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

HEVC / H.265 Video Codec: The Complete Technical Guide

PC By Pablo Cirre • Updated May 7, 2026

What Is HEVC / H.265?

HEVC — High Efficiency Video Coding — is the video compression standard that succeeded H.264/AVC. Jointly developed by the ITU-T Video Coding Experts Group (VCEG) and the ISO/IEC Moving Picture Experts Group (MPEG) under the Joint Collaborative Team on Video Coding (JCT-VC), HEVC was finalized in 2013 and published as ISO/IEC 23008-2 and ITU-T H.265.

HEVC's defining achievement: equivalent visual quality to H.264 at approximately half the bit rate. This efficiency gain comes at the cost of dramatically higher computational complexity — HEVC encoding is 5–10× slower than H.264 at equivalent quality settings.

HEVC is the dominant codec for:

4K UHD streaming (Netflix, Disney+, Amazon Prime — all use HEVC for 4K)
4K Blu-ray discs (BDMV spec mandates HEVC for UHD Blu-ray)
HEIF images (the still-image application of HEVC — default on Apple devices since iOS 11)
Video surveillance (H.265 halves storage requirements)
Broadcast TV (DVB-T2, ATSC 3.0, ISDB-T2 all support HEVC)

Key Improvements Over H.264

Larger Coding Unit (CTU vs. Macroblock)

H.264 uses 16×16 macroblocks. HEVC introduces the Coding Tree Unit (CTU) with a maximum size of 64×64 pixels. The CTU can be recursively split into smaller Coding Units (CU) down to 8×8, allowing the encoder to use large units for flat regions (one prediction for a 64×64 block) and small units for detail areas. This quadtree structure dramatically reduces overhead for high-resolution content.

Improved Intra Prediction

H.264 has 9 intra prediction directions. HEVC expands to 35 intra prediction directions (33 angular modes + planar + DC), allowing much better prediction of textured surfaces, diagonal edges, and complex patterns.

Advanced Motion Compensation

Merge mode: inherit motion vectors from neighboring blocks without explicit signaling.
Advanced TMVP: temporal motion vector predictor from previously encoded frames.
Overlapped Block Motion Compensation (OBMC) (in some profiles).
1/4-pixel precision for luma, 1/8-pixel for chroma.

In-Loop Filters

HEVC adds two in-loop filters applied during encoding (which means reference frames are deblocked/denoised, improving prediction quality):

Deblocking filter: removes blocking artifacts at CTU boundaries.
Sample Adaptive Offset (SAO): reduces ringing artifacts by classifying samples and adding offsets — a novel addition not present in H.264.

HEVC Profiles

Profile	Bit depth	Chroma format	Use case
Main	8-bit	4:2:0	Consumer streaming, broadcasting
Main 10	8/10-bit	4:2:0	HDR video (most common profile)
Main Still Picture	8-bit	4:2:0	HEIF images
Main 12	12-bit	4:2:0	Professional mastering
Main 4:2:2 10	10-bit	4:2:2	Professional production
Main 4:4:4	8-12 bit	4:4:4	High quality post-production
Scalable Main	—	—	Streaming with multiple quality layers
Multiview Main	—	—	3D/VR video
Screen Content Coding	—	—	Screen recordings, text, graphics

Main 10 is the dominant profile for consumer use — it is required for HDR10 content and supported by virtually all modern hardware decoders.

HEVC and HDR

HEVC is the first codec standardized alongside major HDR technologies:

HDR10: HEVC Main 10, static metadata (MaxCLL, MaxFALL) via SEI messages.
HDR10+: Dynamic metadata per scene (Samsung/Amazon).
Dolby Vision: Dual-layer HEVC with BL (base layer) + EL (enhancement layer), or single-layer with RPU metadata.
HLG (Hybrid Log-Gamma): BBC/NHK HDR system for broadcast.

HDR requires 10-bit encoding (Main 10 profile) to provide sufficient precision for the BT.2020 wide color gamut.

ffmpeg HEVC Encoding

# Software encode (libx265) — high quality, slow
ffmpeg -i input.mp4 -c:v libx265 -crf 28 -preset medium output.mp4

# Hardware encode — NVIDIA NVENC
ffmpeg -i input.mp4 -c:v hevc_nvenc -rc:v vbr -cq:v 28 output.mp4

# Hardware encode — Intel QSV
ffmpeg -i input.mp4 -c:v hevc_qsv -global_quality 25 output.mp4

# Hardware encode — Apple VideoToolbox (macOS)
ffmpeg -i input.mp4 -c:v hevc_videotoolbox -q:v 60 output.mp4

# 4K HDR10 encoding
ffmpeg -i input.mkv \
  -c:v libx265 -crf 20 -preset slow \
  -x265-params "hdr-opt=1:repeat-headers=1:colorprim=bt2020:transfer=smpte2084:colormatrix=bt2020nc:master-display=G(13250,34500)B(7500,3000)R(34000,16000)WP(15635,16450)L(10000000,50):max-cll=1000,400" \
  -c:a copy output_hdr10.mkv

