Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
WebRTC動画をトランスコードする / Transcoding video streams ...
Search
Harukasan
PRO
February 05, 2018
Technology
5
1.7k
WebRTC動画をトランスコードする / Transcoding video streams from WebRTC
SkyWay UG Tokyo #2
2018-02-05
Harukasan
PRO
February 05, 2018
Tweet
Share
More Decks by Harukasan
See All by Harukasan
Successor to PicoRabbit: Ruby Programming Envorinment / RubyKaigi 2025 follow up
harukasan
PRO
1
230
Write your own mrbgem, Create your own device
harukasan
PRO
1
270
PicoRabbit: a Tiny Presentation Device Powered by Ruby
harukasan
PRO
2
660
pixivを支える技術 / 技育CAMPアカデミア
harukasan
PRO
3
570
20240401 新卒研修 - ピクシブにおける技術領域
harukasan
PRO
1
910
ピクシブのコンテンツ配信基盤技術 / pixiv TECH SALON
harukasan
PRO
5
5.8k
Goにおける画像ファイル処理 / golang.tokyo #19
harukasan
PRO
7
6.8k
ImageFluxを支えるリモート開発 / 20171202
harukasan
PRO
2
1.9k
YAPC::Fukuoka 前夜祭LT / Yet Another Pawoo Commit logs
harukasan
PRO
0
3.1k
Other Decks in Technology
See All in Technology
モブプログラミング再入門 ー 基本から見直す、AI時代のチーム開発の選択肢 ー / A Re-introduction of Mob Programming
takaking22
5
800
組織のSREを推進するためのPlatform EngineeringとEKS / Platform Engineering and EKS to drive SRE in your organization
chmikata
0
190
[JAWSDAYS2026]Who is responsible for IAM
mizukibbb
0
150
Abuse report だけじゃない。AWS から緊急連絡が来る状況とは?昨今の攻撃や被害の事例の紹介と備えておきたい考え方について
kazzpapa3
1
200
白金鉱業Meetup_Vol.22_Orbital Senseを支える衛星画像のマルチモーダルエンベディングと地理空間のあいまい検索技術
brainpadpr
2
260
AIエージェント・エコノミーの幕開け 〜 オープンプロトコルが変えるビジネスの未来 〜
shukob
0
110
vLLM Community Meetup Tokyo #3 オープニングトーク
jpishikawa
0
220
ブラックボックス観測に基づくAI支援のプロトコルのリバースエンジニアリングと再現~AIを用いたリバースエンジニアリング~ @ SECCON 14 電脳会議 / Reverse Engineering and Reproduction of an AI-Assisted Protocol Based on Black-Box Observation @ SECCON 14 DENNO-KAIGI
chibiegg
0
160
マルチアカウント環境でSecurity Hubの運用!導入の苦労とポイント / JAWS DAYS 2026
genda
0
140
Kaggleの経験が実務にどう活きているか / kaggle_findy
sansan_randd
7
1.3k
越境する組織づくり ─ 多様性を前提にしたチームビルディングとリードの実践知
kido_engineer
2
150
型を書かないRuby開発への挑戦
riseshia
0
200
Featured
See All Featured
How to Think Like a Performance Engineer
csswizardry
28
2.5k
Pawsitive SEO: Lessons from My Dog (and Many Mistakes) on Thriving as a Consultant in the Age of AI
davidcarrasco
0
80
The Illustrated Children's Guide to Kubernetes
chrisshort
51
52k
Balancing Empowerment & Direction
lara
5
930
Building Better People: How to give real-time feedback that sticks.
