Ukukhiqizwa kwezithombe okusebenzisa i-AI sekungenye yezindlela ezibonakala kakhulu zomjaho phakathi kwezinkampani ezinkulu zobuchwepheshe. usenqume ukuthatha isinyathelo ngesibuyekezo esijulile se Izithombe ze-ChatGPT, uhlelo lwayo lokudala okubonakalayo oluhlanganisiwe, esimweni lapho amamodeli afana ne-Nano Banana Pro ka-Google ayebusa ingxenye enkulu yengxoxo.
Ngalokhu kwethulwa, inkampani ngemuva kwe-ChatGPT ifuna ithuluzi layo lidlule ekubeni yisengezo sengxoxo esilula futhi lisebenze njengesici esiphelele. isitudiyo sokudala esihlanganisiwe sangempela, esheshayo, enembe kakhudlwana futhi ene-interface eyenzelwe kusukela ekuqaleni ukusebenza nezithombe esikhundleni sokukhawulelwa embhalweni.
Imodeli entsha ye-GPT Image 1.5: isivinini nokunemba njengezimpawu zayo
Inhliziyo yesibuyekezo yile Isithombe se-GPT 1.5Imodeli entsha ye-OpenAI yezithombe. Le nkampani ithi ingakhiqiza okuqukethwe okubonakalayo kuze kufike ku- ngokushesha izikhathi ezine kunenguqulo yangaphambilini, into ebonakala kakhulu emahoreni okusebenza kakhulu nakumadivayisi eselula, lapho ngaphambili kwakungavamile ukuthi inqubo iphazanyiswe noma ithathe isikhathi eside lapho kushintsha izinhlelo zokusebenza.
Ngaphezu kokusebenza, intuthuko eyinhloko isekulandeleleni imiyalelo. Uhlelo luchaza imiyalelo ngokunembe kakhudlwana. iziphakamiso eziyinkimbinkimbi kanye nobudlelwano obuqondile bendawoukuze izicelo ezifana nokushintsha into eyodwa kuphela, ukulungisa ukukhanya, noma ukushintsha izingubo zomuntu zingasabangeli izinguquko ezingalindelekile endaweni esele.
I-OpenAI ichaza ukuthi i-GPT Image 1.5 iqeqeshwe ukugcina izinto ezibalulekile zesithombe zingaguquguquki, njengokuthi ukuhlonza ubuso, ukwakheka okuphelele, noma i-palette yombalangisho nangemva kwemijikelezo eminingana yokuhlela okuboshwe ngentambo. Leli phuzu libaluleke kakhulu ekusetshenzisweni kobungcweti, lapho ukungaguquguquki kokubona kungeyona into eyenzeka ngenkani, kodwa kuyimfuneko.
Ukuhlela amachashazi nochungechunge: shintsha kuphela okubalulekile
Enye yezindawo lapho amamodeli angaphambilini ahluleka khona kwaba ukuhlela okuqondiwe kwezindawo ezithileUkushintsha isigqoko, ukulungisa ukukhanya, noma ukwengeza isici ngemuva kungagcina kuxube yonke indawo. Izithombe ezintsha ze-ChatGPT zibhekana ngqo nale nkinga.
Imodeli iyakwazi engeza, susa, hlanganisa, hlanganisa futhi uguqule izinto ngaphakathi kwesithombe esifanayo ngenkathi kugcinwa zonke ezinye izingxenye ezibalulekile zizinzile. Empeleni, lokhu kusho ukukwazi ukucela izenzo ezinjengokuthi: ukushintsha umbala wehembe, ukulungisa isigqoko, ukulungisa uphawu lomgwaqo, noma ukuguqula iloli libe yiloli lomlilo ngaphandle kokuphazamisa yonke indawo ezungezile.
Ukuziphatha kwezingcingo nakho kuye kwaqiniswa izinhlelo zochungechungeKuze kube manje, ushintsho lwesithathu noma lwesine luvame ukubangela ukuthi imodeli "iphinde isungule" ngokuphelele isithombe. Nge-GPT Image 1.5, ithuluzi ligcina isitayela, ukuma, kanye nesigcawu ngokwethembeka okukhulu, ukuze ukwazi ukuphindaphinda ngesisekelo esifanayo ngaphandle kokuqala kusukela ekuqaleni ngokuguqulwa ngakunye.
