Lawa njengamanje amawebhusayithi angcono kakhulu okuthola amavidiyo akhiqizwe yi-AI

Siphila esikhathini lapho i- amavidiyo akhiqizwe ubuhlakani bokwenziwa Bangena ezinkundleni zokuxhumana, izinhlelo zokusebenza zemiyalezo, nezindaba ngesivinini sombani, futhi akulula ngaso sonke isikhathi. hlukanisa ukolweni namakhobaIzindaba ezinhle ukuthi namuhla kunezimpawu, izindlela, namathuluzi asiza ukuhlukanisa phakathi kokuqukethwe okuyiqiniso nezinto zokwenziwa noma ezisetshenziswayo. Amawebhusayithi okuthola amavidiyo adalwe nge-AI ngisho nalapho umphumela ubonakala ungenasici ekuqaleni.

Lesi sihloko sihlanganisa, ngendlela esebenzayo nebanzi kakhulu, okungcono kakhulu esikubone kuwebhu ukuthola amavidiyo adalwe nge-AI: izinkomba ezibonakalayo, ukuhlaziya imethadatha, izinkundla zamahhala nezomsebenzi, ngisho nezincomo ezisemthethweni nezokuqinisekisa okwenziwa ngesandla.

Ithini ividiyo ekhiqizwe yi-AI futhi kungani ibalulekile?

Uma sikhuluma ngamavidiyo e-AI, sibhekisela ezincekwini ezilalelwayo nezibukwayo ezidalwe noma ezishintshwe ngamamodeli akhiqizayo nezindlela ezithuthukile (njengama-deepfakes, umbhalo-kuya-vidiyo, noma ama-avatar e-hyperrealistic). Kungaba iziqeshana zokwenziwa ngokuphelele noma amavidiyo wangempela anamasegimenti aguquliweIsibonelo, ngokushintsha ubuso ngokukholisayo noma ngokuhlanganisa izwi.

Ukuhambisana kusobala: lokhu okuqukethwe kungase kungabizi lutho, kulawule imibono, noma konakalise isithunzi. Ngokocwaningo olucashunwe yi-Amazon Web ServicesIngxenye enkulu yokuqukethwe ku-inthanethi isivele yenziwe nge-AI, okwandisa ukuphuthuma kwamakhono namathuluzi okuqinisekisa athembekile.

Obunye ubuchwepheshe sebaziwa kakhulu. I-Sora, ijeneretha yevidiyo emenyezelwe yi-OpenAIIthembisa imiphumela engokoqobo ekhulayo, futhi izinkundla ezifana ne-Runway ne-Pika Labs zivumela abasebenzisi ukuthi bakhe iziqeshana ezivela embhalweni. Ngaleso sikhathi, izinsizakalo ze-avatar ezifana ne-Synthesia zinikeza abethuli bedijithali abanengqondo kakhulu, futhi akukho ukushoda kwabahleli be-AI abathinta kabusha okuqoshiwe okuyiqiniso ngemiphumela emangalisayo. Ukucacisa le mephu kukusiza ukuthi uqonde ukuthi ungabheka kuphi lapho kuphakama izinsolo.

Izimpawu ezibonakalayo nezizwakalayo ezikhaphela ividiyo yokwenziwa

Ngaphambi kokufuna usizo kumawebhusayithi ukuze uthole amavidiyo akhiqizwe yi-AI, isihlungi sakho sokuqala kufanele sibe ukubhekwa. Nakuba amamodeli ethuthuka, amaphutha noma imikhondo ecashile isavela uma wazi ukuthi ubheke kuphi. Lezi yizimpawu ezijwayelekile kumavidiyo akhiqiziwe noma ashintshiwe:

• Ukuvumelanisa izindebe ezingabazekayoUkunyakaza komlomo akufani nomsindo.
• Ukugqolozela okuxakile nokucwayiza: amehlo omile, ukugqolozela, noma ukucwayiza okungajwayelekile.
• Ukukhanya okungahambisani nethunzi: ukuboniswa okungahambisani, ingemuva "eliphefumulayo".
• Isimo sobuso esingezona ezemveloLapho uhleka, umemeza, noma ubonisa imizwa enamandla, kukhona okuklabalasayo.
• Izandla neminwe enenkinga: i-anatomy engalungile ngokucashile noma ukuthinta okungenakwenzeka.
• I-aesthetics "ephelele kakhulu".: ubunono obungahambisani nomongo wevidiyo.

