Why is the audio out of sync after converting?

Common causes: (1) variable framerate source rendered as constant (use -vsync vfr to preserve VFR); (2) different audio sample rates not resampled (add -ar 48000 ); (3) container limitations (MP4 with variable framerate misbehaves — prefer MKV during editing, encode to MP4 only at the end). Always run ffprobe on both source and output to compare timing.

Why is the audio out of sync after converting?

Common causes: (1) variable framerate source rendered as constant (use -vsync vfr to preserve VFR); (2) different audio sample rates not resampled (add -ar 48000 ); (3) container limitations (MP4 with variable framerate misbehaves — prefer MKV during editing, encode to MP4 only at the end). Always run ffprobe on both source and output to compare timing.

Guide

AV1 Video Codec: The Complete Technical Guide

PC By Pablo Cirre • Updated May 7, 2026

What Is AV1?

AV1 (AOMedia Video 1) is an open, royalty-free video codec developed by the Alliance for Open Media (AOM) — a consortium including Google, Mozilla, Microsoft, Amazon, Netflix, Cisco, Apple, ARM, Intel, NVIDIA, AMD, and Samsung. AV1 was finalized in 2018 as the successor to VP9 and the royalty-free alternative to HEVC.

AV1's founding premise: create a codec that matches or exceeds HEVC compression efficiency while being completely free of patent royalties and openly specified. Independent testing has confirmed AV1 achieves 30–50% better compression than H.264 and 20–30% better than HEVC at equivalent visual quality.

Why AV1 Was Created

HEVC's fragmented patent pools (multiple incompatible licensors demanding separate fees) made it untenable for web browsers, streaming services, and device manufacturers seeking a single open standard. AV1 was specifically designed to be royalty-free by having member companies pool their patents and commit to royalty-free licensing.

The founding AOM members serve as both contributors (developing the codec technology) and implementors (deploying it in their products), creating a closed loop of development and adoption.

AV1 Technical Architecture

AV1 builds on VP9 but adds dozens of new coding tools:

Superblocks and Partition Sizes

Like HEVC's CTU, AV1 uses a hierarchical partition structure. Superblocks can be up to 128×128 pixels (larger than HEVC's 64×64), which is especially efficient for 4K and 8K content. Partition shapes include rectangles, T-shapes, and irregular splits — far more flexible than HEVC's pure quadtree.

Intra Prediction

AV1 has 56 directional intra prediction modes plus smooth/paeth/DC modes. The Paeth predictor uses the pixel above, to the left, and above-left to compute a prediction that follows diagonal gradients — excellent for natural images.

Compound Prediction (Inter Frames)

AV1's most powerful feature: compound inter prediction combines two reference frames with per-pixel blending masks:

DIFFWTD (difference-weighted compound): weights based on pixel difference between the two references
WEDGE: geometric partition blending masks
SEG: combined motion masks

This allows AV1 to model complex motion (e.g., an object moving against a static background) with a single prediction rather than two separate regions.

Film Grain Synthesis

Film grain (natural texture in film-originated content) is extremely hard to compress efficiently. AV1 removes grain before encoding, stores grain parameters as metadata, and re-applies synthesized grain at decode time. This can reduce file size by 10–20% for film content while preserving perceived texture.

Loop Restoration Filter

AV1 adds the Wiener filter (an optimal linear restoration filter) and the Constrained Directional Enhancement Filter (CDEF) as post-processing steps to remove blocking and ringing artifacts. These filters operate on the reconstructed frame and feed back into the reference buffer, improving subsequent predictions.

Reference Frames

AV1 can reference up to 7 different frames (vs. H.264's 16 references from a limited pool and HEVC's similar structure). Most importantly, AV1 allows reference frames in both directions (bi-predictive) from an extended pool, enabling more accurate motion compensation for complex motion.

