Anna Ridler (2020) by Bella Riza
Anna Ridler is an artist and researcher, who is interested in working with collections of information to create new and unusual narratives in a variety of mediums.
Listed by Artnet as one of nine “pioneering artists” exploring AI’s creative potential, her work has been exhibited at institutions such as the V&A Museum, Ars Electronica, HeK Basel, Impakt and the Barbican Centre.
Here, Anna talks through the intersection of new technologies and filmmaking in her latest generative film project, ‘Let Me Dream Again', developed as part of the AMI Grants.
Anna Ridler (2020) by Bella Riza
Tell us about yourself?
I'm an artist and researcher.
I’ve always been interested in examining technology through historical moments, and seeing whether one can draw parallels between those historical moments and how technology is acting now.
"Let Me Dream Again" (2020) by Anna Ridler
What is 'Let Me Dream Again'?
‘Let Me Dream Again’ is a series of experiments that combine generative films made using a model trained on early European-American cinema with parts of the original dataset of cinema footage.
It is estimated that between 80 – 90% of silent films are lost forever. This project explores the parallels between early modern cinema and machine learning, to try to recreate some of these films that no longer exist.
This inverts the way I have previously worked with machine learning and early cinema. Where in other projects I worked with early cinema to experiment with the disintegration of meaning, this project looks at new meanings that can be created from datasets of these early films.
Why early cinema?
In early cinema there is this looseness. There are no formal definitions of what is a 'shot' or how things should be made. That looseness fits quite nicely with generative art. There's not yet a formality to say ‘this is correct’ or ‘this isn't correct.’
"Let Me Dream Again" (2020) by Anna Ridler
Early filmmakers had to invent a film language, much like how artists working with machine learning generated imagery are creating a new form of making work that does have prescribed rules. In both cases, there has been a heavy emphasis on hardware.
Early cinema, for example, was much more concerned about the machine that created it, rather than content. Both were considered niche technologies in their infancies, and both try to record and reflect the world as seen by those in control.
Why 'Let Me Dream Again'?
The title of the project, 'Let Me Dream Again,' references a turn-of-the-century film of the same name, which contains one of the first examples of a dream sequence ever portrayed on film.
Early film theorists such as Ricciotto Canudo (1879–1923) and Jean Epstein (1897–1953) also wrote on the dreamlike quality of film.
Detail: t-SNE visualization of Anna Ridler's 'Let Me Dream Again' (2020) by Anna Ridler
How did you build your dataset?
To start, I assembled a training set of over 3 million still frames.
A generative transformer model will try to replicate the structure of the text that you train it on, so it’s worth paying special attention to what this data is.
Machine learning models need plenty of data to train on to make any decent predictions.
To visualize more than 3 million film stills, Anna Ridler employs a technique known as t-distributed Stochastic Neighbor Embedding (t-SNE). Digital files are organized by attributes, and then projected onto a six-dimensional space, and represented in three dimensions.
For this project, I do not want to try and hide (or disguise) the fact that this has been created using machine learning. Rather, I want to bring it out to explore each material in light of each other and how machine learning has given early cinema the potential for a new aesthetic existence.
Explore Let Me Dream Again by Anna Ridler
Exhibit created by the Artists + Machine Intelligence team
Photography by Bella Riza
Special thanks to Anna Ridler, Holly Grimm, Cyril Diagne, and Parag K. Mital