Impi ye ubuhlakani bokwenziwa busetshenziswa ezithombeni Akuseyona nje isithembiso sesikhathi esizayo: sekufikile, futhi kuyinto engathi sína kakhulu. Ekhoneni elilodwa lendandatho sinayo... Isithombe se-ChatGPT 1.5, imodeli entsha ye-OpenAI enika amandla Izithombe ze-ChatGPT; ngakolunye uhlangothi, Nano Banana Pro, the Umkhiqizi wezithombe odumile kanye nomhleli we-Google uhlanganiswe kuyo yonke indawo yayo.Umncintiswano ukhulu kakhulu kangangokuthi i-OpenAI ize yakhuluma ngaphakathi ngesimo "sekhodi ebomvu" ngemuva kokuthuthuka kwe-Google.
Uma usebenza ngokuqukethwe okubonakalayo — kungakhathaliseki ukuthi uchwepheshe wokudala, umakethi, unjiniyela noma umuntu othanda ukuzilibazisa nje— Cishe uyazibuza ukuthi iyiphi kulezi zinhlobo ezimbili ezokunikeza imiphumela engcono. Ake sichaze, ngokuthula kodwa ngokuqondile, ukuthi i-ChatGPT Image 1.5 kanye ne-Nano Banana Pro ziqhathaniswa kanjani ezindaweni ezibaluleke ngempela: isitayela, i-photorealism, amafomethi, isixazululo, isivinini, intengo, ukuhlela, izinkomba, kanye nolwazi lomsebenzisi.
Umongo: Kungani Isithombe se-ChatGPT 1.5 sibaluleke kangaka
Ukufika Isithombe se-ChatGPT 1.5 Lokhu akusikho ukubuyekezwa okuncane. I-OpenAI ikwethula njengemodeli yayo entsha ebonakalayo, etholakala kokubili kuthebhu yezithombe ze-ChatGPT kanye nange-API. Umyalezo osemthethweni ucacile: izithombe ezilandela imiyalelo eduze, ukuhlela okuhle, kanye nokukhiqiza kuze kube kane ngokushesha kunenguqulo yangaphambilini.
Lokhu kwethulwa nakho kuza ngesikhathi esibucayi kakhulu. Kulandela ukusunduza kwe-Google nge IGemini 3 Futhi ngomkhiqizi wayo obonakalayo ohamba phambili—obizwa ngokuthi i-Nano Banana—i-OpenAI yavula uhlobo lohlelo lokuphahlazeka. Ngokuhambisana nezinyathelo ezinkulu njengokwethulwa kwe I-GPT-5.2Le nkampani kwakudingeka ibonise ukuthi ingahola nasemkhakheni wokubona, hhayi nje kumamodeli olimi.
Ngakho-ke, i-GPT Image 1.5 ayigcini ekudaleni izithombe kusukela ekuqaleni: iba yisisekelo se Izithombe ze-ChatGPTithebhu ezinikele ngaphakathi kohlelo lokusebenza kanye newebhusayithi esebenza njengeqiniso "isitudiyo sokudala" esihlanganiswe ku-ChatGPTUkusuka lapho ungakha, uhlele, usebenzise izitayela, uzame iziphakamiso eziphakanyisiwe futhi uthole ngisho nemibono emisha ngenxa yokuphakelayo okwenzelwe ukuphefumulela.
Ukunamathela kwisitayela: ubani ohlonipha kakhulu isitayela osicelayo
Enye yezivivinyo eziveza kakhulu ukuhlola amamodeli ezithombe ukuhlola ikhono lawo lokulandela isitayela esithile sezithombeEnye yeziqhathaniso ezicacile phakathi kwalezi zinhlobo ezimbili yasebenzisa isisho esinemininingwane eminingi: indawo yaseLos Angeles ene-skateboarding, echazwa njengezithombe zasemgwaqweni eziqophayo kusukela ngasekupheleni kweminyaka yama-90, kufilimu engu-35mm, enezithombe eziqoshiwe. Iphalethi le-Kodak Portra 400Ikhamera yohlobo lwe-Leica M, ukukhanya kwemvelo, umehluko othambile, imibala engokoqobo ethulile, uhlamvu lwefilimu olubonakalayo, ubumnene obuncane emaphethelweni futhi, okubaluleke kakhulu, akukho HDR noma ukucija kwedijithali kwesimanje.
