	Demonstrate RGB8/RGBA8/GRAY8 output
`progress`	Progress callbacks and cancellation
`subtitles`	Extract subtitles as SRT/WebVTT/raw text
`remux`	Lossless container format conversion
`validate`	Media file validation report
`async_extraction`	Async frame streaming and audio extraction (`async-tokio`)
`parallel_extraction`	Parallel frame extraction across threads (`parallel`)
`scene_detection`	Scene change detection (`scene-detection`)
`hw_acceleration`	Hardware-accelerated decoding (`hw-accel`)
`gif_export`	Export video frames as animated GIF (`gif`)
`waveform_analysis`	Generate audio waveform data (`waveform`)
`loudness_analysis`	Analyze audio loudness levels (`loudness`)
`audio_iterator`	Lazy audio sample iteration
`write_video`	Encode image sequences into video files (`video-writer`)
`transcode`	Re-encode audio between formats (`transcode`)
`keyframe_analysis`	GOP/keyframe structure analysis
`vfr_detection`	Variable frame rate detection
`packet_inspect`	Raw packet-level demuxer inspection
`subtitle_search`	Search subtitle text content

Run an example:

cargo run --example metadata -- path/to/video.mp4

Advanced Usage

Container Remuxing

use unbundle::Remuxer;

// Convert MKV to MP4 without re-encoding
Remuxer::new("input.mkv", "output.mp4")?.run()?;

// Exclude subtitles during remux
Remuxer::new("input.mkv", "output.mp4")?
    .exclude_subtitles()
    .run()?;

Custom Output Format

use unbundle::{ExtractionConfig, FrameRange, MediaUnbundler, PixelFormat};

let config = ExtractionConfig::new()
    .with_pixel_format(PixelFormat::Rgba8)
    .with_resolution(Some(1280), Some(720));

let mut unbundler = MediaUnbundler::open("input.mp4")?;
let frames = unbundler.video().frames_with_config(
    FrameRange::Interval(30),
    &config,
)?;

Read Metadata

use unbundle::MediaUnbundler;

let unbundler = MediaUnbundler::open("input.mp4")?;
let metadata = unbundler.metadata();

println!("Duration: {:?}", metadata.duration);
println!("Format: {}", metadata.format);

if let Some(video) = &metadata.video {
    println!("Video: {}x{}, {:.2} fps, {} frames",
        video.width, video.height,
        video.frames_per_second, video.frame_count);
    println!("Codec: {}", video.codec);
}

if let Some(audio) = &metadata.audio {
    println!("Audio: {} Hz, {} channels, codec: {}",
        audio.sample_rate, audio.channels, audio.codec);
}

// List all audio and subtitle tracks
if let Some(tracks) = &metadata.audio_tracks {
    println!("{} audio track(s)", tracks.len());
}
if let Some(tracks) = &metadata.subtitle_tracks {
    println!("{} subtitle track(s)", tracks.len());
}

Validate Media Files

use unbundle::MediaUnbundler;

let unbundler = MediaUnbundler::open("input.mp4")?;
let report = unbundler.validate();

if report.is_valid() {
    println!("File is valid");
} else {
    println!("{report}");
}

Probe Media Files

use unbundle::MediaProbe;

// Quick metadata inspection without keeping the file open
let metadata = MediaProbe::probe("input.mp4")?;
println!("Duration: {:?}", metadata.duration);

// Probe multiple files at once
let results = MediaProbe::probe_many(&["video1.mp4", "video2.mkv"]);

Chapter Metadata

use unbundle::MediaUnbundler;

let unbundler = MediaUnbundler::open("input.mkv")?;
let metadata = unbundler.metadata();

if let Some(chapters) = &metadata.chapters {
    for chapter in chapters {
        println!("[{:?} → {:?}] {}",
            chapter.start, chapter.end,
            chapter.title.as_deref().unwrap_or("Untitled"));
    }
}

Frame Metadata

use unbundle::MediaUnbundler;

let mut unbundler = MediaUnbundler::open("input.mp4")?;

