World Labs: Umholi kwezobuhlakani bokuqonda indawo nama-Large World Models.

#Quasa #QUA #worldlabs

I-World Labs (evezwe ku-Quasa.io/projects/world-labs) iyinkampani ephambili kwezobuhlakani bokuqonda indawo eyakha ama-Large World Models (LWMs) ngo-2026.

Isungulwe yingqalabutho ye-AI u-Dr. Fei-Fei Li (ovame ukubizwa ngokuthi “uMama we-AI”) kanye no-Justin Johnson, u-Christoph Lassner no-Ben Mildenhall, i-World Labs yakha izinhlelo ze-AI zesizukulwane esilandelayo ezikwazi ukubona, ukukhiqiza, ukucabanga ngomhlaba we-3D nokusebenzisana nawo — zisusa i-AI ezithombeni eziyisicaba ze-2D ziyiyise ezindaweni ze-3D ezihlala zikhona futhi ezingahlolwa.

Ingqikithi yalokho okulethwa yi-World Labs iwukwenza kube ngokoqobo ukuguquka “kwamaphikseli → abe yimihlaba ye-3D ephilayo”, kugxilwe kakhulu ebuhlakani bokuqonda indawo. 

Amaphuzu abalulekile ahlanganisa:

• I-Marble — umkhiqizo wabo ohamba phambili okhiqiza imihlaba ye-3D enokunemba okuphezulu, ehlala ikhona futhi evumela ukusebenzisana, kusetshenziswa umbhalo, izithombe, amavidiyo noma izinhlelo zokuhlelwa kwezinto;
• Ama-Large World Models (LWMs) — amamodeli athuthukile asebenzisa izinhlobo eziningi zolwazi futhi aqonda ijiyomethri, imithetho yemvelo nobudlelwano bezinto ngokwendawo;
• Ukuhlela Nokunweba Ngokuqondile — abasebenzisi bangahlola, bahlele, bahlanganise futhi banwebe imihlaba ekhiqiziwe ngesikhathi sangempela;
• Ukusetshenziswa Ekukhiqizeni Nasekusunguleni — amathuluzi anamandla amafilimu, i-VFX, ukukhiqiza kusetshenziswa izindawo ezibonakalayo ezenziwe ngekhompyutha, imidlalo, i-AR/VR, ukuklama izakhiwo, ubuchwepheshe bamarobhothi nokuklama;
• Imiphumela Yekhwalithi Ephezulu — izinto ze-3D ezingathunyelwa kwezinye izinhlelo, ezinokuvumelana okuqinile nemininingwane eminingi.

Ifaneleka kahle kubenzi bamafilimu, abathuthukisi bemidlalo, abaculi be-3D, abaklami bezakhiwo, onjiniyela bamarobhothi, amaqembu okukhiqiza asebenzisa izindawo ezibonakalayo ezenziwe ngekhompyutha, nanoma yimuphi umsunguli noma inkampani edinga ukuvala igebe phakathi kokucatshangwayo nokungokoqobo kwe-3D okuvumela ukusebenzisana.

Ngo-2026, i-World Labs isiqoqe uxhaso lwezimali olukhulu (oluhlanganisa oluvela ku-Autodesk, ku-Cisco nakwabanye) futhi iyaqhubeka nokwedlula imingcele ngokuthuthukisa ukukhiqizwa kwemihlaba, ukusebenza okungcono ngesikhathi sangempela nezindawo ze-3D ezilawuleka kakhudlwana.

Izimboni zokusungula nezobuchwepheshe zijabule:
“I-World Labs ichaza kabusha okungenzeka ngokunika i-AI ukuqonda kwangempela kwendawo.”
“I-Marble isivumela ukuba sikhiqize imihlaba ephelele engahlolwa ngomyalo owodwa — okuyigxathu elikhulu ekukhiqizeni kusetshenziswa izindawo ezibonakalayo ezenziwe ngekhompyutha.”
“Ekugcineni, sekukhona i-AI engagcini nje ngokukhiqiza izithombe — yakha izindawo ze-3D eziphilayo nezivumela ukusebenzisana.”

Ivelela ikakhulukazi ekwakheni imihlaba ye-3D ehlala ikhona futhi okungahambahambwa kuyo, inikeza ikhono eliqinile lokucabanga ngobudlelwano bezinto ngokwendawo, yenza kube nokuhlela okuvumela ukusebenzisana kwangempela, futhi isheshisa izinqubo zomsebenzi kumafilimu, emidlalweni, ekuklameni nasebuchwephesheni bamarobhothi.

Ububi: Njengoba iyinkampani yocwaningo oluphambili, ukufinyelela ezicini ezithuthuke kakhulu kuvame ukukhawulelwa noma kube sesigabeni sokuhlola se-beta; ukukhiqiza imihlaba emikhulu kakhulu/eyinkimbinkimbi kungadinga amandla amaningi ekhompyutha; futhi njengamathuluzi amaningi akhiqiza okuqukethwe kwe-3D, imiphumela ingase isadinge ukucutshungulwa okwengeziwe ukuze isetshenziswe ekukhiqizeni kokugcina.

Sekukonke, kulabo abasebenza phambili kwezobuchwepheshe be-AI yokuqonda indawo nokwakhiwa kokuqukethwe kwe-3D ngo-2026, i-World Labs ingenye yezinkampani ezithembisa kakhulu nezinezifiso ezinkulu kulo mkhakha. Umsebenzi wabo kuma-Large World Models umelela igxathu elikhulu elibheke ku-AI ye-3D ehlakaniphe ngempela. Thola nomvuzo ongu-1 QUA nge-Quasa!

Izinkanyezi ezingu-4.8/5 (ivelele ngombono, ukusungula nekhwalithi yobuchwepheshe; kunamaphuzu amancane mayelana nokufinyelela kwamanje nezidingo zokuthuthukiswa okwengeziwe).

Qalisa: https://quasa.io/projects/world-labs