Ukuba sobala kokuqukethwe nakho kuyabalulekile: umongo ongakholeki noma isehlakalo esimangalisa ngokweqile sidinga ukuqinisekiswa okukabili. Uma kubonakala kungakholakali noma kulula kakhulu, sola.Qhathanisa imithombo bese ubheka ezinye izimpawu.

Isebenza kanjani i-AI-powered video detector

Izitholi zesimanje zihlanganisa ukufunda komshini, i-digital forensics, nokuhlolwa kwemethadatha. Eziphelele kakhulu zihlola amaleveli amaningana evidiyo. ukukhomba amaphethini iso lomuntu eligejayo.

1. Layisha noma uxhume kuvidiyoUngalayisha ifayela noma unamathisele i-URL eqondile ukuze uqale ukuhlolwa.
2. Ukuhlaziywa kwamapharamitha amaningi: ukungaguquguquki okubonakalayo, amaphethini okunyakaza, ama-artifact edijithali, amasiginesha imethadatha, nokulandelelwa kokuminyanisa.
3. Umbiko Wobuqiniso: amaphuzu angenzeka, incazelo yokutholiwe kanye, uma kusebenza, imephu yokushisa yezindawo ezisolisayo.
4. Ukuhlukaniswa kozimele ngozimele: kuyasiza uma kufanele ubhekisise lapho kuvela khona okudidayo.

Amanye amawebhusayithi athola amavidiyo adalwe yi-AI ayawacubungula ngesikhathi sangempela noma emizuzwini embalwa, ngisho namavidiyo ayinkimbinkimbi. Ukunemba okuphezulu kushiwo ezimweni ezithile (ngaphezulu kwama-95%).Kodwa-ke, kufanelekile ukukhumbula ukuthi alukho uhlelo olungenaphutha nokuthi imiphumela incike kakhulu ohlotsheni lokukhohlisa, ikhwalithi yefayela, kanye nobude besikhathi.

Amathuluzi namawebhusayithi okuthola amavidiyo akhiqizwe yi-AI

Kumawebhusayithi okuthola amavidiyo adalwe yi-AI, kunezinketho zamahhala nezikhokhelwayo, izinga elilula noma elichwepheshile. Lezi zinkundla nezinsiza sezizuze amandla:

Iskena se-Deepware

I-Deepware Inikeza isithwebuli samahhala esinenketho yezinhlelo ezithuthukile. Ikuvumela ukuthi ulayishe ividiyo noma unamathisele isixhumanisi. futhi ibuyisela isinqumo sayo emizuzwini embalwa, kuye ngokuthi ubude besikhathi nomthwalo wesistimu.

Attestiv.Video

Inguqulo yamahhala (ngokubhaliswa) kwe Imikhosi Ikukhawulela ekuhlaziyeni okumbalwa ngenyanga namavidiyo amafushane, kodwa Ikhiqiza umbiko wobuqiniso onamaphuzu asuka ku-0 kuye ku-100.Ukuhlola okuhlukahlukene kubonisa ukuthi izibalo ezingaphezulu kuka-85/100 ziphakamisa amathuba aphezulu okukhohlisa, ngamamephu okushisa agqamisa ukungahambisani (isb., ukucwayiza noma ukukhonta kwezinwele).