wjessup
370
20k
Designing for Timeless Needs
cassininazir
0
150
ラッコキーワード サービス紹介資料
rakko
1
2.6M
Evolving SEO for Evolving Search Engines
ryanjones
0
150
Leveraging LLMs for student feedback in introductory data science courses - posit::conf(2025)
minecr
1
190
AI: The stuff that nobody shows you
jnunemaker
PRO
3
360
Efficient Content Optimization with Google Search Console & Apps Script
katarinadahlin
PRO
1
370
How Fast Is Fast Enough? [PerfNow 2025]
tammyeverts
3
480
Transcript
Transcoding video streams from WebRTC MICHII Shunsuke pixiv Inc. SkyWay
UG Tokyo #2 2018-02-05
MICHII Shunsuke | @harukasan ImageFluxࣄۀ෦ ෦ ϐΫγϒςΫϊϩδʔζגࣜձࣾࣥߦһ ImageFluxͰGoΛॻ͍ͨΓΠϯϑϥपΓΛͬͨΓͱ͔͍Ζ͍Ζ
ࠓͳ͢͜ͱ • WebRTCͰૹΒΕͯ͘ΔH.264/AVCಈըετϦʔϜΛऔΓग़͢ํ๏ • H.264/AVCಈըετϦʔϜͲͷΑ͏ʹૹΒΕͯ͘Δͷ͔
ࠓͳ͞ͳ͍͜ͱ • WebRTCͱͳʹ͔ • WebRTCͰಈըΛετϦʔϛϯά͢Δํ๏ • WebRTC SFU Sora͕ͬͯ͘ΕΔ͜ͱ ଓཱ֬ɺηογϣϯͷཧɺSRTPͷ҉߸Խղআ……
• SkyWayͰͲ͏ͬͯΔͷ͔……ʢRTPΛऔΓग़ͤΕେৎʣ • H.264/AVCҎ֎ͷVP9, HEVC, AV1ͱ͍ͬͨಈըϑΥʔϚοτʹ͍ͭͯ
ͳͥWebRTCಈըΛτϥϯείʔυ͢Δͷ͔ • WebRTCಈըΛHLS৴͢ΔαʔϏεΛ։ൃதʂ https://speakerdeck.com/imageflux/imageflux-live • pixiv Sketch LiveͷόοΫΤϯυͱͯ͠Քಇத • αʔϏεʹڵຯ͋Δํੋඇ͓͕͚͍ͩ͘͞ʂ
ΤϯδχΞืू͍ͯ͠·͢ʂ
ImageFlux LiveͷΈ Transcoder Transcoder Transcoder RTP stream Streamerd Msgpack stream
TS TS TS ৴ऀ WebRTC SFU Sora WebRTC API H.264 bytestreamΛτϥϯείʔυ͢Δ RTPετϦʔϜΛσίʔυͯ͠ H.264 bytestream formatʹͯ͠ msgpackʹͭΊΔ
H.264 bytestreamΛτϥϯείʔυ͢Δ ImageFlux LiveͷΈ Transcoder Transcoder Transcoder RTP stream Msgpack
stream TS TS TS ৴ऀ WebRTC SFU Sora WebRTC API RTPετϦʔϜΛσίʔυͯ͠ H.264 bytestream formatʹͯ͠ msgpackʹͭΊΔ Streamerd
WebRTCͰԿΛ௨৴͍ͯ͠Δͷ͔ Client WebRTC Client ηογϣϯཧ ϦΞϧλΠϜετϦʔϛϯά ଓཱ֬ʢNATӽ͑ʣ
WebRTCͰԿΛ௨৴͍ͯ͠Δͷ͔ Client WebRTC Client ηογϣϯཧ ϦΞϧλΠϜετϦʔϛϯά ଓཱ֬ʢNATӽ͑ʣ
WebRTCͰԿΛ௨৴͍ͯ͠Δͷ͔ Client WebRTC Client SDP SRTP ηογϣϯใ ಈըετϦʔϜ