Ukuguqulwa kokudala: kusukela ku-selfie kuya ku-poster ye-movie
Ngaphandle kokunemba kwayo kobuchwepheshe, i-OpenAI isunduza Izithombe ze-ChatGPT endaweni yokudala ecacile. Uhlelo luvumela abasebenzisi ukuthi balayishe izithombe zabo, futhi, ngesicelo esilula, bathole isithombe ngemizuzwana embalwa. izinguqulo eziguquliwe ezithembekileKusukela esikhangisweni sama-90s kuya endaweni eseTimes Square phakathi nobusika noma idolobha laseJapan elinobuhle be-cyberpunk.
Imodeli iyakwazi futhi ukudala kabusha izitayela zobuciko ezithile, njengamaphosta ama-movie akudala, imidwebo yesitayela se-anime, noma izingoma ezibukeka njengomlando, zihlonipha izici ezibalulekile zomuntu wokuqala. Umqondo uwukuthi umsebenzisi angazibona "ezimweni ezihlukene kakhulu, ngaphandle kokulahlekelwa umuzwa wokuthi ungumuntu ofanayo.
Le ndlela ikhumbuza lokho amamodeli afana ne-Nano Banana asevele ekunikeza, kodwa i-OpenAI izama ukuzihlukanisa ngokubheja ku- ukuguqulwa komqondo okulawulwa kakhululapho uhlelo lugcina khona umongo wesithombe esiyisisekelo ngenkathi lushintsha izingubo, indawo, ukukhanya noma isikhathi ngokuhambisana okukhulu okubonakalayo.
Izithombe ze-ChatGPT zivalelisa isitayela esiphuzi futhi zithuthukisa izigcawu eziyinkimbinkimbi
Isikhathi eside, kwakulula kakhulu ukubona ukuthi isithombe sakhiwe ngezinguqulo zokuqala ze-ChatGPT: zazivame kakhulu imibala efudumele, ukuqeda okukhilimu, kanye nethoni ephansi ephuzi okwembule imvelaphi yayo yokwenziwa. Ukuqhathaniswa kwangaphakathi okuboniswe yi-OpenAI kanye nokuhlolwa okuzimele, kuqhathaniswa nezinye izindlela ezifana Umdali wesithombe se-BingKubonakala sengathi leso sici sishiywe ngemuva.
Imodeli entsha inikeza ububanzi bombala obungathathi hlangothi futhi obuhlukahlukeneLokhu kwenza izithombe zibukeke njengezithombe ezivamile ngaphandle kokuthi umsebenzisi ucele ngokusobala okuhlukile ku-prompt. Lokhu kusiza izithombe ukuthi zibonakale "zingenalo uphawu" oluningi futhi ziwusizo kakhulu ezimweni lapho kudingeka khona ukuba ngokoqobo noma ukuhlanganiswa nezinto ezikhona zezithombe.
Kuye kwenziwa intuthuko ekumelweni kwe- izigcawu ezinezici eziningi ezincanenjengezixuku noma izizinda ezicebile ngemininingwane. Ubuso obusemaqenjini amakhulu manje sebuhluke kakhulu komunye nomunye, bunezimo zemvelo kanye nokubonakaliswa, futhi amaphutha ajwayelekile njengezandla, imivimbo emincane, noma ukuphindaphinda okungajwayelekile ancishisiwe.
Izithombe ze-ChatGPT zikuvumela ukuthi ufake umbhalo ngaphakathi kwezithombe: gxuma kumaphosta, ama-infographic kanye nama-mockup
Ukukhiqiza umbhalo ofundekayo ngaphakathi kwesithombe ngokomlando kube ngenye yezindlela zika-Achilles zokukhiqiza i-AI. I-OpenAI ithi i-GPT Image 1.5 ithatha isinyathelo esibalulekile kule ndawo, nge ukuhumusha kokuthayipha okuhambisana kakhulu kunezinhlobo zangaphambilini.
Imodeli ingakwazi ukusingatha amabhlogo amancane ombhalo aminyeneLokhu kuvula ithuba lokudala amaphosta, ama-infographic, ama-mockup ephephandabeni, noma imiklamo enamathebula namafomethi ohlobo lwe-markdown anezinga lokufundeka, yize lingaphelele, eliseduze nokuthile okusebenzisekayo ngaphandle kokulungiswa kabusha okujulile.