Transform Coding

AV1 supports multiple transform types beyond the standard DCT:

Asymmetric Discrete Sine Transform (ADST): better for content with gradients from boundary
Flip transforms: variants of DCT/ADST for different signal orientations
Identity transform: pass-through for content where spatial correlation is low

AV1 Encoders

Encoder	Speed	Quality	Use case
libaom-av1	Very slow	Best	Offline encoding, archiving
SVT-AV1	Fast	Very good	Production encoding, Netflix
rav1e	Moderate	Good	Rust-based, open source
aomenc	Very slow	Reference	Research, comparison

SVT-AV1 (Scalable Video Technology for AV1, Intel/Netflix open-source) is now the recommended encoder for production use — it achieves near-libaom quality at 20-30× faster encoding.

ffmpeg AV1 Encoding

# Software encode — libaom (best quality, very slow)
ffmpeg -i input.mp4 -c:v libaom-av1 -crf 35 -b:v 0 output.webm

# Software encode — SVT-AV1 (fast, high quality)
ffmpeg -i input.mp4 -c:v libsvtav1 -crf 35 -preset 6 output.mkv

# Software encode — rav1e
ffmpeg -i input.mp4 -c:v librav1e -qp 70 output.mkv

# Hardware encode — NVIDIA AV1 (RTX 4000 series)
ffmpeg -i input.mp4 -c:v av1_nvenc -cq 35 output.mp4

# Hardware encode — Intel AV1 (Arc, 12th gen+)
ffmpeg -i input.mp4 -c:v av1_qsv -global_quality 35 output.mp4

# Hardware encode — AMD AV1 (RX 7000 series)
ffmpeg -i input.mp4 -c:v av1_amf -qp_i 35 output.mp4

AV1 CRF/Quality Settings

Encoder	Quality flag	Range	Good starting point
libaom	`-crf`	0 (best) – 63 (worst)	30–40
SVT-AV1	`-crf`	0–63	28–38
rav1e	`-qp`	0–255	60–100
NVENC	`-cq`	0 (auto) – 51	30–40

Browser and Hardware Support

Environment	AV1 Support
Chrome	Since 70 (2018), hardware decode since 2021
Firefox	Since 67 (2019)
Edge	Since 88 (2021)
Safari	Since 16 (2022, macOS) — hardware only
Android (Chrome)	Since Android 10
iOS/iPadOS	Since iOS 16 (2022)
YouTube	AV1 for supported browsers/devices since 2018
Netflix	AV1 for Android since 2020, TV/web growing
NVIDIA GPU (decode)	RTX 30 series+
Intel GPU (decode)	11th gen+
AMD GPU (decode)	RX 6000 series+
Apple Silicon	M2 and later

AV1 in Practice

YouTube: uses AV1 for most content when the client supports it, with VP9 as fallback and H.264 for legacy devices. AV1 allows YouTube to serve the same quality at significantly lower bandwidth.

Netflix: adopted SVT-AV1 for their cloud encoding pipeline. Announced ~5 billion AV1 streams served since 2020.

Facebook/Meta: uses AV1 for video in Facebook, Instagram, and WhatsApp.

Discord: uses AV1 for Go Live game streaming.

WebRTC: AV1 is now a required codec in WebRTC specifications (RFC 9274), enabling real-time AV1 video calls.

AV1 vs. HEVC: When to Choose Which

Scenario	Choice	Reason
Web/browser streaming	AV1	Royalty-free, browser native
4K streaming	Either	Netflix/YouTube use AV1; Blu-ray/broadcast use HEVC
UHD Blu-ray authoring	HEVC	Mandatory spec requirement
Mobile apps (Android)	AV1 or HEVC	AV1 hardware decode on Android 10+
Apple ecosystem	HEVC	Better hardware support, Apple optimized
Ultra-slow offline encode	AV1	Best ratio, unlimited time
Real-time streaming	HEVC or H.264	Wider hardware encoder support
Budget hardware	H.264	Universal support, lowest complexity

Converting to/from AV1

H.264/HEVC → AV1: ffmpeg -i in.mp4 -c:v libsvtav1 -crf 35 -preset 6 out.mkv
AV1 → H.264 (for compatibility): ffmpeg -i in.mkv -c:v libx264 -crf 23 out.mp4
AV1 → HEVC: ffmpeg -i in.webm -c:v libx265 -crf 28 out.mkv
Video → AV1 + Opus (web-optimized WebM): ffmpeg -i in.mp4 -c:v libaom-av1 -crf 40 -b:v 0 -c:a libopus -b:a 128k out.webm