I-InVID WeVerify

Akuwona "ukhiye owodwa", kodwa isandiso sesiphequluli se hlukanisa ividiyo ibe ngozimele abangukhiye, hlaziya izithombe, futhi ulandele umsuka. I-InVID WeVerify Kubalulekile ezintathelini nabahlola amaqiniso abafuna ukuhlola mathupha.

Uhlelo olunamandla e-AI ngokumelene nesizukulwane esigcwele: akufani

Kubalulekile ukuhlukanisa phakathi kwe-AI lokho isheshisa ukuhlela kanye ne-AI ekhiqiza ividiyo ephelele. Amathuluzi afana ne-Descript, Filmora, noma i-Adobe Premiere Pro asebenzisa i-AI ukuze ahlanze umsindo, asuse ukuthula, noma enze uhlaka kabusha, ngaphandle kokudala ividiyo kusukela ekuqaleni.

Isinyathelo esimaphakathi siqukethe izixazululo lokho khiqiza izingxenye eziyingxenye (izikripthi, ama-avatar akhulumayo noma ama-montages anezinto ezigcinwe kungobo yomlando), njenge-Google Vids, Pictory noma i-Synthesia, ezidinga ukuthinta kabusha ngesandla.

Ukweqa kokugcina ukuthembeka okuphezulu kombhalo kuya kuvidiyo, lapho uthayipha okufunayo futhi uthole isiqeshana esiseduze kokugcina. Uma lesi sigaba sesisakazeke ngokugcwele, inselele yokuqinisekisa izoba nkulu nakakhulu. futhi inhlanganisela yamasignali namathuluzi kuyoba ebalulekile.

Imikhuba emihle yokuhlola impilo yansuku zonke

Ngale kwezitholi namawebhusayithi okuthola amavidiyo adalwe nge-AI, ukucabanga okujulile kubalulekile. Sebenzisa le mikhuba Ukunciphisa ubungozi:

• Qaphela noma yini ethusayo uze uyiqinisekise ngemithombo ethembekile.
• Bheka umthombo: amaphrofayela asemthethweni, iziteshi zangempela, idethi yokushicilela, nomongo.
• Phinda ubukele, unake amehlo, izindebe, izandla, izithunzi, nokunyakaza kwekhamera.
• Xhumana nabahloli bamaqiniso abafana ne-Chequeado, i-AFP Factual noma i-Snopes uma ividiyo isakazwa kakhulu.
• Faka isandiso se-InVID uma usebenzisa ulwazi oluningi kumanethiwekhi futhi udinga ukulihlunga ngokushesha.

Lezi zinqubo, ezihlanganiswe nethuluzi lokuhlaziya lapho kudingeka, Banikeza isivikelo esiqinile ekukhohlisweni komsindo nokubonwayo. ngaphandle kokukhathazeka noma ukuwela ku-paranoia.

Amafomethi, ukusebenza, nezikhathi zokuhlaziya

Empeleni, amawebhusayithi amaningi okuthola amavidiyo adalwe yi-AI ayamukela amafomethi adumile njengeMP4, AVI noma MOVkanye nezixhumanisi eziqondile ezisekelweni. Isikhathi sokuphendula ngokuvamile sisuka kumasekhondi siye kumaminithi ambalwa, kuye ngobude bevidiyo nomthwalo wesistimu.

Kwezinye izimo, Ukucutshungulwa cishe kusesikhathini sangempela.Ikakhulukazi lapho ukuhlaziya ubungozi bokuqala kwenziwa. Ukuze uthole imibiko ebanzi enamamephu okushisa kanye nokuhlukaniswa kozimele ngozimele, ukulinda kungase kube kude kancane.

Idatha, ukuthobela kanye nokwenza izinto obala

E-Europe, imithethonqubo iza ngamandla: Umthetho we-AI uzodinga ukulebula kokuqukethwe okukhiqizwayo Imayelana nokuhlinzeka ngokusobala mayelana nomsuka. Lokhu akusizi abasebenzisi kuphela, kodwa futhi kulinganisa izinqubo kumidiya, ukukhangisa, nemfundo.