Kulabo abasebenza kwezokumaketha, ezemfundo, ezokuhweba nge-inthanethi, noma okuqukethwe kwedijithali, lokhu kuthuthukiswa kusho ukunciphisa isikhathi esichithwa ku- izinhlamvu ezingalungile noma amagama angapheleleEzimweni lapho kunesidingo sokukhiqiza izinto ezibonakalayo ezinemiyalezo ecacile elungele ukushicilelwa, iqiniso lokuthi imodeli ngokwayo ikhiqiza umbhalo ohlanzekile ngokwanele liba yinto ehlukanisayo.
Okuhlangenwe nakho okusha komsebenzisi: isigaba sezithombe esizinikele ku-ChatGPT
Ukubuyekezwa akugcini nje kumodeli; kuthinta nendlela esetshenziswa ngayo. I-OpenAI ingeze isici esisha kubha eseceleni ye-ChatGPT. ingxenye ethile ebizwa ngokuthi “Izithombe”Lokhu kusebenza kokubili kuhlelo lokusebenza lweselula kanye nenguqulo yewebhu. Umgomo ukuhlukanisa ulwazi olubonakalayo engxoxweni yendabuko nokwenza kube lula kulabo abangafuni ukubhekana nezinkinga eziyinkimbinkimbi ukuthi bazulazule.
Kusukela kulesi sikhala esisha, umsebenzisi uthola izitayela ezichazwe ngaphambilini, iziphakamiso zezitayela, kanye namathempulethi Ngemisebenzi ejwayelekile njengokudala imikhonzo, ukubuyisela izithombe ezindala, ukushintsha phakathi kwezitayela zobuciko ezahlukene, noma ukudala ukuhlukahluka komkhiqizo ofanayo, le ndlela yehlisa isithiyo sokungena kubantu abangenalo ulwazi lobuchwepheshe.
Esinye isici esiwusizo ukuthi isigaba sezithombe sisebenza njenge indawo yokugcina impahla ephakathi yazo zonke izinto ezidalwe ngumsebenzisi ezibonakalayo. Ukusuka lapho kulula ukubuyekeza izinguqulo zangaphambilini, ukuphinda isitayela esinokuqukethwe okusha, noma ukuqhubeka nokuhlela isithombe esivele senziwe, into ewusizo kakhulu emisebenzini eqhubekayo.
Kusukela ku-accessory ebamba amehlo kuya kuthuluzi lokusebenza elibonakalayo
I-OpenAI ngokwayo iyavuma ukuthi, kuze kube manje, ukwenziwa kwezithombe ngaphakathi kwe-ChatGPT kusebenza njenge- ukubamba iso kakhulu ngaphakathi kwesixhumi esibonakalayo esenzelwe umbhalo okusebenza njengendawo yokusebenza ebonakalayo eqinile. Ngalesi sibuyekezo, inkampani ihlose ukwenza igxathu elifanele: ukusuka ezithombeni "zokuhlola" zemidiya yezenhlalo ukuya kuthuluzi elingasetshenziswa ezinqubweni zangempela.
Ukuthuthuka kokuvumelana nokuphindaphinda kunomthelela oqondile emikhakheni efana ukuklama, ukumaketha, ukuhweba nge-inthanethi noma uphawuIzinkampani ezidinga ukuvumelanisa umqondo ofanayo wokudala nezakhiwo eziningi, ukuhlola ukuhlukahluka komkhiqizo, noma ukugcina ukuvumelana kwama-logo nezinto zenkampani ezingcezu eziningi zithola inzuzo ecacile kulolu hlobo lokulawula.
Amapulatifomu okudala asebenza eYurophu, njengabahleli bewebhu namathuluzi okuklama asekelwe efwiniSebavele bahlanganisa lawa mamodeli emisebenzini yabo yokusebenza. Kulesi sigaba, ukuzibophezela kwe-OpenAI endaweni ebonakalayo ebanzi kakhulu kungaba yinto efanele kokubili ama-SME afuna ukusheshisa ukukhiqizwa kwezinto ezibonisa izithombe kanye namaqembu okuxhumana kwangaphakathi ezinkampanini ezinkulu.
