Online Workshop: Adventures in Latent Space – Exploring LLMs and Real-Time Image Generation

Radiona.org is inviting you to international online workshop on AI titled ‘Adventures in Latent Space – Exploring LLMs and Real-Time Image Generation’ led by Marlon Barrios Solano which is set to take place on Tuesday and Thursday, 10th and 12th December 2024 from 6 to 9pm CET. Marlon Barrios Solano is maker-in-residence at Center for Arts, Migration, and Entrepreneurship (CAME), University of Florida

Apply here: https://forms.gle/Avw7BreAS5RJdm2H7
The workshop is free of charge.

Duration: 6 hours total, divided into two online sessions (3 hours each)

Workshop Overview:

Adventures in Latent Space is an interactive, no-code workshop designed to introduce participants to the creative potential of Large Language Models (LLMs) and real-time image generation using open-source models. This workshop aims to provide a hands-on exploration of AI’s latent space through engaging activities and conceptual discussions, perfect for artists, creative technologists, and anyone interested in the intersection of technology and the arts.

Objectives:

  • Understand the basics of latent space and its significance in generative AI.
  • Learn how to utilize LLMs for storytelling and creative ideation.
  • Experience real-time image generation using intuitive, open-source tools and prompts.
  • Engage in the embodied exploration of AI responses through participatory activities.

Session 1: Navigating Latent Spaces with LLMs
Time:
Tuesday, 10th December 2024, 6-9pm
Duration: 3 hours

Agenda:

  1. Introduction to Latent Space and Generative AI (30 min)
  2. Definition and significance of latent space.
  3. Overview of LLMs and how they relate to creative processes.
  4. Real-world applications in art and storytelling.
  • Hands-On Interaction with LLMs (45 min)
  • Guided exercises using open-source LLM interfaces to generate creative text outputs.
  • Group exploration of prompts and AI-generated storytelling.
  • Interactive challenges to experiment with AI as a co-creator.
  • Exploring AI in Storytelling and Conceptual Art (45 min)
  • Techniques for developing narrative structures with LLMs.
  • Using AI responses to inspire or transform artistic projects.
  • Integrating LLM-generated text into multimedia artworks.
  • Discussion and Reflection (30 min)
  • Share experiences and insights from the session.
  • Group discussion on the potential challenges and benefits of using LLMs in creative contexts.

Wrap-Up for Session 1:

  • Recap of key takeaways.
  • Brief overview of real-time image generation and preview of Session 2.

Session 2: Real-Time Image Generation and Interactive Practice
Time: Thursday, 12th December 2024, 6-9pm
Duration:
 3 hours

Agenda:

1.  Introduction to Real-Time Image Generation Tools (30 min)

  • Overview of open-source tools such as DALL-E, Stable Diffusion, and RunwayML.
  • Explanation of how prompts influence outputs and the creative process.

2. Hands-On Image Creation (60 min)

  • Interactive demonstration of generating images based on text prompts.
  • Group activities exploring variations in prompt design and output.
  • Real-time feedback and adjustments to prompts to shape visual results.

3. Creative Integration: Text Meets Image (45 min)

  • Combining LLM-generated text with image generation for cohesive multimedia projects.
  • Examples and exercises for creating thematic connections between visual and textual outputs.
  • Collaborative mini-projects where participants develop short narratives accompanied by AI-generated visuals.

4. Showcase and Interactive Feedback (30 min)

  • Participants present their final projects or share their creative process.
  • Peer feedback and group discussion on the potential of latent space in artistic practices.

Conclusion:

  • Final thoughts on the importance of AI as a creative tool.
  • Tips for further exploration and resources for self-guided learning.

Pre-Requisites:

  • No coding experience required.
  • A computer with a stable internet connection and access to online tools.
  • A willingness to engage and experiment with creative processes.

Outcome:
Participants will leave with a practical understanding of how to navigate and use latent spaces in LLMs and real-time image generation for creative projects. They will gain insights into combining AI-driven text and visuals for innovative storytelling and artistic practices, using accessible, open-source models.

Marlon Barrios Solano (c)

About the workshop leader:
Marlon Barrios Solano is a Venezuelan-American interdisciplinary artist, creative technologist and researcher with a background in dance, software engineering and cognitive science working with generative AI, machine learning, creative coding and interactive performance. Since August 2024, he is the Maker-in-Residence focused at CAME Center for Arts, Migration and Entrepteneurship foocused on AI, Art and Diasporas at the University of Florida. He is currently an artist-in-residence with the Rewilding Cultures program at Radiona.org in Zagreb, Croatia, and at Lake Studios Berlin, Germany, where he is a founding member.

Marlon’s work investigates computational creativity and synthetic cognition designing hybrid complex systems where language and self-organizing processes generate the aesthetic experience. His artistic practice includes machine learning for embodied digital interaction, generative AI, movement and voice, creative coding and app development, generative writing, bottom-up robotics, dance improvisation and vipassana meditation. His works are deployed as dynamic browser based art, open source code, improvisational scores, sound environments, AI chatbots and video that he organizes as installations, workshops, participatory performances and lectures. All the work is published and distributed with an open source license.

In 2007, Marlon launched dance-tech.net, a social network for interdisciplinary explorers of movement performance, innovators, and emergent performance practices. He has produced more than 200 video interviews as one of the first international video bloggers dedicated to the intersection of dance, arts, and digital technology, published several essays, and created pioneering XR projects for knowledge transmission within dance/art festivals. In 2011 movimiento.org was created in Spanish and Portuguese and supported by the South American Network of Dance. These networks have received support from Motionbank and the Forsythe Company (Germany), Panorama Festival (Brazil), Transmediaakademy (Germany), Lake Studios Berlin, among many others.

He was a research associate at the Inter-University of Dance/University of The Arts (UDK/HZT) in Berlin, Germany, from 2013 to 2016, a 2017 Hombroich Fellow in Germany, and an artist/researcher in residence at ICK Amsterdam, Netherlands, in 2013-14 and at the Gilles Jobin Company in Geneva, Switzerland, from 2009-2012.

Marlon has written essays for the books Dance In the Head/Tanz im Kopf and Transmission in Motion: The Technologizing of Dance edited by Johannes Birringer and Maiike Bleeker respectively. Some of his interviews have been transcribed and published by Contact Quarterly.

As a contemporary dancer and improviser in New York (1994-2001), Marlon has collaborated with choreographers such as Lynn Shapiro, Merian Soto, Dean Moss, Bill Young, and Susan Marshall, and musicians such as Philip Glass, John Zorn, and Erik Friedlander. He has participated in numerous festivals and taught both nationally and internationally. He studied and practiced improvisation with Nancy Stark Smith, Jennifer Monson, Bebe Miller, KJ Holmes, and David Zambrano. In Venezuela (1981-1994), he apprenticed with DanzaLuz, DanzaHoy, and danced professionally with Aktion Colectiva and Rajatabla Danza. He co-founded Danza Contemporánea de Maracaibo with Yasmin Villavicencio in 1986.

Marlon holds an MFA in Dance and Technology from The Ohio State University, USA (2004) and completed the General Assembly Software Engineering Immersive Program 2021. He is also certified in Vipassana/Mindfulness Meditation by Spirit Rock Meditation Center, a certified 200 Hours Embodyoga Teacher and has studied one year of the Somatic Experiencing Certification Program.

Organizers: Radiona.org

Supported by: City of Zagreb, Ministry of Culture and Media of Republic Croatia, Kultura nova Foundation