// Get a frame with its decode metadata
let (image, info) = unbundler.video().frame_with_info(0)?;
println!("Frame {}: keyframe={}, type={:?}, pts={:?}",
    info.frame_number, info.is_keyframe, info.frame_type, info.pts);

Thumbnail Generation

use std::time::Duration;
use unbundle::{MediaUnbundler, ThumbnailConfig, ThumbnailGenerator};

let mut unbundler = MediaUnbundler::open("input.mp4")?;

// Single thumbnail at a timestamp
let thumb = ThumbnailGenerator::at_timestamp(&mut unbundler, Duration::from_secs(5), 320)?;

// Contact-sheet grid
let config = ThumbnailConfig::new(4, 3); // 4 columns × 3 rows
let grid = ThumbnailGenerator::grid(&mut unbundler, &config)?;
grid.save("contact_sheet.png")?;

// Smart thumbnail (picks frame with highest visual variance)
let smart = ThumbnailGenerator::smart(&mut unbundler, 10, 320)?;

GIF Export

use std::time::Duration;
use unbundle::{FrameRange, GifConfig, MediaUnbundler};

let mut unbundler = MediaUnbundler::open("input.mp4")?;

let config = GifConfig::new().width(320).frame_delay(10);
unbundler.video().export_gif(
    "output.gif",
    FrameRange::TimeRange(Duration::from_secs(0), Duration::from_secs(5)),
    &config,
)?;

// Or export to memory
let bytes = unbundler.video().export_gif_to_memory(
    FrameRange::Interval(10),
    &config,
)?;

Audio Waveform

use unbundle::{MediaUnbundler, WaveformConfig};

let mut unbundler = MediaUnbundler::open("input.mp4")?;
let waveform = unbundler.audio().generate_waveform(
    &WaveformConfig::new().bins(1000),
)?;

for bin in &waveform.bins {
    println!("min={:.3} max={:.3} rms={:.3}", bin.min, bin.max, bin.rms);
}

Loudness Analysis

use unbundle::MediaUnbundler;

let mut unbundler = MediaUnbundler::open("input.mp4")?;
let loudness = unbundler.audio().analyze_loudness()?;
println!("Peak: {:.1} dBFS, RMS: {:.1} dBFS", loudness.peak_dbfs, loudness.rms_dbfs);

Audio Sample Iteration

use unbundle::MediaUnbundler;

let mut unbundler = MediaUnbundler::open("input.mp4")?;
let iter = unbundler.audio().sample_iter()?;
let mut total_samples = 0u64;
for chunk in iter {
    let chunk = chunk?;
    total_samples += chunk.samples.len() as u64;
}
println!("Total mono samples: {total_samples}");

Audio Transcoding

use unbundle::{AudioFormat, MediaUnbundler, Transcoder};

let mut unbundler = MediaUnbundler::open("input.mp4")?;

// Re-encode audio from the source format to MP3
Transcoder::new(&mut unbundler)
    .format(AudioFormat::Mp3)
    .run("output.mp3")?;

Video Writing

use unbundle::{MediaUnbundler, FrameRange, VideoWriter, VideoWriterConfig, VideoCodec};

let mut unbundler = MediaUnbundler::open("input.mp4")?;
let frames = unbundler.video().frames(FrameRange::Interval(30))?;

let config = VideoWriterConfig::new(1920, 1080)
    .fps(24)
    .codec(VideoCodec::H264);
VideoWriter::new(config).write("output.mp4", &frames)?;

Keyframe & GOP Analysis

use unbundle::MediaUnbundler;

let mut unbundler = MediaUnbundler::open("input.mp4")?;
let gop = unbundler.video().analyze_gops()?;
println!("Keyframes: {}, Avg GOP size: {:.1}", gop.keyframes.len(), gop.average_gop_size);

VFR Detection

use unbundle::MediaUnbundler;

let mut unbundler = MediaUnbundler::open("input.mp4")?;
let vfr = unbundler.video().analyze_vfr()?;
println!("VFR: {}, mean FPS: {:.2}", vfr.is_vfr, vfr.mean_fps);

Packet Inspection

use unbundle::MediaUnbundler;

let mut unbundler = MediaUnbundler::open("input.mp4")?;
for pkt in unbundler.packet_iter()? {
    let pkt = pkt?;
    println!("stream={} pts={:?} size={} key={}",
        pkt.stream_index, pkt.pts, pkt.size, pkt.is_keyframe);
}

API Documentation

Complete API documentation is available at docs.rs/unbundle.