WebRTCͰૹΒΕͯ͘ΔಈըΛ τϥϯείʔυ͢Δʹ 1. ΫϥΠΞϯτͱηογϣϯใʢSDPʣΛΓͱΓ͢Δ 2. ҉߸Խ͞Ε͍ͯΔετϦʔϜʢSRTPʣΛ҉߸Խղআ͢Δ 3. ಈըετϦʔϜʢRTPʣΛσίʔυ͢Δ 4. औΓग़ͨ͠H.264ετϦʔϜΛτϥϯείʔυ͢Δ
WebRTCͰૹΒΕͯ͘ΔಈըΛ τϥϯείʔυ͢Δʹ 1. ΫϥΠΞϯτͱηογϣϯใʢSDPʣΛΓͱΓ͢Δ 2. ҉߸Խ͞Ε͍ͯΔετϦʔϜʢSRTPʣΛ҉߸Խղআ͢Δ 3. ಈըετϦʔϜʢRTPʣΛσίʔυ͢Δ 4. औΓग़ͨ͠H.264ετϦʔϜΛτϥϯείʔυ͢Δ
1, 2WebRTC SFU Sora͕ͬͯ͘Ε·͢
RTP (Realtime Transfer Protocol) RFC3550 RTP: A Transport Protocol for
Real-Time Applications • ԻɺಈըͳͲͷσʔλΛϦΞϧλΠϜʹసૹ͢ΔͨΊͷϓϩτίϧ • UDP্ʹ࣮͞Ε͍ͯΔ • TCPΛ͍ͬͯͳ͍ͷͰɺॱং੍ޚ੍ޚΛRTPϨΠϠͰ࣮Ͱ͖Δ →ݴ͍͑ΕࣗͰ࣮͢Δඞཁ͕͋Δ • SRTPʢͬ͘͟Γݴ͑ʣRTPΛ҉߸Խͨ͠ͷ
࣮ମUDPͷബ͍ϥούʔ • UDPʹബ͍ϔομ͕͍͍ͭͯΔ͚ͩ • ϖΠϩʔυͷූ߸ԽํࣜϖΠϩʔυͷछྨ͝ͱʹҟͳΔ 0 1 2 3 4
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 V P X CC M payload type sequence number timestamp synchronization source (SSRC) identifier contributing source (CSRC) identifier payload ….
ϖΠϩʔυͷผํ๏ • ϖΠϩʔυͷϑΥʔϚοτpayload typeͰผ͢Δ • payload type7bit͔͠ͳ͘ɺ͙͢ʹΓͳ͘ͳͬͨͷͰɺݱࡏͰ 96-127ΛಈతʹׂΓͯͯ͏ʢRFC3551ʣ • payload
typeͱϑΥʔϚοτͷϚοϐϯάSDPͷattributeͱͯ͠ ΓͱΓ͞ΕΔ • طʹొ͞Ε͍ͯΔͷIANAͰཧ͞Ε͍ͯΔ https://www.iana.org/assignments/rtp-parameters/rtp-parameters.xhtml
ॱংอূ • RTPUDPͰૹΒΕͯ͘ΔͷͰॱ൪อূ͞Ε͍ͯͳ͍ • RTPϔομʹγʔέϯε൪߸ʢsequence numberʣؚ͕·Ε͍ͯΔͷ Ͱ͜ΕΛͬͯιʔτ͢Δඞཁ͕͋Δ • ;ͭ͏ʹ300ݸ͙Β͍ޙΕͯύέοτ͕ඈΜͰ͖ͨΓͯ͠ਏ͍……͕ ͋·ΓͭͱԆͷݪҼʹͳΔͷͰέʔεόΠέʔε
• ͍·ͷ࣮Ͱύέοτ͕ඈΜͩ߹100ݸ͘Β͍ͭΑ͏ʹ͍ͯ͠Δ
ϑΥʔϚοτ͝ͱʹϖΠϩʔυͷܗࣜҟͳΔ • ͦΕͧΕҟͳΔRFCͰඪ४Խ͞Ε͍ͯΔ RFC6184 RTP Payload Format for H.264 Video
RFC7587 RTP Payload Format for the Opus Speech and Audio Codec draft-uberti-payload-vp9-01 RTP Payload Format for VP9 Video
RTP Payload Format for H.264 Video • H.264 NALU (NAL
Unitʣ1͕ͭ1ͭͷUDPύέοτʹؚ·ΕΔ 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 V P X CC M payload type sequence number timestamp synchronization source (SSRC) identifier contributing source (CSRC) identifier NAL Header Payload….
