Adobe's AI text-to-3D world concept is a glimpse of where Firefly could go next

Adobe's been rolling out AI tool after AI tool in the past couple of years, from Firefly's image and video generation models to the recent much-debated addition of AI Relight in Photoshop.

The publication of a huge batch of patent filings shows that the creative tech giant has plenty of ideas about where it could take things next, and that includes into field where Adobe doesn't currently have a strong presence.

Adobe patent for AI text to 3D world generator

(Image credit: Adobe / USPTO)

As noticed by the patent tracking site Patentlyze, the US Patent and Trademark office has just published a series of patent filings from Adobe. One of them describes a method for creating navigable 3D worlds from text descriptions.

Latest Videos FromCreative Bloq

An example use cited describes a user typing "a foggy Japanese garden at dusk" to generate said 3D environment. A non-specified AI model would first build an intermediate "spherical" 2D picture of the space, like a 360 view produced by stitching 3D scans. That would then be fed back into a second AI process that would use it as a basis for the construction of 3D geometry.

The output would be viewable from multiple perspectives, and the user would be able to move forwards and backwards to explore the environment.

Even with today's procedural tools that allow artists to generate assets using rules, building 3D environments is a time consuming process that requires potentially hundreds of individual assets that are digitally sculpted and then painted with surface details.

Adobe has a suite of 3D programs in its Substance collection of apps. Substance Painter is fairly popular for that part of the process, but Substance 3D modeler remains a relatively niche choice compared to the likes of ZBrush and Blender, with its main advantage being the seamless switching between desktop and VR. For now, it's one of the Adobe programs that hasn't added Firefly-powered generative AI tools.

Adobe's not the only company exploring the idea of text-to 3D world generation. For the moment the nearest thing are tools like Blockade Labs' Skybox AI , which generates 360-degree scenes from text prompts, but these are quite some way from navigable environments for games or VR and AR.

This is just one of the dozens of patents from Adobe published in recent weeks. Other filings include concepts for an AI tool that builds new camera angles from a single photo, one that finds matching materials across 3D models and a tool for rendering a 3D garment preview.

A patent application doesn't necessarily signal a development intention. Companies file patents for various reasons, including to try to prevent competitors from using an idea and to show shareholders that they're being innovative.

Joe Foley
Freelance journalist and editor

Joe is a regular freelance journalist and editor at Creative Bloq. He writes news, features and buying guides and keeps track of the best equipment and software for creatives, from video editing programs to monitors and accessories. A veteran news writer and photographer, he now works as a project manager at the London and Buenos Aires-based design, production and branding agency Hermana Creatives. There he manages a team of designers, photographers and video editors who specialise in producing visual content and design assets for the hospitality sector. He also dances Argentine tango.

You must confirm your public display name before commenting

Please logout and then login again, you will then be prompted to enter your display name.