“Venus” is a music video directed by Christoph Pretzsch, Sophie Kreutzer and Patrick Aere as part of the LICHTSTÜCKE project by the via-nova-chor München.
Inspired by movies like Interstellar and Stanley Kubrick’s 2001: A Space Odyssey, the video explores the intersection of music, space and the human imagination. Mixing self-recorded footage with generative visuals created in TouchDesigner and Resolume, as well as some AI-generated video sequences, Venus visualizes Michael Ostrzyga’s composition as an abstract space landscape that mirrors the complexity and beauty of the piece.
LICHTSTÜCKE was a collaboration from students at LMU München (Art and Media program) and the Hochschule für Technik Berlin (Screen-Based Media program). It was sponsored by Amateurmusikfonds, BMCO and Bundesregierung für Kultur and Medien. The video premiered at Technikum in München and was also presented at chor.com 2024.
Which AI tools have been used? What were the prompts?
For the AI-generated sequences, Runway ML was used as the primary tool. It was used specifically for the moments in the video where self-built alternatives were unfeasible, such as long shots of the spaceship taking off or floating in outer space, because in these cases, creating 3D renders would have been too time intensive for our small team. For the prompting, we referenced films such as Cristopher Nolan’s Interstellar and Stanley Kubrick’s 2001: A Space Odyssey, indicating the type of field sizes („totale“), colors („black and white“), resolution („ultra high definition/Unreal Engine“) and action („disintegrating“/„falling apart“). To fine-tune the results, we worked iteratively, first creating images and then using them as input to create short video sequences.
Another instance where we used AI was to create an image of a female astronaut. We decided not to use one of the choir members to avoid putting too much focus on one of the choir members and instead having a more generalized visual. We then worked with said visual and manipulated it using a compute shader in TouchDesigner called particlesGpu, creating our own animation. There, we did were more vague with the prompts, allowing the machine to surprise us.
Reflection of the process: What was the intention? How was the experience with the interaction with the AI tool? How are the results evaluated? Happy or disappointed? Did AI meet the expectations?
We were satisfied with the results, and surprised even that the AI tool was able to create such sophisticated and detailed visuals of what we had in mind. Still, a lot of iterations were necessary to create the specific images we needed for the video. Another challenge was the length of the video sequences, because the common output was only a few seconds long. Had we worked with self-built 3D renders in Blender for example, we could have done longer shots, but given the time constraints and budget, working with the AI tool worked just fine.
For our „Venus“ music video, the focus was not on using only AI visuals, but a combination of digital and analog techniques. What enriched our video at the end was using a Super 8 camera, generative visuals, VJing software, etc. in combination with the AI sequences. The result was a timeless look-and-feel that the viewer could not categorize into being of the past (analog film) or the future (AI film), but somewhere in between.
Patrick Aere is a Swiss-Costa Rican senior media designer and visual artist based in Munich, Germany. He is part of the new media art collective Fobin.Kollektiv and NFT collective The Drops. He initially pursued his passion for art by studying Fine Arts at the University of Costa Rica (UCR) and later at the Ludwig-Maximilians Universität (LMU) in Germany. To further his growth and knowledge in design, he continued his education by participating in design workshops at institutions such as the Copenhagen Institute of Interaction Design (CIID) and Veritas University. In his personal artistic practice, he explores the relationship between humans and digital technology through illustration, generative art, artificial intelligence, physical computing, VJing and animations.
Christoph Pretzsch has a background in computer Science, communication design and is currently studying Multimedia Art at LMU. In his works he fuses his passions for the arts and technology, and he is a member of the Fobin.Kollektiv. He does not constrict himself to one medium, but passionately enjoys combining and intercepting different forms of digital and analog media. These include, but aren't limited to: 3D Animation, Generative Visuals, Interactive Media Installations, Sculpture, and Video Art.
Sophie Kreutzer completed her arts and anthropology studies in 2019 at the Freie Universität Berlin and has since worked with various social and cultural organizations. She is currently studying art education at Ludwig-Maximilian-Universität in Munich and is a member of the Fobin.Kollektiv. Alongside her studies, she works as a self-taught freelance graphic designer and is involved in organizing music and cultural events.