Best Practices

Use SVT-AV1 for production encoding — near-libaom quality at 20-30× speed.
Use CRF 28–38 with SVT-AV1 (-preset 6 for balance, -preset 4 for slower/better).
Use WebM or MKV containers for AV1 — MP4 support for AV1 is improving but not universal.
For web: pair AV1 with Opus audio in WebM for a fully royalty-free stack.
Hardware encode (NVENC/QSV AV1) for real-time or near-real-time scenarios.
Test with your target devices — AV1 software decoding on mobile can be CPU-intensive.
For HDR10, AV1 carries HDR metadata via ITU-T T.35 SEI messages.
Enable Film Grain Synthesis (--film-grain-denoise=1) for film-originated content.
Prefer AV1 over VP9 for all new workflows — better ratio, hardware support growing fast.
Keep H.264 as a fallback for devices that do not support AV1 hardware decoding.

Related conversions

Common video conversions that pair well with this guide:

Frequently Asked Questions

libaom-av1 is the reference encoder developed by the Alliance for Open Media — it produces the highest quality output and is used as the benchmark, but it is extremely slow (often 50-100× slower than real-time for 1080p). SVT-AV1 (Scalable Video Technology for AV1) is a parallel, production-grade encoder originally developed by Intel and Netflix, now maintained by the open-source community. SVT-AV1 at preset 6 is 20-30× faster than libaom while achieving within 2-5% of libaom's quality. SVT-AV1 is the recommended encoder for all production workflows. Use libaom only when you have unlimited time and need maximum compression (e.g., archiving a large video library where you run the encoder once and playback thousands of times).

Libaom-av1 is the reference encoder developed pelo Alliance para abrir Media — it produces the altaest quality output e is used como o benchmark, mas it is extremely slow (often 50-100× mais lento que real-time para 1080p). SVT-AV1 (Scalable Video Technology para AV1) is a parallel, production-grade encoder originally developed by Intel e Netflix, now maintained pelo open-source community. SVT-AV1 at preset 6 is 20-30× mais rápido que libaom while achieving within 2-5% of libaom's quality. SVT-AV1 is the recommended encoder para all production workflows. usar libaom only when you have unlimited time e need máximo compressão (e.g., archiving a large video library where you run the encoder once e playback thousands of times).

Libaom-av1 is the reference encoder developed by the Alliance für öffnen Media — it produces the hochest quality output und is used als das benchmark, aber it is extremely slow (often 50-100× langsamer als real-time für 1080p). SVT-AV1 (Scalable Video Technology für AV1) is a parallel, production-grade encoder originally developed by Intel und Netflix, now maintained by the open-source community. SVT-AV1 at preset 6 is 20-30× schneller als libaom while achieving within 2-5% von libaom's quality. SVT-AV1 is the recommended encoder für all production workflows. verwenden libaom only when you have unlimited time und need maximal Komprimierung (e.g., archiving a large video library where you run the encoder once und playback thousands von times).

Libaom-av1 is the reference encoder developed by the Alliance para abrir Media — it produces the altaest quality output y is used como el benchmark, pero it is extremely slow (often 50-100× más lento que real-time para 1080p). SVT-AV1 (Scalable Video Technology para AV1) is a parallel, production-grade encoder originally developed by Intel y Netflix, now maintained by the open-source community. SVT-AV1 at preset 6 is 20-30× más rápido que libaom while achieving within 2-5% de libaom's quality. SVT-AV1 is the recommended encoder para all production workflows. usar libaom only when you have unlimited time y need máximo compresión (e.g., archiving a large video library where you run the encoder once y playback thousands de times).

AV1 is the most efficient (royalty-free, ~30% smaller than H.265) but encoding is slow. H.265 (HEVC) saves ~30–50% over H.264 and is supported by every modern phone and desktop. H.264 remains the safest baseline for legacy compatibility. Rule of thumb: archives → AV1, daily use → H.265, broadest reach → H.264.

Yes, independent tests consistently show AV1 compresses approximately 20-30% better than HEVC at equivalent visual quality (meaning the same perceptual quality in AV1 requires 20-30% fewer bits than HEVC). Compared to H.264, AV1 is 30-50% more efficient. These gains vary by content type: film-originated content with film grain synthesis sees larger gains; screen recordings and animation see smaller gains. The Moscow State University (MSU) codec comparison, Netflix's internal benchmarks, and Xiph's tests all confirm this range. The caveat: these gains are only realized when using high-quality encoders (libaom or SVT-AV1) at slow presets; fast presets reduce the advantage.

