Topics
Latest
AI
Amazon
Image Credits:koto_feja / Getty Images
Apps
Biotech & Health
Climate
Image Credits:koto_feja / Getty Images
Cloud Computing
Commerce
Crypto
Image Credits:Luma
endeavor
EVs
Fintech
Image Credits:Luma
fundraise
Gadgets
Gaming
Government & Policy
Hardware
layoff
Media & Entertainment
Meta
Microsoft
Privacy
Robotics
security measures
societal
place
Startups
TikTok
transport
Venture
More from TechCrunch
Events
Startup Battlefield
StrictlyVC
Podcasts
Videos
Partner Content
TechCrunch Brand Studio
Crunchboard
Contact Us
A couple of years ago , Alex Yu and Amit Jain came together to found a party that ’d allow citizenry entrance object in 3D using their smartphones — no additional equipment required . At the clock time , Yu was an AI researcher at UC Berkeley while Jain was an Apple employee building out theVision Pro‘s multimedia system experiences .
Their company , Luma , launched a smartphone app in 2021 , which quickly gained adhesive friction — going on to attract millions of users ( just over two million as of publication clip ) . But now , as reproductive AI technical school overflow the channel , Yu and Jain hope to evolve Luma into something handsome — and , with any luck , better — than they primitively envisioned .
Luma today announced that it ’ll start leveraging a compute cluster of ~3,000 Nvidia A100 GPUs to train new AI models that can — in Yu ’s Son — “ see and understand , show and explain and eventually interact with [ the ] world . ”
Meet Genie , a research trailer of an all - Modern form of generative three-D foot model#MadeWithGenie
💬 Create 3D thing in seconds on Discord ⚡ ️ Prototype in various styles 🎨 Customize stuff 🆓 Free during research preview
Try it now 👇 https://t.co/NVgbsvn2XUpic.twitter.com/qm7PkdMEAp
— Luma AI ( @LumaLabsAI)November 1 , 2023
Join us at TechCrunch Sessions: AI
Exhibit at TechCrunch Sessions: AI
The first phase of this plan entail creating models open of generating 3D objects from text edition descriptions ; Luma launched one such model on its Discord server earlier this twelvemonth , call Genie . The next will be recrudesce “ next gen ” generative AI models that address what Yu characterizes as the “ uncanny valley ” problem in current - gen GenAI .
“ We believe that multimodality is critical for intelligence . To go beyond language models , the next unlock will come from visual modality , ” Yu told TechCrunch in an email interview . “ [ However , ] AI demand to get a lot chic to fork up the potential the world see in it .
To realize thisvisionvision ( pardon the pun ) , Luma has raised $ 43 million in a Series B round with participation from Andreessen Horowitz among other backer old and newfangled . According to a rootage familiar with the issue , the round esteem Luma at between $ 200 million and $ 300 million ; Luma ’s war chest now stands at more than $ 70 million .
Luma ’s current focus — launch 3D model - creating AI models — is an increasingly competitive space . There ’s object - crafting political platform like 3DFY and Scenario , as well as startup such as Hypothetic , Kaedim , Auctoriaand Mirage . Stability AI recentlylauncheda 3-D - model - generate standalone puppet as did newer ventureAtlas . Even incumbent like Autodesk and Nvidia are set out to dip their toes in the sector with apps like Get3D , which converts effigy to 3D models , and ClipForge , which generates models from schoolbook description .
So how will Luma ’s tool place upright apart ? Fidelity primarily , Yu says .
“ Current models are all being trained on two - dimensional images and , when inquire to generate conniption , they mangle place , bodies and drive , ” he enunciate . “ It ’s very difficult to beget anything ordered and usable in the first few effort , limiting where you could expend the outputs … [ We ’re bringing ] about the most modern generative photorealistic technolog[ies ] in an nonrational app . ”
That ’s forebode a batch considering it ’s too soon in Luma ’s challenging new roadmap . An improved version of Genie launches today , but future , more subject procreative AI model are a ways aside .
Luma ’s wasting no time though , planning to duplicate its 24 - someone workforce by the end of next year while piece together a model - running host cluster of “ thousands ” of GPUs . Perhaps it ’ll make headway after all ; time will say .
“ We ’ve been growing the squad across productive AI research , engineering , design and Cartesian product in rules of order to bring our vision to life , and plan to speed up the pace here significantly play along this round , ” Jain said . “ With Genie , for the first time make 3D things at scale of measurement has become possible with AI , and that ’s grown to 100,000 drug user in just four week … [ But we want to]build vastly more adequate to , intelligent , and utilitarian visual exemplar for our substance abuser . ”