# Convert H.264 to HEVC (quality-preserving transcode)
ffmpeg -i input_h264.mp4 -c:v libx265 -crf 24 -preset medium -c:a copy output_hevc.mp4

CRF Values for libx265

CRF	Quality	File size vs. CRF 28
18	Visually lossless	~3× larger
22	Excellent	~1.5× larger
24	Very good	~1.2× larger
28	Good (default)	Reference
32	Acceptable	~0.7×
36	Low	~0.5×
51	Worst	Smallest

HEVC Licensing: The Patent Minefield

HEVC's biggest practical problem is its patent licensing situation. Three patent pools and numerous individual licensors claim essential patents:

HEVC Advance: ~USD 0.60 per device, or 0.2% of revenue
Velos Media: additional license required
Via LA: another pool with different terms
Individual licensors: Qualcomm, Sisvel, others

This fragmentation led many companies to seek HEVC-free alternatives, driving adoption of AV1.

HEVC Container Formats

HEVC is typically stored in:

HEVC Annex B (.hevc, .265): raw bitstream, used in broadcasting
MP4/ISOBMFF (.mp4, .m4v): standard web and consumer container
Matroska (.mkv): preferred for archiving and PC playback
MPEG-TS (.ts, .m2ts): broadcast and Blu-ray
MOV (.mov): Apple's container

HEVC vs. H.264 vs. AV1

Property	H.264	HEVC	AV1
Year	2003	2013	2018
Bit rate vs. H.264	Reference	~50% lower	~30–50% lower
Encoding complexity	Reference	~5× higher	~20× higher
Decoding complexity	Reference	~2× higher	~3-5× higher
HW decode	Universal	Most modern	Growing (2021+)
License	Free (patents expired)	Expensive, fragmented	Royalty-free
4K streaming	Common	Dominant	Growing
Browser support	Universal	Partial	Growing

Best Practices

For 4K/HDR content, HEVC (Main 10) is the current standard — all streaming platforms and UHD Blu-ray use it.
Use CRF 24–28 with libx265 for a good quality/size balance.
Use hardware encoders (NVENC, QSV, VideoToolbox) for real-time or near-real-time encoding.
Use software encoder (libx265) for archiving — better quality at same file size.
Use .mkv container for HEVC archives — MKV handles HEVC streams cleanly.
For HDR10, set correct metadata (MaxCLL, MaxFALL, mastering display) for proper tone mapping.
For 1080p and below, consider AV1 or H.264 — HEVC licensing costs may not be justified.
Preset slow or slower gives ~10–15% better compression than medium at significant time cost.
Use --pools flag in x265 to control thread usage on multicore systems.
For broadcast, use HEVC Annex B in MPEG-TS containers with the appropriate profile level tier.

Related conversions

Common video conversions that pair well with this guide:

Frequently Asked Questions

HEVC's efficiency gains come directly from increased computational complexity. The quadtree Coding Tree Unit (CTU) structure requires the encoder to evaluate many possible block sizes (64×64, 32×32, 16×16, 8×8) and choose the best at every position. The 35 intra prediction directions (vs. H.264's 9) multiply the number of options to test. Advanced merge mode and TMVP require more complex motion estimation searches. The Sample Adaptive Offset (SAO) filter needs per-frame optimization. All these decisions happen in a rate-distortion optimization loop — the encoder tries many possibilities and picks the best quality-per-bit combination. Encoding 5-10× slower than H.264 at medium preset; using the "fast" preset reduces this to 2-3× but sacrifices some compression efficiency.

HEVC's efficiency gains come directly de increased computational complexoity. The quadtree Coding Tree Unit (CTU) structure requires the encoder to evaluate many possible block sizes (64×64, 32×32, 16×16, 8×8) e choose the best at every position. The 35 intra prediction directions (vs. H.264's 9) multiply the number of options to test. Advanced merge mode e TMVP require more complexo motion estimation searches. The Sample Adaptive Offset (SAO) filter needs per-frame optimization. All these decisions happen em um rate-distortion optimization loop — the encoder tries many possibilities e picks the melhor qualidade-per-bit combination. codificação 5-10× mais lento que H.264 at medium preset; usando the "fast" preset reduces this to 2-3× mas sacrifices some compressão efficiency.

