Acapulco
114’, 2023-2024. W/ Pedro Gallego.
In the context of research fellowship at Collegium Helveticum ETH. More here.
Employing methodologies rooted in critical pedagogy, Moreschi crafted an immersive experience for the “seers” who participate in this film. Leading experts from around the world received printed postcards by mail featuring some images used to train IA/ Computer Vision, while crowd workers who tag these images participated in several experimental reading and conversation sessions. The culmination is a visual odyssey that invites speculation on novel approaches to image comprehension, advocating for a paradigm shift in computer training methods towards more nuanced, human-centric perspectives.
Part of the methodology applied in the film was discussed in this article:
Pereira, Gabriel, and Bruno Moreschi. 2023. “Living With Images From Large-Scale Data Sets: A Critical Pedagogy For Scaling Down.” Photographies 16 (2): 235–61. https://doi.org/10.1080/17540763.2023.2189285.
Part of the methodology applied in the film was discussed in this article:
Pereira, Gabriel, and Bruno Moreschi. 2023. “Living With Images From Large-Scale Data Sets: A Critical Pedagogy For Scaling Down.” Photographies 16 (2): 235–61. https://doi.org/10.1080/17540763.2023.2189285.
PART 1
I WOULD LIKE TO PURCHASE YOUR EYES TO REST MINE
PART 2
AN IMAGE IS A FACTORY
PART 3
DE-CANONIZATION
Coordinated by Bruno Moreschi.
Directed by Bruno Moreschi and Pedro Gallego.
Edited by Pedro Gallego.
Screenplay by Bruno Moreschi and Caroline Carrion.
Music by Carla Boregas.
Design by Guilherme Falcão.
Sound Treatment by David Menezes.
Translations by Mariana Mendes.
Sending postcards by Pablo Figueiroa.
Research completed during Bruno Moreschi's research fellowship as an early-career fellow at Collegium Helveticum 2023/2024. The Collegium Helveticum is the joint Institute for Advanced Studies of ETH Zurich, the University of Zurich, and the Zurich University of the Arts (ZHdK). Professor Felix Stalder (ZHdK) was the associate fellow of the project.
1,5990,611 images coming from the following computer vision datasets:
Flickr-Faces-HQ, ImageNet, Labeled Faces in the Wild, LAION-400-Million Open Dataset, Leaf Detection, Microsoft COCO 2017, Google Open Images, Pascal VOC 2012, Stanford Dogs, Tencent ML-Images, This Person Does Not Exist, Visual Genome, Wildfire Smoke Dataset.
The film is currently being shown in cinemas, museums and universities in Europe, creating opportunities not only for the exhibition of an artistic work, but for collectively rethinking what Computer Vision is and what it can be.