Kulolo vivinyo, Izithombe ze-ChatGPT 1.5 kuphela Wagxila engxenyeni ebalulekile yesitayela: okusanhlamvu kwefilimu okufakiwe kanye nokubukeka okubonakalayo okungokwemvelo umuntu angakulindela ku-Kodak Portra 400 film roll. Isithombe esiphume sahlonipha imiphetho ethambile, ukungabikho kobukhali obukhulu bedijithali kanye nomuzwa wedokhumentari ojwayelekile ngasekupheleni kweminyaka yama-90.
I-Nano Banana Pro, yona, idale isithombe esikhangayo, kodwa esinobuhle obuncane ngokusobala. idijithali futhi ibukhaliIsithombe besihlanzekile kakhulu futhi sibukhali, singenawo umugqa we-analog kanye nokuthungwa okungaphelele kancane njengoba isisho besidinga. Ngamanye amazwi, ezingeni elijulile lesitayela, imodeli ye-Google yaphambuka emgomweni eyayihlosiwe: yayiyisithombe esihle sesimanje kunefilimu engu-35mm ethembekile.
Lolu hlobo lwecala lusikisela ukuthi i-ChatGPT Images 1.5 ine- ukunamathela kangcono emiyalweni yesitayela esiyinkimbinkimbiLokhu kubaluleke kakhulu kubathwebuli bezithombe, abaqondisi bezobuciko, noma abadali abafuna ukudala kabusha izinkathi ezithile, ama-emulsion, noma imisebenzi ye-analog.
Ukwenziwa kwezithombe ngendlela ye-Photorealism ekuthwebuleni izithombe zeselula: ukuhlolwa kwe-iPhone
Esinye isimo esibalulekile yi-photorealism yesimanje, lapho ikhono lazo zombili izinhlobo zokulingisa isitayela se-a Isithombe sithathwe nge-smartphone yesimanjeIsaziso esasetshenziswa kwakuyi: isithombe esingokoqobo sowesifazane osemusha ethala elithule e-Amsterdam, ekuseni ngoMashi obandayo, enobubanzi obungajulile futhi ethathwe nge-iPhone.
Kulokhu kuqhathanisa, umphumela ubonakala sengathi uthambekele ohlangothini lwe- Nano Banana ProIsithombe esikhiqizwe yimodeli ye-Google sasibukeka njengesithombe esithathwe ngocingo: ubukhali nokufiphala okuhambisana nekhamera ye-smartphone, ukwakheka okukholekayo, kanye nesimo esiveza kahle lowo muzwa obandayo, wasekupheleni kobusika oceliwe kulesi saziso.
Ngaphezu kwalokho, uNano Banana wengeze imininingwane, nakuba ingaphelele, ekhuluma kahle ngokuqonda kwakhe okuphathelene nesimo: wafaka phakathi iphosta yekhefi yangempela yase-AmsterdamIndawo eqondile yayingalungile, kodwa yayiveza umuzwa wokuba sendaweni yangempela edolobheni, okuqinisa ubuqiniso bendawo.
Ngakolunye uhlangothi, izithombe ze-ChatGPT 1.5 zikhiqize isithombe esine- isici se-"AI" esengeziweNgamaphutha amancane kodwa abonakalayo esweni eliqeqeshwe kahle: ukuthungwa okuthile okubonakala sengathi kuyipulasitiki kancane, ukufiphala okungokwemvelo, kanye nesimo esijwayelekile esingahambisani kahle nesithombe sangempela se-iPhone. Kalula nje: i-Nano Banana ithole ubuhle be-smartphone obukholisayo kakhulu.
Isiphetho esiyingxenye: uma okubaluleke kakhulu kuwe ukukhiqiza izithombe zeselula ezithembekile (kumanethiwekhi, ama-prototype omkhankaso, ama-mockup emikhiqizo ngokomongo, njll.), i-Nano Banana okwamanje izwakala iphambili ekubonakaleni kwangempela okusheshayo.
