A published Adobe patent application describes a possible path from text input to an explorable 3D environment. However, it confirms neither a concrete product development nor the availability of a tool.
A published Adobe patent application describes a method by which an explorable 3D environment could be generated from text input. The outlined approach first goes through a spherical 2D image. This does not confirm that Adobe is working on a corresponding tool or intends to release it.
From text via a 2D image to 3D geometry
According to Creative Bloq, the concept provides for two processing steps. First, an AI model is supposed to generate a spherical 2D image of the environment from the description. This image then serves as the basis for a further AI step that creates 3D geometry from it.
According to the described method, the text input would therefore not directly produce a finished 3D world. Instead, the spherical image would be an intermediate step between description and geometry. The patent application thus sets out a possible technical workflow for text-to-3D environments.
Exploring the environment from different perspectives
The generated environment is supposed to be viewable from multiple perspectives. In addition, it is supposed to be possible to move forwards and backwards within it. This distinguishes the described result from a single image: what is envisaged is an environment that you can view from different angles and explore in both directions.
The application thus describes the intended properties of the environment. However, it does not prove that such a method has already been implemented as an Adobe tool or can be used.
What the patent application means for you in practice
For you, the application is above all an insight into a possible approach to using text input to create navigable 3D worlds. However, no available feature can be derived from the description. According to the source, a concrete intention to develop or release something also does not necessarily follow from a patent application.
The difference is important: a published patent document describes an idea or a method, but it is not a product announcement. Whether Adobe actually develops the outlined approach further or makes it accessible as a tool cannot be determined from the application alone.
I myself once tried out a few 3D texts with Photoshop in the past and I quite liked them, but it was a lot of work. That's why I now find the AI approach very exciting and will definitely take a look at it myself. Will you too?
Sources
