I-GLIGEN Grounded Generation
I-GLIGEN (Isizukulwane Solimi Olugxilisiwe-siya-Sesithombe) ikuvumela ukuthi ulawule ngqo lapho izinto zivela khona esithombeni esikhiqiziwe ngokuphakela amabhokisi okuhlanganisa amamodeli namalebula ahambisana nokwaziswa kombhalo.
Uhlolojikelele
It turns vague text-to-image into precise, layout-controllable synthesis.
I-Deep Dive
Amamodeli ajwayelekile ombhalo uye esithombeni alwa nokulawula indawo: cela 'ikati elingakwesokunxele lenja' futhi ngokuvamile uthola ukubekwa okungalungile. I-GLIGEN, eyethulwe ngo-2023, ixazulula lokhu ngokwengeza okokufaka okuyisisekelo okufana namabhokisi ahlanganisayo abhangqwe nombhalo noma amabhizinisi ezithombe, amaphuzu angukhiye, noma izithombe eziyisethenjwa. Okubaluleke kakhulu, imisa izisindo zemodeli yokusabalalisa eqeqeshwe kusengaphambili futhi ijove izendlalelo zokuzinaka ezintsha eziqeqeshekayo ezimunca amathokheni aphansi. Lokhu kusho ukuthi yakhela phezu kwemodeli efana ne-Stable Diffusion ngaphandle kokucekela phansi ulwazi lwayo olufundiwe, futhi isango liqala eduze kukaziro ukuze ukuziphatha kwemodeli eyisisekelo kulondolozwe kusenesikhathi ekuqeqeshweni. Umphumela uba ukukhiqizwa komhlaba ovulekile: ungabeka izinto ezichazwe ngokungafanele ezindaweni ezishiwo, futhi ihlanganisa imiqondo nezakhiwo ezingabonakali ngesikhathi sokuqeqeshwa kokuphansi.
I-Technical Insight
I-GLIGEN imele inhlangano ngayinye eyisisekelo njengophawu oluhlanganisa umbhalo wayo noma ukushumeka kwesithombe nolwazi lwaso lwendawo, njengezixhumanisi ezine zebhokisi elihlanganisayo elifakwe ikhodi ngezici ze-Fourier. Lawa mathokheni aphansi angena ku-U-Net efriziwe yokusabalalisa ngokusebenzisa izendlalelo zokuzinaka ezifakwe esangweni ezibekwe phakathi kwamabhulokhi okuzinaka akhona kanye namabhulokhi okunaka okuphambanayo. Isango elifundekayo, eliqaliswe kuqanda, lilawula ukuthi isisekelo sinethonya elingakanani esizukulwaneni, ngakho ukwengeza ukulawula kwehlisa isithunzi futhi ukuqeqeshwa kuhlala kuzinzile.
I-Strategic Impact
Isivinini nesikali
I-Visual AI ingakwazi ukuhlola, ukutholwa, nokumaka imisebenzi esikalini.
Yakha ukukhetha
Amathimba aqanjiwe angakwazi ukulinganisa imiqondo ngokushesha ngezibuyekezo ezimbalwa ezenziwa mathupha.
Ithimba kanye nokusebenza komsebenzi
Imisebenzi ingasebenzisa amasiginali wesithombe nawevidiyo obekunzima ukuwenza ngaphambilini.
Ikusasa le-GLIGEN Grounded Generation
Isizukulwane esisekelwe phansi nesilawuleka ngokuhleleka sesiba yizinga kumathuluzi okukhiqiza. Lindela isimo sendawo sesitayela se-GLIGEN ukuze uhlangane nezinye izindlela zokulawula ezifana ne-ControlNet nokwaziswa kwesifunda, futhi udlulele kuvidiyo ne-3D lapho ukubekwa kwento ngokuhamba kwesikhathi nesikhala kubaluleke nakakhulu. Njengoba amamodeli amukela ukuxhumana okulandela imiyalelo, isilawuli sesakhiwo sokudonsa nokuwisa kanye namagrafu esigcawu esicaciswe ngolimi kuzokwenza ukwakheka okunembayo kufinyeleleke ngaphandle kwamaqhinga onjiniyela osheshayo.
