{"id":114343,"date":"2024-03-15T18:25:48","date_gmt":"2024-03-15T16:25:48","guid":{"rendered":"https:\/\/prohoster.info\/blog\/novosti-interneta\/vypusk-kodirovshhika-svt-av1-2-0-i-dekodirovshhika-dav1d-1-4-dlya-formata-video-av1"},"modified":"2024-03-15T18:25:48","modified_gmt":"2024-03-15T16:25:48","slug":"vypusk-kodirovshhika-svt-av1-2-0-i-dekodirovshhika-dav1d-1-4-dlya-formata-video-av1","status":"publish","type":"post","link":"https:\/\/prohoster.info\/sq\/blog\/novosti-interneta\/vypusk-kodirovshhika-svt-av1-2-0-i-dekodirovshhika-dav1d-1-4-dlya-formata-video-av1","title":{"rendered":"L\u00ebshimi i kodifikuesit SVT-AV1 2.0 dhe dekodifikuesit dav1d 1.4 p\u00ebr formatin e videos AV1","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"
\u00cbsht\u00eb publikuar \u00e7lirim i bibliotek\u00ebs SVT-AV1 2.0 (Scalable Video Technology AV1) me zbatime t\u00eb koduesit dhe dekoduesit t\u00eb formatit t\u00eb kodimit t\u00eb videove AV1, p\u00ebr t\u00eb cilin p\u00ebrdoren burimet e p\u00ebrshpejtimit t\u00eb harduerit t\u00eb pranishme n\u00eb CPU-t\u00eb moderne t\u00eb Intel. Projekti \u00ebsht\u00eb krijuar nga Intel n\u00eb partneritet me Netflix me q\u00ebllimin p\u00ebr t\u00eb arritur nj\u00eb nivel performanc\u00eb t\u00eb p\u00ebrshtatsh\u00ebm p\u00ebr rikodimin e videove n\u00eb koh\u00eb reale dhe p\u00ebr p\u00ebrdorim n\u00eb sh\u00ebrbimet q\u00eb ofrojn\u00eb video me k\u00ebrkes\u00eb (VOD). Aktualisht, zhvillimi po kryhet n\u00ebn mbik\u00ebqyrjen e aleanc\u00ebs Open Media (AOMedia), q\u00eb mbik\u00ebqyr zhvillimin e formatit t\u00eb kodimit t\u00eb videove AV1. M\u00eb par\u00eb, projekti ishte zhvilluar brenda projektit OpenVisualCloud, i cili gjithashtu zhvillon koduesit SVT-HEVC dhe SVT-VP9. Kodi shp\u00ebrndahet n\u00ebn licenc\u00ebn BSD. <\/p>\n
SVT-AV1 mund t\u00eb b\u00ebhet p\u00ebr sistemet q\u00eb zbatojn\u00eb \u00e7do arkitektur\u00eb, p\u00ebr t\u00eb cilat ka nj\u00eb kompilues q\u00eb mb\u00ebshtet standardin C99, por performanca m\u00eb e mir\u00eb arrihet n\u00eb sistemet x86_64, p\u00ebr t\u00eb cilat p\u00ebrdoren optimizime assembly bazuar n\u00eb instrukcionet SIMD (preferohet q\u00eb CPU t\u00eb ket\u00eb mb\u00ebshtetje p\u00ebr AVX2, por si minimum mjafton edhe SS2). Konsumi i memories varet nga numri i b\u00ebrthamave t\u00eb procesorit t\u00eb angazhuara gjat\u00eb kodimit, t\u00eb rregulluara nga opsioni \u00ab\u2014lp\u00bb. P\u00ebr shkak t\u00eb kompleksitetit t\u00eb algoritmeve n\u00eb AV1, kodimi i k\u00ebtij formati k\u00ebrkon n\u00eb m\u00ebnyr\u00eb t\u00eb konsiderueshme m\u00eb shum\u00eb burime se formatet e tjera, q\u00eb nuk lejon p\u00ebrdorimin e koduesit standard AV1 p\u00ebr rikodim n\u00eb koh\u00eb reale. P\u00ebr shembull, koduesi standard nga projekti AV1 k\u00ebrkon 5721, 5869 dhe 658 her\u00eb m\u00eb shum\u00eb llogaritje krahasuar me koduesit x264 (profili \u00abmain\u00bb), x264 (profili \u00abhigh\u00bb) dhe libvpx-vp9. <\/p>\n
Nd\u00ebr ndryshimet n\u00eb versionin e ri t\u00eb SVT-AV1: <\/p>\n
\n
\u00cbsht\u00eb b\u00ebr\u00eb kalimi n\u00eb nj\u00eb num\u00ebrim t\u00eb ri versionesh, sipas t\u00eb cilit numri i par\u00eb n\u00eb version do t\u00eb ndryshoj\u00eb me \u00e7do ndryshim API\/ABI.\n
