Mode 01 • System
Speech to video
We turn spoken sentences into evolving visuals in real-time, collected using a microphone.
PORTAL
An art installation that enables the real-time generation of a continuous video from written or spoken words.
System
Mode 02 • System
Text to video
We generate a continuous video from queued text prompts written by the audience.
Moderation
The installation comes with three levels of moderation.
Moderation layer
Natural moderation
The algorithms used for visual creation are trained on a curated dataset, allowing objectionable content to be excluded without removing creative range.
Moderation layer
Automatic moderation
A ban list can be configured with forbidden words or phrases. Refused prompts are rejected immediately and the participant can submit a new one.
Moderation layer
Human moderation
A live interface allows manual approval, rejection, prioritization, and timing control for each prompt when an event requires tighter supervision.
Style
We can adapt the visual generation to your artistic direction, from color palette to motifs and overall atmosphere. A small set of inspiration references is enough to shape a dedicated style.
Gallery
Each project is specific, and each event is unique.
Each installation showcases a different approach to real-time AI video generation, offering inspiration for how you can integrate this experience into your own projects.
Contact
Want to design your own live experience?
The interactive project brief loads on demand. You can already reach the team directly by email if you want to discuss a live activation now.
Email ObviousAbout Us
We are Obvious, a French trio of artists and researchers working with artificial intelligence to create art. Inspired by Renaissance workshops, we operate at the crossroads of academic research and artistic production.
Our research lab at Sorbonne University, funded by the French National Research Agency, develops image, video, and sound generation systems. We were behind the first AI-created artwork to go through Christie’s in 2018.