Izilinganiso zokubukeka: ukuguquguquka kwefomethi kusebenza kahle ku-Google
Ngale kwesitayela, ekukhiqizeni kobungcweti, okulandelayo kubaluleke kakhulu: ukuguquguquka kwefomethiLapha, i-Nano Banana Pro inezinsizakalo zayo zokusekela zendabuko zesamba se- Izilinganiso zesici ezingu-8 kwehlukile:
- 1:1 (isikwele)
- 3:2 (indawo yakudala)
- 2:3 (okuqondile okujwayelekile)
- 4:3
- 3:4
- 16:9 (i-panoramic ejwayelekile)
- 9:16 (okuqondile kwezindaba nezimfushane)
- 21:9 (i-panoramic ephezulu)
Ngakolunye uhlangothi, izithombe ze-ChatGPT 1.5 zinikeza ukuhambisana kuphela ne- Izilinganiso zesici ezingu-3Nakuba ihlanganisa izinto eziyisisekelo, ayiphumeleli uma iqhathaniswa nohlobo lwe-Nano Banana, ikakhulukazi uma udinga okuqukethwe okulungiselelwe amapulatifomu amaningi edijithali anezidingo ezithile (amabhanela, ama-reels, izikhangiso zokubonisa, amaphephadonga, njll.).
Ekusetshenzisweni kwansuku zonke, lo mkhawulo usho ukuthi nge-ChatGPT Images 1.5 kuzodingeka nquma futhi uhlanganise kaningiKanti nge-Nano Banana ungakhiqiza ngqo ngefomethi yokugcina oyifunayo, wonge isikhathi futhi ugweme ukulahlekelwa okukhulu kokufometha.
Isixazululo esiphezulu: ubani ohamba ibanga elide kakhulu ngekhwalithi yokukhipha
Kepha isinqumo, Nano Banana Pro uthatha ubuholi futhi amathuba ezobuchwephesheIzithombe ze-ChatGPT 1.5 zivumela:
- Isinqumo mayelana 1K ngefomethi yesikwele.
- Isixazululo esilinganiselwe se 1,5K ngamafomethi e-landscape noma e-portrait.
I-Nano Banana Pro, ngakolunye uhlangothi, uyavuma:
- 1K isisekelo.
- 2K njengesisombululo sokusebenza esisezingeni eliphezulu kakhulu.
- 4K ngezinqubo ze ukukhushulwa okuhlanganisiwe, eyenzelwe ukusetshenziswa lapho kudingeka khona incazelo ephezulu (ukuphrinta, izikrini ezinkulu, ukunqampuna okunamandla, njll.).
Lokhu kusho ukuthi, ngezindleko ezifanayo ezimweni eziningi, i-Nano Banana ihlinzeka Amaphikseli engeziwe kanye nendawo eyengeziwe yokuhlela ngemuva.Ngomsebenzi odinga ukwandiswa noma ukuhlanza okukhulu lapho usondeza, umehluko phakathi kokuhlala ku-1K/1,5K noma ukukwazi ukufika ku-2K/4K awulula.
Isivinini sokukhiqiza: Ithiyori ye-OpenAI vs. izinombolo ezingavuthiwe
I-OpenAI ithi i-ChatGPT Images 1.5 iyi- kuze kube kane ngokushesha kunenguqulo yayo yangaphambilini. Empeleni, uma kuqhathaniswa ne-Nano Banana, ukuqhathanisa kuncike esixazululweni, kodwa idatha yethebula icacile.
Kokuhamba kusukela 1K:
- Izithombe ze-ChatGPT 1.5: phakathi kwemizuzwana engama-30 nengama-45 ngesithombe ngasinye.
- Nano Banana Pro: phakathi kwemizuzwana eyi-10 neyi-15.
Ngamanye amazwi, i-Nano Banana ikhiqiza cishe i-1K ngokushesha kathathu Izithombe ze-ChatGPT 1.5. Lapho i-GPT ithatha isigamu somzuzu noma ngaphezulu, imodeli ye-Google iletha imiphumela ngemizuzwana engaphezu kweyi-10, into ephawuleka kakhulu uma uphinda izinhlobo eziningi.
Uma siya ku- izinqumo eziphakeme:
- Izithombe ze-ChatGPT 1.5: kufika ku-1,5K, ngezikhathi zemizuzwana engama-50 kuya kwengama-60.
- Nano Banana ProI-2K ngemizuzwana engama-30 kuya kwengama-60, kanye ne-4K ngemizuzu embalwa ngokukala.