Uma usebenza enhlanganweni, cabanga ngenqubomgomo yangaphakathi: Ukuqeqeshwa ekuqinisekiseni, ukusebenzisa ngendlela efanele imitshina, kanye nokubonisana nochwephesheIzinkampani ezikhethekile ezifana ne-Atico34 zinikeza ukwesekwa ukuze kuqinisekiswe ukuthi konke lokhu kuyahambisana nokuvikelwa kwedatha kanye nezibopho zomthetho.

Imibuzo evamise ukubuzwa mayelana namawebhusayithi okuthola amavidiyo akhiqizwe yi-AI

• Yikuphi ukunemba kwangempela engingakulindela kumtshina wevidiyo eku-inthanethi? Kuncike esimweni, kodwa ezinye izinsiza zibika amazinga okunemba angaphezu kwama-95% kumafomethi athile kanye nokukhohlisa. Noma kunjalo, khumbula ukuthi ama-deepfakes ayavela, futhi alikho ithuluzi elinembile ngo-100%.
• Imaphi amafomethi wevidiyo avame ukusekelwa? Iningi lisebenza ngamafayela e-MP4, AVI, kanye ne-MOV, kanye nezixhumanisi eziqondile ezivela ezisekelweni ezidumile. Njalo hlola uhlu lokusebenzisana lwesevisi ohlela ukuyisebenzisa.
• Ingabe amavidiyo alungiswe kancane angatholwa? Yebo. Izitholi zamanje zingakwazi ukubona amasegimenti ashintshiwe yi-AI ngaphakathi kwesiqeshana sangempela, ikakhulukazi ngokungahambisani kwasendaweni noma ama-artifact ezindaweni ezithile.
• Kuthatha isikhathi esingakanani ukuhlaziya? Ivamise ukusuka kumasekhondi kuye kumaminithi, ihluka kuye ngobude bevidiyo, ubunkimbinkimbi bayo, kanye nomthwalo wesistimu ngaleso sikhathi.
• Yiziphi izinhlobo zokukhohlisa abazikhombayo? Okuphelele kakhulu kuhlukanisa phakathi kokujula kobuso, ukuhlanganisa izwi, ukudluliswa kwesitayela, kanye nokukhiqizwa kwezinga lesigcawu sokwenziwa, ngokusebenza okuhlukahlukene esigabeni ngasinye.

Ku-ecosystem lapho okokwenziwa kanye nomuntu sebevele bedansa eduze kakhulu, kuwukuhlakanipha ukuqhubeka ngokuqapha: Ihlanganisa ukubuka, amathuluzi, ubuhlakani, kanye namazinga okuqinisekisa acacile. ukuze ugweme ukuwela ogibeni, futhi ukhumbule ukuthi inani alikho ekudaleni i-AI, kodwa ekuyisebenziseni ngokuzibophezela nangokusobala.

Isihloko esihlobene:

I-Pinterest yenza kusebenze izilawuli ukuze kwehliswe okuqukethwe kwe-AI kokuphakelayo

UDaniel Terrasa

Umhleli okhethekile kwezobuchwepheshe kanye nezindaba ze-inthanethi onolwazi olungaphezu kweminyaka eyishumi kumidiya ehlukene yedijithali. Ngisebenze njengomhleli kanye nomdali wokuqukethwe kwe-e-commerce, ukuxhumana, ukumaketha ku-inthanethi kanye nezinkampani zokukhangisa. Ngike ngabhala kumawebhusayithi ezomnotho, ezezimali neminye imikhakha. Umsebenzi wami nawo uwuthando lwami. Manje, ngokusebenzisa izihloko zami ku Tecnobits, ngizama ukuhlola zonke izindaba namathuba amasha izwe lobuchwepheshe elisinikeza lona nsuku zonke ukuze sithuthukise izimpilo zethu.