Essential Types

Type	Description
`MediaUnbundler`	Main entry point — opens a media file and provides access to extractors
`VideoExtractor`	Extracts video frames as `DynamicImage`
`AudioExtractor`	Extracts audio tracks as bytes or files
`SubtitleExtractor`	Extracts text-based subtitle tracks
`FrameRange`	Specifies which frames to extract (range, interval, timestamps, etc.)
`ExtractionConfig`	Configure threading, progress callbacks, cancellation, pixel format, resolution, HW accel

Stream & Iteration Types

Type	Description
`FrameIterator`	Lazy, pull-based frame iterator
`AudioIterator`	Lazy pull-based audio sample iterator (mono f32)
`AudioChunk`	A chunk of decoded audio samples with timing
`PacketIterator`	Lazy raw-packet-level demuxer iterator
`FrameStream`	Async stream of decoded frames via Tokio (feature: `async-tokio`)
`AudioFuture`	Async audio extraction future (feature: `async-tokio`)

Configuration Types

Type	Description
`FrameOutputConfig`	Pixel format and resolution settings for frame output
`PixelFormat`	Output pixel format (RGB8, RGBA8, GRAY8)
`AudioFormat`	Output audio format (WAV, MP3, FLAC, AAC)
`SubtitleFormat`	Output subtitle format (SRT, WebVTT, Raw)
`ThumbnailConfig`	Grid thumbnail configuration (columns, rows, width)
`GifConfig`	Animated GIF export configuration (width, delay, repeat) (feature: `gif`)
`WaveformConfig`	Waveform generation settings (bin count, time range) (feature: `waveform`)
`SceneDetectionConfig`	Scene detection threshold configuration (feature: `scene-detection`)
`VideoWriterConfig`	Video writer settings (FPS, resolution, codec, CRF) (feature: `video-writer`)
`HwAccelMode`	Hardware acceleration mode selection (feature: `hw-accel`)

Metadata Types

Type	Description
`MediaMetadata`	Container-level metadata (duration, format)
`VideoMetadata`	Video stream metadata (dimensions, frame rate, codec)
`AudioMetadata`	Audio stream metadata (sample rate, channels, codec)
`SubtitleMetadata`	Subtitle stream metadata (codec, language)
`ChapterMetadata`	Chapter information (title, start/end times)
`FrameInfo`	Per-frame decode metadata (PTS, keyframe flag, picture type)
`FrameType`	Picture type enum (I, P, B, etc.)
`KeyframeInfo`	Keyframe position metadata (packet number, PTS, timestamp)
`GopInfo`	GOP structure analysis result (keyframes, sizes, statistics)
`VfrAnalysis`	Variable frame rate detection result (min/max/mean FPS)
`PacketInfo`	Per-packet metadata (stream index, PTS, DTS, size, keyframe)

Utility Types

Type	Description
`MediaProbe`	Lightweight stateless media file probing
`ThumbnailGenerator`	Thumbnail generation helpers (single, grid, smart)
`Remuxer`	Lossless container format conversion
`ValidationReport`	Result of media file validation
`ProgressCallback`	Trait for receiving progress updates
`ProgressInfo`	Progress event data (current, total, percentage, ETA)
`CancellationToken`	Cooperative cancellation via `Arc<AtomicBool>`
`OperationType`	Identifies the operation being tracked
`UnbundleError`	Error type with rich context
`SubtitleEvent`	A single decoded subtitle event (text, start/end time)
`BitmapSubtitleEvent`	A bitmap subtitle event with image and timing