Isihluthulelo lapha yilokho I-Nano Banana ifinyelela ku-2K cishe ngesikhathi esifanayo esithatha i-GPT ukukhiqiza isithombe esingu-1,5K. Kuma-studio, ama-ejensi, noma amaqembu omkhiqizo adinga izithombe eziningi, leyo nhlanganisela ye Isixazululo esiphezulu kanye nokulinda okuncane Kuyimpikiswano enamandla kakhulu.
Intengo ye-API: ubani onikeza okuningi ngemali efanayo
Ngokuphathelene nezindleko ngokusetshenziswa ngakunye nge-API, indaba iyaziphinda: Izipesheli ze-Nano Banana ukulungiswa okuphezulu ngentengo efanayo ekhombisayoNgokusho kokuqhathaniswa okutholakalayo:
- para 1KI-ChatGPT Images 1.5 kanye ne-Nano Banana Pro zisebenza ngendlela efanayo cishe $0,15-$0,17 ngesithombe ngasinye.
- En 2KIzithombe ze-ChatGPT 1.5 azinazo izinketho, kuyilapho i-Nano Banana igcina lolo hlu lwe- $0,15-0,17.
- para 4K (ngokusebenzisa i-upscaling), i-Nano Banana itholakala eduze $0,26-0,28 ngesithombe ngasinye.
Ihunyushwe yaba amagama asebenzayo, i-Nano Banana API ikunikeza 2K ngentengo enikezwa yi-ChatGPT Images 1.5 nge-1KLokhu kwenza ikhambi le-Google libe inketho engcono kakhulu kulabo abakhiqiza izithombe eziningi noma abadinga izinto zokuphrinta noma izibonisi ezinabantu abaningi.
Amakhono okuhlela: ukunemba kanye nokuhlonipha okwangempela
Enye yezinguquko ezinkulu zalesi sizukulwane esisha esibonakalayo esivela ku-OpenAI yi- Ukuhlela okunembile kwezinto ezithileNgokomlando, amamodeli ezithombe ayenezinkinga ezinkulu zokushintsha ingxenye yesigcawu kuphela: ubungacela ukulungiswa okuthile bese uhlelo lugcina luchaza kabusha ingxenye yesithombe.
Ngezithombe ze-ChatGPT 1.5Izithombe ze-ChatGPT zithembisa okuphambene nalokho: ungabuza, isibonelo, "shintsha kuphela isimo sobuso" noma "shintsha ukukhanya ngaphandle kokuthinta uhlaka" bese imodeli ikwazi ukwenza lokho. ukugcina izici nokwakheka kupheleleIzici zobuso, ukukhanya, amathoni, uhlaka... konke kugcinwa ngaphandle kokuthi ubonise okuthile okuhlukile kusaziso.
I-OpenAI igcizelela ukuthi uhlelo manje seluqonda kangcono kakhulu lokho ofuna ukukushintsha nalokho ofuna ukukushintsha. Akufanele kushintshweEmpeleni, lokhu kuvumela ukuhamba komsebenzi like:
- Ukulungisa indlela umlingiswa abonakala ngayo ngaphandle kokuklama kabusha ubuso bakhe.
- Shintsha umbala wezingubo, izimpawu, noma izimoto kuphela.
- Shintsha ukukhanya okuphelele ngenkathi ugcina uhlaka nokwakheka.
- Ukuguqula izigcawu zansuku zonke zibe amaphosta e-movie noma imifanekiso yesitayela esithile, ihlonipha isihloko esiyinhloko.
Ngakolunye uhlangothi, i-Nano Banana Pro nayo iyasekela Ukuhlela okuvamile futhi kuphendula ngokuthembekile enjongweni yomsebenzisiEzivivinyweni lapho isithombe semoto yasebusuku kwakufanele siguqulwe sibe yindawo yasemini, womabili amamodeli enze kahle, yize abanye abahlaziyi babebheka umphumela we-Google njengongcono kancane ngokombono wobuhle.
Etafuleni lokuqhathanisa, kucatshangelwa kokubili i-GPT Image 1.5 kanye ne-Nano Banana. Banikeza ukuhlela okuqinile kanye nokunamathela okuhle enhlosweniNgakho-ke, lesi sigaba sibonakala silingana, i-GPT ithola inzuzo ekulawuleni okuhle ngaphakathi kwesixhumi esibonakalayo esisha se-ChatGPT Images.