H.264/AVC NAL (Network Abstraction Layer) • H.264/AVCͰNALʢNetwork Abstraction Layerʣ͕ఆٛ͞Ε͍ͯΔ •
NALಈըετϦʔϜΛ͍ΖΜͳϝσΟΞͰૹ͍͢͠Α͏ʹநԽ͠ ͍ͯΔϨΠϠʔ • ؆୯ʹ͍͏ͱɺಈըετϦʔϜΛNAL UnitʢNALUʣͱݺΕΔͻͱ͔ͨ ·Γʹܗ͍ͯ͠Δ
H.264/AVC Bytestream Format (Annex B) • ΄ͱΜͲͷσίʔμH.264/AVC Bytestream Format (Annex
B)Λೖྗͱ ͯ͠͏͚͚ͭΔ • ISO/IEC 14496-10ͷAnnex Bʹࡌ͍ͬͯͨͷͰAnnex B formatͱݺΕΔ • جຊతʹNALUΛ00 00 00 01Ͱܨ͛ΕΑ͍ ͭ·ΓɺRTPͷϖΠϩʔυΛܨ͛Ε͍͍ͩͨΑͦ͞͏ 00 00 00 01 NALU 00 00 00 01 NALU 00 00 00 01 NALU …
ׂύέοτͱ݁߹ύέοτ • UDPͷϑϨʔϜαΠζҰൠతʹ1500byte͔͠ͳ͍ͷͰɺ݁߹ͨ͠Γɺ ׂͯ͠ෳͷύέοτʹೖΕͨΓ͢Δ͜ͱ͕͋Δ • ݁߹ύέοτͱɺׂύέοτNAL Unit typeͷ24-29Ͱఆٛ͞Ε͍ͯΔ NAL Unit
Packet Packet Type Name Section Type Type ------------------------------------------------------------- 1-23 NAL unit Single NAL unit packet 5.6 24 STAP-A Single-time aggregation packet 5.7.1 25 STAP-B Single-time aggregation packet 5.7.1 26 MTAP16 Multi-time aggregation packet 5.7.2 27 MTAP24 Multi-time aggregation packet 5.7.2 28 FU-A Fragmentation unit 5.8 29 FU-B Fragmentation unit 5.8 https://tools.ietf.org/html/rfc6184
ׂͨ͠Γ݁߹ͨ͠Γ͢Δ • ඇΠϯλʔϦʔϒϞʔυͰSTAP-AͱFU-A͕ΘΕΔͷͰ͜ͷ2ݸ͚ͩ ରԠ͢ΕΑ͍ • STAP-A: ෳͷNALUΛαΠζΛઌ಄ʹ͚ͭͯ݁߹ͨ͠ύέοτ • FU-A: ෳͷNALUΛׂ͠ɺ֤ύέοτʹFUϔομΛ͚ͭͨͷ
جຊతʹstart bit (s)͕͍͍ͭͯΔύέοτ͔Βend bit (e)͕͍͍ͭͯΔύ έοτ·Ͱ݁߹͢Δ 0 1 2 3 4 5 6 7 8 S E - Type FUϔομ
RTP͔ΒH.264 byte streamΛσίʔυ͢Δʹ 1. SDPΛಡΜͰpayload typeͱϑΥʔϚοτͷϚοϐϯάΛऔΓग़͢ 2. RTPύέοτΛόοϑΝϦϯάͯ͠ɺγʔέϯε൪߸ॱʹฒͼସ͑Δ 3. ݁߹ύέοτͩͬͨ߹ׂͯ͠ෳͷόΠτετϦʔϜʹ͢Δ
4. ׂύέοτͩͬͨ߹શύέοτ͕όοϑΝϦϯά͞ΕΔͷΛ·ͬ ͯ1ͭͷNALUΛऔΓग़͢
SDP (Session Description Protocol) RFC4566 SDP: Session Description Protocol •
ηογϣϯใΛදݱ͢ΔͨΊͷϓϩτίϧ • WebRTCΛ͡ΊɺʢWebRTCͷϕʔεʹͳͬͨʣSIPɺRTSPɺϝʔϧͳͲ ͘ΘΕ͍ͯΔ • ಡΈͮΒ͍͜ͱͰ༗໊ʁ