Jan\u00eb b\u00ebr\u00eb ndryshime n\u00eb API t\u00eb lidhura me kalimin n\u00eb indikimin e p\u00ebrfundimit t\u00eb rrjedh\u00ebs (EOS \u2014 End Of Stream) n\u00eb kadrin e fundit n\u00eb vend t\u00eb p\u00ebrdorimit t\u00eb nj\u00eb kadrin bosh, duke mund\u00ebsuar k\u00ebshtu eliminimin e vones\u00ebs p\u00ebr pritjen e kadrave t\u00eb panevojshme. Ndryshimi n\u00eb API tashm\u00eb \u00ebsht\u00eb reflektuar n\u00eb baz\u00ebn e kodit t\u00eb FFmpeg.\n
\u00cbsht\u00eb hequr moda e tre kalimeve p\u00ebr bitratein e ndryshuesh\u00ebm (3-pass VBR), n\u00eb vend t\u00eb saj tani p\u00ebrdoret mekanizmi i VBR me shum\u00eb kalime. Moda shum\u00ebkalimike e VBR \u00ebsht\u00eb reduktuar n\u00eb ekzekutimin e dy kalimeve p\u00ebr t\u00eb siguruar integrimin me FFmpeg.\n
Jan\u00eb shtuar optimizime n\u00eb kodues, si rezultat i t\u00eb cilave efikasiteti i kompresimit t\u00eb preset\u00ebve M9-M13 \u00ebsht\u00eb rritur me 1-4%, nd\u00ebrsa konsumimi i memories n\u00eb presetin M5 \u00ebsht\u00eb zvog\u00ebluar me 20-35% n\u00eb modin LP 8 dhe me 1-5% n\u00eb modet e tjera. Konsumimi i memories n\u00eb presetet e tjera \u00ebsht\u00eb zvog\u00ebluar me 1-5%.\n
\u00cbsht\u00eb b\u00ebr\u00eb optimizimi i kompromisit cil\u00ebsi\/shpejt\u00ebsi p\u00ebr presetet q\u00eb vendosin nj\u00eb nivel t\u00eb lart\u00eb cil\u00ebsie. Puna e presetit MR, q\u00eb ofron cil\u00ebsi referuese, \u00ebsht\u00eb p\u00ebrshpejtuar me 100%.\n
Nj\u00eb tjet\u00ebr gj\u00eb q\u00eb mund t\u00eb p\u00ebrmendet \u00ebsht\u00eb lansimi i projektit dav1d 1.4.1, ku komunitetet VideoLAN dhe FFmpeg po zhvillojn\u00eb bibliotek\u00ebn me realizimin e nj\u00eb dekoduesi alternativ t\u00eb lir\u00eb p\u00ebr formatin e kodimit t\u00eb videos AV1. Biblioteka dav1d mb\u00ebshtet t\u00eb gjitha funksionalitetet e AV1, duke p\u00ebrfshir\u00eb llojet e avancuara t\u00eb subdysbikimit dhe t\u00eb gjitha parametrat e kontrollit t\u00eb thell\u00ebsis\u00eb s\u00eb ngjyr\u00ebs (8, 10 dhe 12 bit) t\u00eb shpallura n\u00eb specifikim. Nj\u00eb tipar ky\u00e7 i dav1d \u00ebsht\u00eb orientimi drejt arritjes s\u00eb performanc\u00ebs maksimale t\u00eb dekodimit dhe siguroi nj\u00eb funksionim cil\u00ebsor n\u00eb modalitetin me shum\u00eb fjede. Kodi i projektit \u00ebsht\u00eb shkruar n\u00eb gjuh\u00ebn C (C99) me p\u00ebrfshirje asamble (NASM\/GAS) dhe shp\u00ebrndahet n\u00ebn licenc\u00ebn BSD. Supportohet arkitekturat x86, x86_64, ARMv7 dhe ARMv8, si dhe sistemet operative FreeBSD, Linux, Windows, macOS, Android dhe iOS. <\/p>\n
Versioni dav1d 1.4 \u00ebsht\u00eb e r\u00ebnd\u00ebsishme p\u00ebr mb\u00ebshtetje t\u00eb arkitekturave t\u00eb reja Loongarch dhe RISC-V, si dhe p\u00ebr p\u00ebrdorimin e optimizimeve t\u00eb tjera mbi baz\u00ebn e instruksioneve AVX-512, p\u00ebrshpejtimin e pun\u00ebs s\u00eb filtrave 6tap n\u00eb sistemet ARM, rritjen e efikasitetit t\u00eb pun\u00ebs \u043c\u043d\u043e\u0433\u043e\u043fot\u00ebshe dhe uljen e madh\u00ebsis\u00eb s\u00eb t\u00eb dh\u00ebnave binare n\u00eb sistemet ARM64, ARM32 dhe RISC-V. \u00cbsht\u00eb eliminuar kjo dob\u00ebsi CVE-2024-1580, q\u00eb \u00e7onte n\u00eb shkrimin jasht\u00eb kufijve t\u00eb tamponit p\u00ebr shkak t\u00eb mbushjes integer gjat\u00eb p\u00ebrpunimit t\u00eb kadrove t\u00eb konsiderueshme. \n Burimi: opennet.ru<\/a> <\/p>","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"excerpt":{"rendered":"