Ukuqaliswa Komhlaba Wangempela
Ukubeka ilogo noma umkhiqizo endaweni eqondile yesikhangiso esikhiqiziwe kusetshenziswa ibhokisi lokuhlanganisa
Ukuqamba izigcawu eziyinkimbinkimbi ngokucacisa ukuthi uhlamvu ngalunye noma into kufanele ihlale kuphi ngaphambi kokunikela
Ukukhiqiza idatha yokuqeqeshwa ukuze kutholwe into enezindawo ezaziwayo zebhokisi leqiniso eliphansi
Ukupenda into echaziwe endaweni edwetshwe umsebenzisi yesithombe esikhona
Izingozi & Guardrails
Amalungelo ezithombe kanye nemvume kungaba ubungozi bezomthetho uma ukuvela kungacacile.
Ukusebenza kwemodeli kungahluka kukho konke ukukhanya, izibalo zabantu, kanye nezindawo.
Okuhle okungelona iqiniso kungase kungabonakali ngaphandle uma izinga lokuzethemba liqashelwa.
Ukuqalisa Umhlahlandlela
Chaza indlela yokwamukela yokunemba, ukukhumbula, nezindleko zamaphutha.
Hlola ngedatha efana nezimo zangempela zokukhiqiza.
Engeza isibuyekezo somuntu ukuze uthole ukuzethemba okuphansi noma izibikezelo zomthelela omkhulu.
Landelela ukukhukhuleka kwemodeli bese uqinisekisa kabusha ngemva kwezinguquko zekhamera noma zesethi yedatha.
Qhubeka Uhlole
Free newsletter
Get the daily AI briefing
Three verified AI stories every weekday morning, written in plain English. Free forever, no ads.
One email each weekday. Unsubscribe in one click. We never sell or share your address.
Test yourself
Take the GLIGEN Grounded Generation quiz
Instant feedback on every answer, and a shareable certificate with a verifiable ID once you pass a course.
Support free AI education. AI Understanding is a 501(c)(3) nonprofit — no ads, no paywall, ever. Make a donation
Umhlahlandlela olandelayo
I-Text-to-3D Generation
Imibuzo evame ukubuzwa
What is GLIGEN Grounded Generation?
I-GLIGEN (Isizukulwane Solimi Olugxilisiwe-siya-Sesithombe) ikuvumela ukuthi ulawule ngqo lapho izinto zivela khona esithombeni esikhiqiziwe ngokuphakela amabhokisi okuhlanganisa amamodeli namalebula ahambisana nokwaziswa kombhalo. Iguqula umbhalo uye-isithombe ongacacile ube ukwakheka okunembayo, okulawulekayo kwesakhiwo.
Iyiphi inkinga i-GLIGEN eyixazulula ngokuyinhloko?
I-GLIGEN yengeza okokufaka okusekelwe phansi okufana namabhokisi ahlanganisayo ukuze ukwazi ukulawula lapho izinto zibekwe khona, iziphi iziyalezo zombhalo ongenalutho oziphatha kabi.
I-GLIGEN ilugcina kanjani ulwazi lwemodeli yokusabalalisa eqeqeshwe kusengaphambili?
I-GLIGEN imisa imodeli eyisisekelo futhi ifake izendlalelo zokuzinaka ezifakwe esangweni, ukuze ulwazi lwangempela olufundiwe luhlale lunjalo.
Kuyini okokufaka okuyisisekelo okujwayelekile okwamukelwa yi-GLIGEN?
I-GLIGEN isekela ukubeka phansi ngamabhokisi ahlanganisayo anombhalo/isithombe, amaphuzu abalulekile, nemifanekiso eyireferensi.
Kungani isango lezendlalelo zokunaka ezintsha ze-GLIGEN liqalwa ukuze libe uziro?
Isango elaqalwa uziro lisho ukuthi indawo ephansi ayinawo umphumela ekuqaleni, igcina imodeli yasekuqaleni kanye nokugcina ukuqeqeshwa kuzinzile njengoba kunyuka amarampu okulawula.
Kusho ukuthini isizukulwane esisekelwe phansi 'somhlaba ovulekile' ku-GLIGEN?
I-GLIGEN yenza okujwayelekile ngale kwesethi yayo yokuqeqeshwa okuyisisekelo, ibeka inoveli izinto ezichazwe ezindaweni ezithile.