HEVC's efficiency gains come directly von increased computational complexity. The quadtree Coding Tree Unit (CTU) structure requires the encoder to evaluate many possible block sizes (64×64, 32×32, 16×16, 8×8) und choose the best at every position. The 35 intra prediction directions (vs. H.264's 9) multiply the number von options to test. Advanced merge mode und TMVP require more complex motion estimation searches. The Sample Adaptive Offset (SAO) filter needs per-frame optimization. All these decisions happen in einem rate-distortion optimization loop — the encoder tries many possibilities und picks the beste Qualität-per-bit combination. Codierung 5-10× langsamer als H.264 at medium preset; using the "fast" preset reduces this to 2-3× aber sacrifices some Komprimierung efficiency.

HEVC's efficiency gains come directly de increased computational complexity. The quadtree Coding Tree Unit (CTU) structure requires the encoder to evaluate many possible block sizes (64×64, 32×32, 16×16, 8×8) y choose the best at every position. The 35 intra prediction directions (vs. H.264's 9) multiply the number de options to test. Advanced merge mode y TMVP require more complex motion estimation searches. The Sample Adaptive Offset (SAO) filter needs per-frame optimization. All these decisions happen en un rate-distortion optimization loop — the encoder tries many possibilities y picks the mejor calidad-per-bit combination. codificación 5-10× más lento que H.264 at medium preset; using the "fast" preset reduces this to 2-3× pero sacrifices some compresión efficiency.

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.

HEVC Main profile supports 8-bit per channel video with 4:2:0 chroma subsampling — standard dynamic range (SDR) content. Main 10 profile additionally supports 10-bit per channel video, providing 4× more tonal gradations per channel (1024 vs 256 levels). This extra bit depth is required for HDR content because HDR displays operate over a much wider luminance range (0.0005 to 10,000 nits for HDR10 vs 0.1 to 100 nits for SDR) that cannot be represented with 8-bit precision without visible banding. Main 10 is the mandatory profile for HDR10, HDR10+, and HLG content. All modern smartphones, Smart TVs, and streaming devices support HEVC Main 10 hardware decoding.

HEVC Main profile suporta 8-bit per channel video com 4:2:0 chroma subsampling — padrão dynamic range (SDR) content. Main 10 profile Adicionalmente suporta 10-bit per channel video, providing 4× more tonal gradations per channel (1024 vs 256 levels). This extra bit depth is required para HDR content because HDR displays operate over a much wider luminance range (0.0005 to 10,000 nits para HDR10 vs 0.1 to 100 nits para SDR) that cannot be represented com 8-bit precision sem visible banding. Main 10 is the mandatory profile para HDR10, HDR10+, e HLG content. All moderno smartphones, smart TVs, e streaming devices support HEVC Main 10 hardware decoding.

HEVC Main profile unterstützt 8-bit per channel video mit 4:2:0 chroma subsampling — Standard dynamic range (SDR) content. Main 10 profile additionally unterstützt 10-bit per channel video, providing 4× more tonal gradations per channel (1024 vs 256 levels). This extra bit depth is required für HDR content because HDR displays operate over a much wider luminance range (0.0005 to 10,000 nits für HDR10 vs 0.1 to 100 nits für SDR) that cannot be represented mit 8-bit precision ohne visible banding. Main 10 is the mandatory profile für HDR10, HDR10+, und HLG content. All modern smartphones, Smart-TVs, und streaming devices support HEVC Main 10 hardware decoding.

HEVC Main profile soporta 8-bit per channel video con 4:2:0 chroma subsampling — estándar dynamic range (SDR) content. Main 10 profile additionally soporta 10-bit per channel video, providing 4× more tonal gradations per channel (1024 vs 256 levels). This extra bit depth is required para HDR content because HDR displays operate over a much wider luminance range (0.0005 to 10,000 nits para HDR10 vs 0.1 to 100 nits para SDR) that cannot be represented con 8-bit precision sin visible banding. Main 10 is the mandatory profile para HDR10, HDR10+, y HLG content. All moderno smartphones, smart TVs, y streaming devices support HEVC Main 10 hardware decoding.

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.

The default CRF for libx265 is 28, which produces smaller files than H.264's default CRF of 23 at equivalent visual quality (because HEVC is more efficient). For general purpose archiving and distribution, CRF 24-28 is recommended. CRF 18-22 gives visually lossless quality for critical content but produces large files. CRF 28-32 is suitable for web delivery where bandwidth is a concern. For 4K HDR content, use CRF 20-24 because high-resolution and HDR video contains more perceptible detail that justifies larger files. The "-preset" flag also matters: "slow" or "slower" gives 10-15% better compression than "medium" but takes 2-4× longer to encode.