Ukutholakala kwezithombe ze-ChatGPT kubasebenzisi, amabhizinisi, kanye nabathuthukisi
I-OpenAI isiqalile ukukhipha izithombe ezintsha ze-ChatGPT ze- iningi labasebenzisi beplatifomu, okuhlanganisa nalabo abanama-akhawunti wamahhalaAbasebenzisi abaningi sebevele bebona isaziso lapho bevula uhlelo lokusebenza olubamema ukuthi bazame umsebenzi wesithombe, kanye nethebhu entsha enikezelwe kumenyu eseceleni ukuze kusetshenziswe ndawonye.
Emkhakheni webhizinisi, inkampani iqinisekisile ukuthi ukufinyelela okuthuthukisiwe kwama-akhawunti ebhizinisi nawebhizinisi kuzoqala ukusebenza kancane kancane, kugxilwe ekuhlanganisweni ngaphakathi ukuhamba komsebenzi ochwephesheEzinhlanganweni zaseYurophu esezivele zisebenzisa i-ChatGPT emisebenzini yangaphakathi, lokhu kusho ukukwazi ukwandisa ukusetshenziswa kwayo kusukela kumbhalo kuya ezintweni ezibonakalayo ezikhiqizwe ngaphansi kweziqinisekiso ezifanayo.
Ngesikhathi esifanayo, i-GPT Image 1.5 iyatholakala nge- I-OpenAI APILokhu kuvumela abathuthukisi ukuthi bahlanganise amakhono okukhiqiza izithombe nokuhlela ezinhlelweni zabo zokusebenza. Le nkampani ithi izindleko zokufaka nokukhipha izithombe ziphansi cishe ngo-20% kunemodeli yangaphambilini, okuyinzuzo enkulu kumaphrojekthi amakhulu noma izinsizakalo ezisebenza ngemingcele emincane.
Ukuncintisana ne-Nano Banana Pro kanye nezinye izinhlobo ezibonakalayo
Isinyathelo se-OpenAI sifike ngesikhathi sokucindezeleka okukhulu kokuncintisana. I-Google icindezele Nano Banana Pro njengenye yamamodeli akhiqiza izinto ezibonakalayo ahamba phambili, ahlanganiswe ohlelweni lwayo lwamathuluzi okudala kanye kuxhumene naye Umndeni wakwaGemini, okuye kwathuthukisa ukusetshenziswa kwayo emhlabeni jikelele.
Lesi simo siholele ekusungulweni [okungacacile] kwezinye izinsizakalo ezincintisanayo. imikhawulo eqinile yabasebenzisi abakhululekileIsibonelo, ngokunciphisa inani lezithombe ezingakhiqizwa ngosuku, ngokwengxenye ngenxa yokufunwa okuphezulu. Ngokuphambene nalokho, i-OpenAI ibonakala ibheja ekuhlanganisweni kokufinyelela okubanzi, isivinini esikhulu, kanye nendawo yokuhlela ecwengekile ukuze igcine futhi idonsele abasebenzisi.
Okwamanje, abanye abadlali abanjengo-xAI nge-chatbot yayo i-Grok noma ochwepheshe bezithombe abahlukahlukene baphokophela phambili. isizukulwane esibonakalayo siba yingxenye ebalulekile empini yokunaka kwabasebenzisi. Isu le-OpenAI lihilela ukuhlanganisa i-ChatGPT "njengohlelo lokusebenza oluphelele," lapho usesho, izwi, umbhalo, izithombe, namavidiyo kuhlangana ndawonye endaweni eyodwa yokungena.
Ngalezi zithombe ezintsha ze-ChatGPT, i-OpenAI ithatha isinyathelo esibalulekile esiya ku- ithuluzi elibonakalayo elivuthiwe kakhuluImodeli esheshayo nenembe kakhudlwana, isikhombikubona esihlukile, kanye namakhono okuhlela aqondiswe ngokusobala emsebenzini wangempela, kokubili ezimweni zomuntu siqu nezobungcweti. Kusazobonakala ukuthi lezi zithuthukisi zizohlanganiswa ngezinga elingakanani empilweni yansuku zonke yabasebenzisi namabhizinisi eSpain naseYurophu, kodwa umyalezo ucacile: isithombe asiseyona nje into ejabulisayo engxoxweni, kodwa sesibe yingxenye ebalulekile yesistimu ye-ChatGPT.