Ukusetshenziswa kwezithombe ezibhekisela kuzo: ukwethembeka vs ubuningi
Enye insimu ebalulekile yomsebenzi wobungcweti ukuphathwa kwe- izithombe eziyisethenjwaImodeli ngayinye inikeza izinzuzo ezahlukene.
Izithombe ze-ChatGPT 1.5 zivumela ukusebenza kuze kufike ku izithombe ezinhlanu zokubhekisela futhi, ngaphezu kwakho konke, inikeza ukulawula okucacile ezingeni ukwethembeka ofuna imodeli iwalandele ngayo. Okusho ukuthi, ungalungisa ukuthi ufuna ukukopishwa okuthembekile kakhulu kwesithenjwa ("ukwethembeka okuphezulu") noma incazelo ekhululekile ethatha kuphela ubuhle obujwayelekile.
Ngakolunye uhlangothi, i-Nano Banana Pro ivelele ngobuningi bayo: isekela kuze kufike ku- Izithombe eziyi-14 ezibhekisela kuzo ngesicelo esisodwa. Lokhu kuvula umnyango wokuhlanganisa izitayela, ukuma, izindawo, kanye nezinto ezivela emithonjeni eminingi ngaphandle kwesidingo sokuhlanganisa iziphindaphindo eziningi kangaka.
Incazelo iyonke yalo mehluko iwukuthi, ekubhekiseni okubonakalayo, ibhalansi yamandlaI-GPT ihamba phambili ekulawuleni ukwethembeka, kuyilapho i-Nano Banana ihamba phambili enanini lezinkomba ezisetshenziswa ngasikhathi sinye. Kuye ngokuthi ubeka phambili ukufana kokulungisa noma ukuhlanganisa okufakwayo okuningi, okukodwa noma okunye kungase kufaneleke kakhulu.
Isixhumi esibonakalayo esisha se-ChatGPT Images: isitudiyo sokudala ngaphakathi kwengxoxo
Ngale kwamakhono alo modeli, i-OpenAI icabange kabusha indlela yokusebenzisana nezithombe ngaphakathi kwe-ChatGPT. Manje sekukhona ithebhu ethile ebizwa ngokuthi Izithombe, kufinyeleleka kokubili kunguqulo yewebhu kanye nohlelo lokusebenza lweselula, oluguqula ulwazi lube yinto eseduze nesitudiyo sokudala kunebhokisi lombhalo elilula.
Kusukela kuleyo thebhu ungakwazi:
- Qala ukusebenza ngokusekelwe esithombeni somuntu siqukungakhathaliseki ukuthi isithombe, isithombe somkhiqizo, noma isigcawu sansuku zonke.
- Hlola izihlungi nezitayela ezilungiselelwe kusengaphambili ukukukhuthaza ngaphandle kokubhala izincomo eziyinkimbinkimbi.
- Sebenzisa iziphakamiso ezivela ezikhuthazweni esezivele zinikezwe uhlelo, ezisebenza njengezifanekiso zokudala.
- Dala kusukela ekuqaleni, hlela lokho osekuvele kwenziwe, noma ucabange kabusha ngesizindalwazi esifanayo ezimweni ezihluke ngokuphelele.
Abaphathi bemikhiqizo ye-OpenAI bachaza lesi sigaba njenge- "Isitudiyo sokudala" sithunyelwe ngaphakathi kwe-ChatGPTYakhelwe ukuze ungadingi ukuba ngumklami noma ube yingcweti yemiyalo ethuthukisiwe. Imayelana nokukhombisa lokho ofuna ukukubona, ukudlala ngokushintshashintsha okusheshayo (ukusebenzisa isivinini esithuthukisiwe), nokuhlola imiphumela sengathi uhlola amasethi ahlukene ekuthathweni kwezithombe.