The default CRF para libx265 is 28, which produces smaller arquivos than H.264's default CRF of 23 at equivalent visual quality (because HEVC is more efficient). para general purpose archiving e distribution, CRF 24-28 is recommended. CRF 18-22 gives visually sem perdas quality para critical content mas produces large files. CRF 28-32 is suitable para web delivery where bandwidth is a concern. para 4K HDR content, usar CRF 20-24 because alta-resolution e HDR video contém more perceptible detail that justifies larger files. The "-preset" flag also matters: "slow" ou "slower" gives 10-15% better compressão than "medium" mas takes 2-4× longer to encode.

The default CRF für libx265 is 28, which produces smaller Dateien than H.264's default CRF von 23 at equivalent visual quality (because HEVC is more efficient). für general purpose archiving und distribution, CRF 24-28 is recommended. CRF 18-22 gives visually verlustfrei quality für critical content aber produces large files. CRF 28-32 is suitable für web delivery where bandwidth is a concern. für 4K HDR content, verwenden CRF 20-24 because hoch-resolution und HDR video contains more perceptible detail that justifies larger files. The "-preset" flag also matters: "slow" oder "slower" gives 10-15% better Komprimierung than "medium" aber takes 2-4× longer to encode.

The default CRF para libx265 is 28, which produces smaller archivos than H.264's default CRF de 23 at equivalent visual quality (because HEVC is more efficient). para general purpose archiving y distribution, CRF 24-28 is recommended. CRF 18-22 gives visually sin pérdidas quality para critical content pero produces large files. CRF 28-32 is suitable para web delivery where bandwidth is a concern. para 4K HDR content, usar CRF 20-24 because alta-resolution y HDR video contains more perceptible detail that justifies larger files. The "-preset" flag also matters: "slow" o "slower" gives 10-15% better compresión than "medium" pero takes 2-4× longer to encode.

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.

HEVC has one of the most complex patent situations in codec history. Three separate patent pools (HEVC Advance, Velos Media, Via LA) plus individual licensors all claim essential patents — meaning device manufacturers must pay multiple separate fees. HEVC Advance charges approximately $0.60 per device plus revenue-based fees. This complexity caused major tech companies (Google, Mozilla, Microsoft, Amazon, Netflix, Cisco, Apple) to form the Alliance for Open Media and develop AV1 as a royalty-free alternative. For users, this matters primarily when distributing software or selling devices: ffmpeg can encode HEVC freely, but distributing HEVC-encoded content commercially may require licensing depending on jurisdiction and use case. Consumer playback of HEVC is generally covered by device manufacturers' licenses.

HEVC has one of the most complexo patent situations in codec history. Three separate patent pools (HEVC Advance, Velos Media, Via LA) plus individual licensors all claim essential patents — meaning device manufacturers must pay multiple separate fees. HEVC Advance charges approximately $0.60 per device plus revenue-based fees. This complexoity caused major tech companies (Google, Mozilla, Microsoft, Amazon, Netflix, Cisco, Apple) to form the Alliance para abrir Media e develop AV1 como um royalty-free alternative. para users, this matters primarily when distributing software ou selling devices: ffmpeg can encode HEVC grátisly, mas distributing HEVC-encoded content commercially may require licensing depending on jurisdiction e usar case. Consumer playback of HEVC is Geralmente covered by device manufacturers' licenses.

HEVC has one des most complex patent situations in Codec history. Three separate patent pools (HEVC Advance, Velos Media, Via LA) plus individual licensors all claim essential patents — meaning device manufacturers must pay multiple separate fees. HEVC Advance charges approximately $0.60 per device plus revenue-based fees. This complexity caused major tech companies (Google, Mozilla, Microsoft, Amazon, Netflix, Cisco, Apple) to form the Alliance für öffnen Media und develop AV1 als ein royalty-free alternative. für users, this matters primarily when distributing Software oder selling devices: ffmpeg can encode HEVC kostenlosly, aber distributing HEVC-encoded content commercially may require licensing depending on jurisdiction und verwenden case. Consumer playback von HEVC is Generell covered by device manufacturers' licenses.

HEVC has one del most complex patent situations in codec history. Three separate patent pools (HEVC Advance, Velos Media, Via LA) plus individual licensors all claim essential patents — meaning device manufacturers must pay multiple separate fees. HEVC Advance charges approximately $0.60 per device plus revenue-based fees. This complexity caused major tech companies (Google, Mozilla, Microsoft, Amazon, Netflix, Cisco, Apple) to form the Alliance para abrir Media y develop AV1 como un royalty-free alternative. para users, this matters primarily when distributing software o selling devices: ffmpeg can encode HEVC gratisly, pero distributing HEVC-encoded content commercially may require licensing depending on jurisdiction y usar case. Consumer playback de HEVC is Generalmente covered by device manufacturers' licenses.

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.