ಡΈͮΒ͍ʁ v=0 o=jdoe 2890844526 2890842807 IN IP4 10.47.16.5 s=SDP Seminar
i=A Seminar on the session description protocol u=http://www.example.com/seminars/sdp.pdf
[email protected]
(Jane Doe) c=IN IP4 224.2.17.12/127 t=2873397496 2873404696 a=recvonly m=audio 49170 RTP/AVP 0 m=video 51372 RTP/AVP 99 a=rtpmap:99 h263-1998/90000 https://tools.ietf.org/html/rfc4566
ಡΈͮΒ͍ʁ v=0 o=jdoe 2890844526 2890842807 IN IP4 10.47.16.5 s=SDP Seminar
i=A Seminar on the session description protocol u=http://www.example.com/seminars/sdp.pdf
[email protected]
(Jane Doe) c=IN IP4 224.2.17.12/127 t=2873397496 2873404696 a=recvonly m=audio 49170 RTP/AVP 0 m=video 51372 RTP/AVP 99 a=rtpmap:99 h263-1998/90000 https://tools.ietf.org/html/rfc4566
Media Description "m=" m=<media> <port> <proto> <proto> <fmt> … <media>:
ϝσΟΞͷछྨʢaudio, video,…) <port>: ૹ৴ݩϙʔτ <proto>: ϓϩτίϧ <fmt>: ϑΥʔϚοτใ ϝσΟΞͷछྨͱϑΥʔϚοτɺૹ৴ݩϙʔτؚ͕·Ε͍ͯΔ ͋·Γ͑Δใͳ͍
Attributes "a=" a=<attribute>:<value> <attribute>: ଐੑ <value>: ͦͷ • SDPʹΛ֦ு͢ΔͨΊʹ༻ҙ͞Ε͍ͯΔύϥϝʔλ •
WebRTCͰ֦ு͞Ε͍ͯΔ͍͍ͩͨa=ʹೖ͍ͬͯΔ
SDP͔ΒRTPͷpayload type mappingΛऔΓग़͢ a=rtpmap:109 opus/48000/2 a=rtpmap:120 H264/90000 • payload typeͷmappingrtpmapͰࢦఆ͞ΕΔ
• ্ͷྫͩͱɺH.264ͷpayload type120ɺOpus109 • 2൪ͷϑΥʔϚοτ/प/νϟϯωϧ H.264/AVCͩͱ90000KHzͰλΠϜελϯϓ͕ͬͯ͘Δ͜ͱ͕Θ͔Δ
RTP͔ΒH.264 byte streamΛσίʔυ͢Δʹ 1. SDPΛಡΜͰpayload typeͱϑΥʔϚοτͷϚοϐϯάΛऔΓग़͢ 2. RTPύέοτΛόοϑΝϦϯάͯ͠ɺγʔέϯε൪߸ॱʹฒͼସ͑Δ 3. ݁߹ύέοτͩͬͨ߹ׂͯ͠ෳͷόΠτετϦʔϜʹ͢Δ
4. ׂύέοτͩͬͨ߹શύέοτ͕όοϑΝϦϯά͞ΕΔͷΛ·ͬ ͯ1ͭͷNALUΛऔΓग़͢ ෮श
·ͱΊ • WebRTCಈըετϦʔϜΛτϥϯείʔυ͢Δʹɺ RTPύέοτΛσίʔυͰ͖ΕΑ͍ • αʔόαΠυͰRTPΛऔΓग़͢ʹWebRTC SFU Sora͕ศར • ImageFlux
Live։ൃதͰ͢ʂ