Dutch design studio Modem Works has unveiled the 'Dream Recorder', a device that reconstructs the user's dreams using AI and shows them as videos.
The Dream Recorder is a device that allows users to speak their dreams out loud immediately after waking up, and the device then converts them into text, which is then interpreted by AI and recreated as a video. Users can choose the visual style they want, and the resulting image is played back as a low-resolution, impressionistic “dreamscape.” This is intentionally implemented in a low resolution to visually express the blurriness and illogical development characteristic of dreams. Modem says the device is designed to be a “phone-free haven” in the bedroom.
The Dream Recorder is not a commercial product, but was developed for research and creation purposes. The entire source code, hardware manual, and even the design files for 3D printing are available on GitHub, and anyone can assemble it using open source. It is a structure that can be assembled without soldering, and can be manufactured with only about 285 euros in parts.
The core of the device is text analysis and video generation. The dream story told by the user is converted into voice-to-text, ChatGPT refines the story structure, and Luma AI then visualizes it. Artist Alexis Jammeh participated in this, giving it an aesthetic composed of dreamy colors and blurred textures.
Modem explained the identity of the dream recorder by saying, “AI does not see the world, but predicts it.” Dreams are originally a series of imperfect and illogical images, and the distorted and fragmented images generated by AI are more like dreams. This is an example of technology’s errors and imperfections actually acting as charm.
The device can store up to a week's worth of dream footage, allowing users to visually track recurring symbols or emotions.