Sim, independent tests consistently show AV1 compresses approximately 20-30% melhor que HEVC at equivalent visual quality (meaning the same perceptual quality in AV1 requires 20-30% fewer bits than HEVC). comparado a H.264, AV1 is 30-50% more efficient. These gains vary by content type: film-originated content com film grain synthesis sees larger gains; screen recordings e animation see smaller gains. The Moscow State University (MSU) codec comparison, Netflix's internal benchmarks, e Xiph's tests all confirm this range. The caveat: these gains are only realized when usando alta-quality encoders (libaom ou SVT-AV1) at slow presets; fast presets reduce the advantage.

Ja, independent tests consistently show AV1 compresses approximately 20-30% besser als HEVC at equivalent visual quality (meaning the same perceptual quality in AV1 requires 20-30% fewer bits than HEVC). im Vergleich zu H.264, AV1 is 30-50% more efficient. These gains vary by content type: film-originated content mit film grain synthesis sees larger gains; screen recordings und animation see smaller gains. The Moscow State University (MSU) Codec comparison, Netflix's internal benchmarks, und Xiph's tests all confirm this range. The caveat: these gains are only realized when using hoch-quality encoders (libaom oder SVT-AV1) at slow presets; fast presets reduce the advantage.

Sí, independent tests consistently show AV1 compresses approximately 20-30% mejor que HEVC at equivalent visual quality (meaning the same perceptual quality in AV1 requires 20-30% fewer bits than HEVC). comparado con H.264, AV1 is 30-50% more efficient. These gains vary by content type: film-originated content con film grain synthesis sees larger gains; screen recordings y animation see smaller gains. The Moscow State University (MSU) codec comparison, Netflix's internal benchmarks, y Xiph's tests all confirm this range. The caveat: these gains are only realized when using alta-quality encoders (libaom o SVT-AV1) at slow presets; fast presets reduce the advantage.

CRF (Constant Rate Factor) is the best default for offline files: ffmpeg picks the bitrate frame-by-frame to maintain perceived quality. Two-pass is only better when you must hit an exact final size (DVD targets). Constant bitrate is for streaming with a fixed channel. For "smallest at quality X" always use CRF.

Film grain is the random texture noise inherent in film-originated content (from photographic film grain and sensor noise). This noise is extremely hard to compress because it is pseudo-random — each frame has different grain that cannot be predicted from previous frames. Traditional codecs allocate significant bits to encoding this noise. AV1's Film Grain Synthesis removes grain from the source before encoding (the encoder de-noises the frame), stores only a compact statistical description of the grain parameters (standard deviation, correlation across color channels), and the decoder re-applies procedurally generated grain matching the original parameters. For film content, this typically saves 10-20% of file size while preserving the perceptual quality of grain.

Film grain is the random texture noise inherent in film-originated content (from photographic film grain e sensor noise). This noise is extremely hard to compress because it is pseudo-random — each frame has different grain that cannot be predicted de previous frames. Traditional codecs allocate significant bits to codificação this noise. AV1's Film Grain Synthesis removes grain de the source antes codificação (the encoder de-noises the frame), stores only a compact statistical description of the grain parameters (standard deviation, correlation across color channels), e the decoder re-applies procedurally generated grain matching the original parameters. para film content, this tipicamente saves 10-20% of tamanho do arquivo while preserving the perceptual quality of grain.

Film grain is the random texture noise inherent in film-originated content (from photographic film grain und sensor noise). This noise is extremely hard to compress because it is pseudo-random — each frame has different grain that cannot be predicted von previous frames. Traditional Codecs allocate significant bits to Codierung this noise. AV1's Film Grain Synthesis removes grain von the source vor Codierung (the encoder de-noises the frame), stores only a compact statistical description des grain parameters (standard deviation, correlation across color channels), und the decoder re-applies procedurally generated grain matching the original parameters. für film content, this typically saves 10-20% von Dateigröße while preserving the perceptual quality von grain.