Feature-Specific Types

Type	Feature	Description
`Transcoder`	`transcode`	Audio re-encoding builder (format, range, bitrate)
`VideoWriter`	`video-writer`	Encodes image sequences into video files
`VideoCodec`	`video-writer`	Supported video codecs (H.264, H.265, MPEG-4)
`WaveformData`	`waveform`	Generated waveform result with per-bin statistics
`WaveformBin`	`waveform`	Single waveform bin (min, max, RMS amplitude)
`LoudnessInfo`	`loudness`	Peak/RMS loudness with dBFS equivalents
`SceneChange`	`scene-detection`	Detected scene change with timestamp and score
`HwDeviceType`	`hw-accel`	Supported HW device types (CUDA, VAAPI, etc.)

Performance

unbundle is designed for efficiency in both single-file and batch processing scenarios:

Smart seeking — Uses FFmpeg's keyframe-based seeking. For sequential access (ranges, intervals), frames are decoded without redundant seeks.
Lightweight decoders — Each extraction call creates a fresh decoder. FFmpeg decoder creation is fast relative to actual decoding work.
Batch optimization — FrameRange::Specific sorts requested frame numbers and processes them in order to minimize seeks.
Memory-efficient streaming — for_each_frame and FrameIterator process frames one at a time without buffering entire frame sets.
Parallel extraction — frames_parallel() (feature parallel) splits frames across rayon threads, each with its own demuxer for true parallelism.
Hardware acceleration — When enabled (feature hw-accel), attempts GPU-accelerated decoding with automatic fallback to software.
Correct stride handling — Properly handles FFmpeg's row padding when converting frames to image buffers.
Zero-copy audio — Uses avio_open_dyn_buf for in-memory audio encoding without temporary files.

Testing

Generate test fixtures first:

# Linux / macOS
bash tests/fixtures/generate_fixtures.sh

# Windows
tests\fixtures\generate_fixtures.bat

Then run tests:

cargo test --all-features

Test Coverage

The test suite includes comprehensive coverage:

Test Module	Coverage
`video_extraction`	Single frames, ranges, intervals, timestamps, specific lists, pixel formats, resolution scaling
`audio_extraction`	WAV/MP3/FLAC/AAC extraction, ranges, file output, multi-track
`subtitle_extraction`	Subtitle decoding, SRT/WebVTT export, multi-track
`metadata`	Container metadata, video/audio/subtitle stream properties
`config`	ExtractionConfig builder, pixel formats, resolution, cancellation
`progress`	ProgressCallback, ProgressInfo fields, CancellationToken
`error_handling`	Error variants, context, invalid inputs, missing streams
`frame_iterator`	FrameIterator, lazy iteration, early exit
`conversion`	Remuxer, stream exclusion, lossless format conversion
`validation`	ValidationReport, warnings, errors, valid files
`chapters`	Chapter metadata extraction, titles, timestamps, ordering
`frame_metadata`	FrameInfo, FrameType, keyframe detection, PTS values
`segmented_extraction`	FrameRange::Segments, multiple disjoint time ranges
`probing`	MediaProbe, probe/probe_many, error handling
`thumbnail`	ThumbnailGenerator, grid, smart selection, aspect ratio
`audio_iter`	AudioIterator, chunk iteration, sample rates
`keyframe_analysis`	GopInfo, KeyframeInfo, GOP statistics
`vfr_analysis`	VfrAnalysis, constant vs variable frame rate
`packet_iter`	PacketIterator, PacketInfo, stream filtering
`subtitle_search`	Subtitle search, case-insensitive matching
`metadata_extended`	Extended metadata: video tracks, colorspace, HDR

Feature-specific tests (require corresponding features enabled):

