AI startup Luma has released Genie: a text-to-3D image model that works “in under 10 seconds”. The announcement coincides with a Series B funding round that brought Luma’s total investments to more than $70 million, the company says.
Genie’s interface will be undoubtedly familiar to anyone who has used an AI tool before. Users type a description of the model they’d like to generate, click “Use Prompt”, and Genie will generate four options of quad mesh models. Out of ideas? Just click one of the user-generated floating images on Genie’s landing page and its text prompt will appear.
Once generated, users can download their image in FBX, GLTF, USDZ, Blend, STL, or OBJ format. They also have the option to generate a high-resolution version of their model, although this takes significantly longer. Our test of a high-res model, for example, was still generating after more than 15 minutes.
Overall, the models do seem to be of higher fidelity than other text-to-3D model generators we’ve seen so far of which Genie joins the growing ranks.
It’s unclear if Genie uses a diffusion model like Point-E and DreamFusion, implicit functions like Shap-E, or another method entirely. The company appears to be holding the technology used to power its generator much closer to its chest than other companies in the field.
Regardless, such generators have large implications for use with 3D printing, even if the technology might not yet fully be realized. As it evolves, we see them being used in everything from the creation of prototypes and educational tools to customized or personalized models, and much more. True to its name, Genie appears to have the potential to be a wish-granter.
Those interested in trying Genie out for themselves can do so through Luma’s website and Discord community, or via the Genie iOS app.
You’ve read that; now read these:
License: The text of "This AI Tool Creates 3D Models from Text Prompts in Seconds" by All3DP is licensed under a Creative Commons Attribution 4.0 International License.