Ukuguqulwa kokudala: kusukela kuma-selfies ajwayelekile kuya ezigcawini zama-movie
Enye yezindawo lapho Izithombe ze-ChatGPT ezine-GPT Image 1.5 zifuna ukuncintisana khona ngqo ne-Nano Banana ise- ukuguqulwa kokudala okusekelwe ezithombeni zangempelaLo mbono wawusuvele ungomunye wamandla emodeli ye-Google, evumela izinto ezifana nokuguqula izithombe ezivamile zibe izikhangiso, izigcawu ze-cinema, noma imifanekiso enesitayela.
Ngemodeli entsha ye-OpenAI, okudingeka ukwenze nje layisha isithombe —isibonelo, isithombe somzimba wonke— futhi sihambisane nesicelo esilula sokufeza izinguquko ezijulile kodwa ezihambisanayo. Ezinye izibonelo zomhlaba wangempela eziye zahlolwa zifaka:
- Thatha isithombe sendoda evamile bese ubuza: "Beka eTimes Square ebusika buka-2025, izingubo nendawo ezungezile ingakholeka ngaleso sikhathi sonyaka.".
- Yihambise ku- Idolobha laseJapan ngobusuku obunemvula obunobuhle be-cyberpunkngezibani ze-neon, ukubonakaliswa kwe-asphalt kanye nesimo sesikhathi esizayo.
- iguqule ibe yi-a Ama-samurai aseJapan anezikhali zendabuko nama-katana, esimweni somlando esingokoqobo esingenazo izimpawu zanamuhla.
- Khiqiza i-a Isikhangiso sesitayela saseYurophu kusukela ngawo-90 esebenzisa isithombe sakhe somzimba wonke njengomlingiswa oyinhloko.
Kuzo zonke lezi zimo, i-ChatGPT Images 1.5 izama ukugcina ubunikazi bomuntu obubonakalayo Ngenkathi umongo, izingubo, kanye nethoni yokulandisa yesehlakalo kushintsha kakhulu. Kwenzeka okufanayo nakwezinye izibonelo eziboniswe yi-OpenAI, lapho abantu, izinja, kanye nezindawo kuhlanganiswa khona ezithombeni zebhayisikobho ngokwethembeka ezikhuthazweni ezinemininingwane eminingi.
Ithola ukunemba nombhalo ngaphakathi kwezithombe
Okunye ukuthuthuka okubalulekile kumodeli entsha ye-OpenAI kuse- ukunemba kwemiyaleloAbantu abaningi baye babhekana nokukhungatheka kokucela imininingwane ethile (“yenza ihembe libe bomvu, isigqoko sibe phuzi, shintsha inombolo esesibonini somgwaqo ibe ngu-15…”) bese bethola imiphumela engasinakanga ingxenye ye-oda, noma eyisebenzise ngendlela engafanele.
Nge-ChatGPT Images 1.5, i-OpenAI ithi lezi zinhlobo zamaphutha zincishisiwe kakhulu. Uma unikeza imiyalelo enemininingwane, ithuluzi kufanele likwazi kangcono... hlonipha isici ngasinye esithile: imibala yezingubo, izinombolo zezimpawu, inhlanganisela yezinto eziningana ohlakeni olulodwa olunesimo esithile, njll.
Ngaphezu kwalokho, ukusebenza kuqiniswe kwenye yezindawo ezaziyinkinga kakhulu ngokomlando kumamodeli ezithombe: ukukhiqizwa kombhalo ofakwe esithombeniLokhu kubalulekile ekuklameni amaphosta, izikhangiso, izinto zokukhangisa, noma izingoma lapho umyalezo obhaliwe ubaluleke njengesithombe noma umfanekiso.
Kukhona futhi ushintsho olukhulu lobuhle: lokho kubukeka “ophuzi futhi okhilimu” Isibonakaliso sokuthi izithombe eziningi ezikhiqizwe yizinguqulo zangaphambilini ze-ChatGPT zikhiqizwe sinciphile kakhulu. Ngaphandle kokuthi ucele ngokusobala lolo hlobo lwethoni efudumele, imiphumela emisha ivame ukuba mnene futhi ihlukahlukene, okwenza kube nzima ukubona ngokushesha ukuthi zivela kumodeli we-OpenAI.