Test Module	Feature Required	Coverage
`scene_detection`	`scene-detection`	Scene change detection, threshold configuration
`async_extraction`	`async-tokio`	FrameStream, AudioFuture, async streaming
`parallel_extraction`	`parallel`	frames_parallel, sequential parity, interval mode
`hw_accel`	`hw-accel`	HW device enumeration, Auto/Software modes
`gif_export`	`gif`	GIF encoding, file and in-memory output
`waveform`	`waveform`	WaveformConfig, bin statistics, time ranges
`loudness`	`loudness`	Peak/RMS loudness, dBFS values
`video_writer`	`video-writer`	VideoWriter, codec selection, frame encoding
`transcode`	`transcode`	Transcoder, format conversion, time ranges

Benchmarks

Run performance benchmarks:

cargo bench --all-features

Criterion benchmarks are located in benches/ and measure:

Frame extraction throughput (single vs parallel)
Seek performance (sequential vs random access)
Audio extraction speed across formats
Iterator overhead vs batch extraction

Troubleshooting

FFmpeg Linking Errors

Problem: error: linking with 'cc' failed or cannot find -lavcodec

Solution:

Ensure FFmpeg development libraries are installed (see Installation)
Set PKG_CONFIG_PATH to point to FFmpeg's .pc files
On Windows, set FFMPEG_DIR environment variable
Verify with: pkg-config --libs --cflags libavcodec

Codec Not Supported

Problem: UnbundleError::CodecNotSupported

Solution:

Check that your FFmpeg build includes the required codec
Run ffmpeg -codecs to list available codecs
Some codecs require FFmpeg to be built with specific flags (e.g., --enable-libx264)

Hardware Acceleration Fails

Problem: Hardware decoding falls back to software or fails entirely

Solution:

Verify GPU drivers are up to date
Check available HW devices: HwDeviceType::available_devices()
Use ExtractionConfig::with_hw_accel(HwAccelMode::Auto) for automatic fallback
Not all codecs/formats support hardware acceleration
Try ffmpeg -hwaccels to list available hardware acceleration methods

Out of Memory

Problem: High memory usage when extracting many frames

Solution:

Use streaming iteration instead of batch extraction: frame_iter() or for_each_frame()
Process frames in smaller batches
Use AudioIterator for large audio files instead of loading entire tracks

Slow Frame Extraction

Problem: Frame extraction is slower than expected

Solution:

Use frames_parallel() (feature parallel) for CPU-bound workloads
Enable hardware acceleration (feature hw-accel) for high-resolution video
Avoid extracting specific frames in random order — sorted access is much faster
Consider using FrameRange::Interval instead of many individual frame numbers

Permission Denied / File Not Found

Problem: Cannot open media file

Solution:

Verify file path is correct and file exists
Check file permissions (readable by current user)
Ensure file is not locked by another process
On Windows, use raw string literals for paths: r"C:\path\to\video.mp4"

Contributing

Contributions are welcome! Please see the GitHub repository for:

Bug reports and feature requests
Pull requests
Discussions and questions

License

MIT — see LICENSE file for details.

unbundle 3.0.0

unbundle

Why unbundle?

Use Cases

Installation

System Requirements

Quick Start

Extract Video Frames

Extract Multiple Frames

Streaming Frame Iteration

Extract Audio

Extract Subtitles

Progress & Cancellation

Features

Core Capabilities

Optional Features

Feature Usage Guide

Examples

Advanced Usage

Container Remuxing

Custom Output Format

Read Metadata

Validate Media Files

Probe Media Files

Chapter Metadata

Frame Metadata

Thumbnail Generation

GIF Export

Audio Waveform

Loudness Analysis

Audio Sample Iteration

Audio Transcoding

Video Writing

Keyframe & GOP Analysis

VFR Detection

Packet Inspection

API Documentation

Essential Types

Stream & Iteration Types

Configuration Types

Metadata Types

Utility Types

Feature-Specific Types

Performance

Testing

Test Coverage

Benchmarks

Troubleshooting

FFmpeg Linking Errors

Codec Not Supported

Hardware Acceleration Fails

Out of Memory

Slow Frame Extraction

Permission Denied / File Not Found

Contributing

License