Film grain is the random texture noise inherent in film-originated content (from photographic film grain y sensor noise). This noise is extremely hard to compress because it is pseudo-random — each frame has different grain that cannot be predicted de previous frames. Traditional codecs allocate significant bits to codificación this noise. AV1's Film Grain Synthesis removes grain de the source antes codificación (the encoder de-noises the frame), stores only a compact statistical description del grain parameters (standard deviation, correlation across color channels), y the decoder re-applies procedurally generated grain matching the original parameters. para film content, this typically saves 10-20% de tamaño de archivo while preserving the perceptual quality de grain.

Common causes: (1) variable framerate source rendered as constant (use <code>-vsync vfr</code> to preserve VFR); (2) different audio sample rates not resampled (add <code>-ar 48000</code>); (3) container limitations (MP4 with variable framerate misbehaves — prefer MKV during editing, encode to MP4 only at the end). Always run <code>ffprobe</code> on both source and output to compare timing.

AV1 is production-ready for web streaming, mobile apps, and desktop applications. YouTube uses AV1 for billions of streams. Netflix uses AV1 for Android and growing platforms. Browser support covers Chrome, Firefox, Edge, Safari (16+), and all major mobile browsers. Hardware decoding is now available on most devices sold from 2021 onwards (NVIDIA RTX 30+, Intel 11th gen+, AMD RX 6000+, Apple M2+). For web/streaming: use AV1 as primary with H.264 fallback. For professional production (broadcast, Blu-ray): still use HEVC or H.264 where standards mandate it. For real-time applications: hardware AV1 encoders (NVENC, QSV AV1) make real-time AV1 feasible on modern GPUs. The only remaining concern is software decoding CPU cost on old hardware.

AV1 is production-ready para web streaming, mobile apps, e desktop aplicativos. YouTube uses AV1 para billions of streams. Netflix uses AV1 para Android e growing plataformas. Browser support covers Chrome, Firefox, Edge, Safari (16+), e all major mobile browsers. Hardware decoding is now disponível on maioria dos dispositivos sold de 2021 onwards (NVIDIA RTX 30+, Intel 11th gen+, AMD RX 6000+, Apple M2+). para web/streaming: usar AV1 as primary com H.264 fallback. para profissional production (broadcast, Blu-ray): still usar HEVC ou H.264 where padrãos mandate it. para real-time aplicativos: hardware AV1 encoders (NVENC, QSV AV1) make real-time AV1 feasible on moderno GPUs. The only remaining concern is software decoding CPU cost on old hardware.

AV1 is production-ready für web streaming, mobile apps, und desktop Anwendungen. YouTube uses AV1 für billions von streams. Netflix uses AV1 für Android und growing Plattformen. Browser support covers Chrome, Firefox, Edge, Safari (16+), und all major mobile browsers. Hardware decoding is now verfügbar on most devices sold von 2021 onwards (NVIDIA RTX 30+, Intel 11th gen+, AMD RX 6000+, Apple M2+). für web/streaming: verwenden AV1 as primary mit H.264 fallback. für professionell production (broadcast, Blu-ray): still verwenden HEVC oder H.264 where Standards mandate it. für real-time Anwendungen: hardware AV1 encoders (NVENC, QSV AV1) make real-time AV1 feasible on modern GPUs. The only remaining concern is Software decoding CPU cost on old hardware.

AV1 is production-ready para web streaming, mobile apps, y desktop aplicaciones. YouTube uses AV1 para billions de streams. Netflix uses AV1 para Android y growing plataformas. Browser support covers Chrome, Firefox, Edge, Safari (16+), y all major mobile browsers. Hardware decoding is now disponible on most devices sold de 2021 onwards (NVIDIA RTX 30+, Intel 11th gen+, AMD RX 6000+, Apple M2+). para web/streaming: usar AV1 as primary con H.264 fallback. para profesional production (broadcast, Blu-ray): still usar HEVC o H.264 where estándars mandate it. para real-time aplicaciones: hardware AV1 encoders (NVENC, QSV AV1) make real-time AV1 feasible on moderno GPUs. The only remaining concern is software decoding CPU cost on old hardware.

Yes if you only change the container: <code>ffmpeg -i in.mkv -c copy out.mp4</code>. This remuxes the stream without re-encoding, takes seconds even for hours of footage. Limitations: codec must be supported by the target container (e.g. you cannot put H.264 in WebM, only VP8/VP9/AV1). To shrink size you must re-encode.