ARTV
Artificial
Reality
Television

ARTV Artificial

ARTV (Artificial Reality Television, also ART Television) is an artistic project

In an era of rapid artificial intelligence development, the ARTV project emerges as a fascinating glimpse into the future of art and entertainment. It is an innovative venture that combines advanced AI technologies to create a fully autonomous, generative audiovisual stream. ARTV is not just a content creation tool – it’s a comprehensive ecosystem that independently generates, composes, and presents unique artistic experiences

This project utilizes state-of-the-art language models, video synthesis systems, and music generators to create a seamless stream of content that continually evolves and surprises the viewer. A key element of ARTV is randomness at every stage of the creative process, ensuring that each experience is unrepeatable and unpredictable.

ARTV showcases AI’s technological capabilities and poses fundamental questions about the nature of creativity, authorship, and the perception of art in the age of AI. In the following sections, we will examine the critical aspects of this project and its ambitious artistic goals.

Prompt Generation
Advanced language models, such as GPT, are used to create creative scene descriptions that serve as stimuli for generated content. These models generate prompts, which are scenario suggestions that are then transformed into visual content.

Video Synthesis
Cutting-edge text-to-video models transform textual descriptions into dynamic visual sequences. This process involves tools and libraries like HuggingFace and GitHub, enabling advanced natural language processing and generating visual content.

Music Generation
AI-generated soundtracks are integrated with visual content, creating a complete multimedia experience. The music is automatically created, adapting to the mood and dynamics of the generated scenes.

Automatic Compilation
The generated clips are seamlessly combined into a continuous stream, simulating the experience of watching television but with content entirely created by AI. Google Colab coordinates and processes data in real-time, enabling smooth content generation and compilation.

Randomness as a Creative Element
ARTV employs randomness at every stage of the creative process, from the random selection of keywords for generating prompts through stochastic processes in generative models to unpredictable connections between image and sound. This randomness is a crucial element of the project, ensuring that every ARTV experience is unique and unrepeatable.

Artistic Goals

• Exploring the potential of AI as an autonomous creator of artistic content
• Challenging traditional concepts of authorship and artistic intention
• Exploring new forms of visual storytelling beyond human imagination
• Reflecting on the relationship between artificial intelligence and human perception and interpretation of art
• Celebrating unpredictability and spontaneity in generative art.

Sketch ARTV version I, July 2024

ARTV doesn’t just create art; it invites the viewer to participate in a continuous act of creation. The built-in randomness ensures that each sequence is unique, placing the viewer as a witness to a constantly evolving artistic process. This demonstrates technological possibilities and a profound exploration of the nature of creativity, perception, and reality in the age of artificial intelligence.

ARTV (Artificial Reality Television and ART Television) is a groundbreaking artistic project redefining the boundaries of generative art and digital media. Artificial intelligence generates the entire audiovisual experience, creating unique, unparalleled experiences for viewers.

Currently, ARTV is in its experimental test phase. Streams are published on YouTube on the ARTV_Artificial channel.

Artwork Usage License / Credits:

1. This work is licensed under Creative Commons Attribution-NonCommercial 4.0 (CC BY-NC 4.0).
2. The copyright owner of this work is Mateusz Beźnic.
3. You are allowed to:
   • You are not allowed to:
   • Use the work for commercial purposes without the author’s consent
   • For any use of the work, you must credit the author, provide the license information, and mention the AI models used.
   • This license is valid indefinitely.
   • Share and copy the work
   • Adapt and modify the work
4. If you have any questions, please get in touch with the author.
5. This work was created using the following AI models:
   • Zeroscope v2 576w and Zeroscope v2 XL (CreativeML Open RAIL-M license)
   • MusicGen (CC-BY-NC 4.0 license)
   • GPT-2 (MIT license)
   • Node Media Server (MIT license)
6. Inspiration:
   • Julian Antonisz’s exhibition at Google Cultural Institute
   • Jak Działa Jamniczek? (How Does the Little Dachshund Work?) reż, J. Antonisz
   • Creative Process: Norman McLaren – NFB
   • Norman McLaren – NFB
   • Julian Bilcke, Ai WebTV
   • Alara Dirik, Text-to-Video: The Task, Challenges and the Current State