Ukutholakala, ubumfihlo, kanye nokulinganisela
Isithombe se-GPT 1.5 siyasetshenziswa kubo bonke abasebenzisi be-ChatGPT, kufaka phakathi uhlelo lwamahhalaKokubili kuwebhu nakuhlelo lokusebenza lweselula. Uma usebenzisa uhlelo lokusebenza, cishe uzobona isaziso esikumema ukuthi uqale ukudala izithombe ngokushesha nje lapho uluvula, kanye nesigaba esisha sezithombe kubha eseceleni.
Lesi sigaba asigcini nje ngokuhlanganisa zonke izithombe ozikhiqizayo, kodwa futhi sihlanganisa iziphakamiso zesitayela nezibonelo Ukuze uqale ngaphandle kokuklama isikhuthazo kusukela ekuqaleni. Ungakhetha isitayela esibonakalayo osithandayo, ulayishe isithombe noma uthathe i-selfie, bese uvumela uhlelo ukuthi lwenze okusele, ulungise noma yini ongayithandi kamuva.
Ngokuphathelene nokucutshungulwa kwedatha, i-OpenAI ikhombisa ukuthi ukuhlela kwenziwa kuphela nge izithombe ozilayishayo kanye nemibhalo oyinikezayoAyisebenzisi imithombo yangaphandle ngaphandle kwemvume yakho kuleyo misebenzi ethile. Inkampani igcina ungqimba lwayo lokuphepha lusebenza. ukulinganisela okuqukethwe kanye nemikhawulo yobuchwephesheNoma kunjalo, ayicacisi ngokunembile amasethi edatha asetshenziswa ukuqeqesha imodeli ebonakalayo.
I-Nano Banana ku-Google ecosystem: ukuhlanganiswa kanye nokusetshenziswa kwangempela
Okwamanje, Nano Banana —isidlaliso esidumile se-Gemini 2.5 Flash Image— sizuze udumo lwaso njengomkhiqizi nomhleli Fast and kulula ukuyisebenzisaI-Google ibilokhu ihlanganisa kancane kancane imikhiqizo efana ne-Gemini, i-Google Photos, i-Search uqobo, kanye ne-Messages, okwenza ibonakale ngokwemvelo kokuhlangenwe nakho kwansuku zonke komsebenzisi.
Umphumela uba ukuthi i-Nano Banana ayisebenzi nje kuphela njengethuluzi elizimele, kodwa futhi njengethuluzi elizimele. ingxenye ehlanganisa konke ngaphakathi kwe-Google ecosystemUkuthinta kabusha izithombe zeselula, ukukhiqiza izinsiza zamadokhumenti e-Workspace, noma ukudala izithombe zokuqukethwe kwewebhu kufinyeleleka kakhulu ezigidini zabasebenzisi abangase bangafinyeleli ngqo ekhasini "lelebhu" le-AI.
Ezilinganisweni zomphakathi, njenge- Isilinganiso sombhalo kuya esithombeni se-LMarenaI-Nano Banana Pro ihambe phambili kafushane kwaze kwaba yilapho i-GPT Image 1.5 ithola indawo ephezulu ngemizuzu embalwa kusukela ikhishwe. Kodwa-ke, ezibonelweni ezisebenzayo ezabiwe ngabasebenzisi kanye nemithombo yezindaba, imiphumela yangempela iyinkimbinkimbi kakhulu: kunezimo lapho i-Nano Banana isabonakala ingcono khona, ikakhulukazi ekubonakaleni kwezithombe ngokushesha, kanye nezinye lapho i-GPT ibonisa ukulawula okukhulu noma ukuceba kwesitayela.
Uma sibheka lesi simo, okuvelayo umncintiswano osondelene kakhulu: I-GPT Image 1.5 iphumelela kakhulu ekulawuleni okuhle, ukunamathela kwesitayela okuyinkimbinkimbi, kanye nesitudiyo sokudala esihlanganisiwengenkathi I-Nano Banana Pro ikhanya ngesivinini, ukuxazululwa, ukuhlukahluka kwefomethi, kanye nokuba ngokoqobo okufana nokwama-smartphone.Ukukhetha okukodwa noma okunye kuzoncika ekutheni ukwazisa kakhulu ukuguquguquka kobuchwepheshe kanye nokuhlanganiswa ne-Google, noma uhlelo lwe-ChatGPT, izinketho ezintsha zokuhlela ezithuthukisiwe, kanye nezinga lokulawula elinikezwa yisiphakamiso se-OpenAI.
