top of page


by Cristian Peñas

An Interactive Journey Through Lucid Dreams

What is Latens

Latens is an application that proposes a very innovative way to use real-time generative AI: using AI-generated images as the frames of an interactive movie. This particular use opens the door to endless possibilities, allowing the user to immerse themselves in an immersive, dynamic, and fully customizable experience.

In essence, Latens could be considered the first 'lucid dream simulator'.

What makes Latens special

Unlike other projects, Latens does not use real-time generation as a 'filter', as do assisted drawing applications and AI rendering applications. Depending on pre-established guides, although it provides more consistent results, limits the true potential and freedom of this technology.

The main proposal of Latens is to allow you to navigate through the scenarios or 'dreams' that the AI imagines for itself, giving you control over them through text prompts, camera movements, and other simple inputs.

The technology behind it

The basis of Latens is the "image-to-image" technique, which simulates a continuous flow of frames. Each generated image is built from the previous one, with small real-time variations, such as zoom or camera movements. This creates the sensation of navigating through a movie that unfolds and adapts to your will. Among the variety of innovative techniques that the project experiments with, this one undoubtedly stands out as the most ambitious of all.

The Concept of "Latent"

"Latens" comes from Latin and means "hidden" or "latent". This word evokes the idea of something that is present but not visible or manifest, something that has a potential or a reality that has not yet been revealed. In this context, Latens can be interpreted as the artistic representation of the "Latent Space" itself, an entropic canvas where all existing possibilities and realities await to be manifested.

First steps

When you access Latens, you encounter a screen where curiously nothing is displayed but "noise". That is the "entropic broth", where any reality you wish is in a latent state, waiting to be manifested. Through a text box, you write what you want to visualize, and you will see how what was once noise begins to materialize. With controls similar to those of a first-person video game, you begin to explore and interact with the environment that is generated before your eyes.

Interaction proposals

Currently, Latens is in an experimental phase and, for now, offers 2 main ways to interact with it:

  • Flux: An exploratory mode where every frame and every pixel shown on screen is completely generated. Flux works through an uninterrupted flow of images generated in real time, which function as the frames or "graphics" of this mode.

  • Limbo: A space with mixed graphics, where you can generate and expand the environment at will by simply pressing a button. The AI-generated images become 3D holograms to simulate the scenario through your prompt.

Future and Evolution of Latens

Latens seeks to advance at the pace of developments in AI, refining the current experience and including new modes of interaction with the latent space, such as a range of creative tools. It also aims to orient towards a more playful theme, with adventures and mini-games, and plans for more accessible mobile versions and a VR adaptation, where the Latens experience would reach its maximum immersion.

The key to innovation and effectiveness

In a constantly evolving sector, small teams with strong synergies and individual developers have a significant advantage. Possessing a broad and centralized knowledge facilitates the generation and implementation of innovative ideas much more effectively. Latens is a clear example of this. Currently designed by a single developer, it benefits from combining and catalyzing experience in areas such as conceptual art, design, VFX, game development, programming, specialization in generative AI, etc.

The potential reach

Thanks to its highly customizable nature, Latens allows users to immerse themselves in experiences tailored to their tastes and preferences. This means that the target audience for Latens is not limited to a specific profile but is broad and diverse. With plans to expand to mobile platforms, PC, and virtual reality (VR), the potential reach of Latens will be further expanded.

The questioning of Generative AI

Latens addresses the use of generative image models in a unique and positive way, convinced that the advantages of these technologies far outweigh their possible drawbacks. The main focus is on exploring their innovative potential, rather than replacing traditional methods. The technique of generating real-time frames to create "interactive movies" is a revolutionary example of this approach, demonstrating the highly constructive use of AI models.

An exciting horizon

With ambition, ingenuity, and a set of creative ideas, Latens is preparing for a very promising future. It wants to be the demonstration of the true potential of generative AI and invites you to be part of the beginning of a new era in digital interaction.

Try it out

(PC Web Browser Only)

bottom of page