About Shadowplay
About Shadowplay
Imagine a film where every single key frame has its own unique soundtrack, a symphony of sound perfectly synchronized with the visual’s narrative, with the film’s cinematography. This is the concept behind Shadowplay, a visual research experiment that plays with the boundaries of film and artificial intelligence.
Shadowplay takes Mircea Sauceanu’s 1967 Romanian drama Meandre as its starting point. Director-researcher Georgia Mihalcea embarked on a journey to explore the question: What if each key image of this classic film had its own score?
The result is a mesmerizing experimental song that expands Meandre‘s narrative and creates the potential of this film’s rediscovery, but also a new horizon for the cultural patrimony development and rejuvenation. Georgia used cutting-edge AI applications like Gemini Advanced and Suno to analyze the cinematography of Meandre and extract metadata related to lighting, composition, feel, theme, and movement. This data was then used to generate a unique musical score and evocative lyrics that capture the essence of the image, while at the same preserving the connection with the overall story.
Shadowplay is not just a technical experiment; it’s an artistic exploration of the interplay between sound and image. The song delves into the emotional depths of Meandre, uncovering hidden layers of meaning and offering a fresh perspective on this cinematic masterpiece.
But what if we maintain the same music, but generate lyrics in more languages? It would result a multicultural playlist resulted only from one image. What if we use more from the ‘escavated’ metadata to create more songs for the same image? It would result a different kind of playlist and story extension. What if we use these experimental songs as references for a music composer and create real songs? In all cases, it would help the film’s rediscovery.
This project also raises intriguing questions about the future of film and the role of AI in creative expression. Can AI help us rediscover classic films and expand their narratives? Can it expand our understanding of cultural patrimony? Shadowplay offers a tantalizing glimpse into the possibilities.
the unseen experiment behind this post
In this experiment, I aimed to make the information about the Shadowplay visual research project more machine-readable. I achieved this by translating the key details into structured data using the JSON-LD format. This format allows machines, like search engines, to easily understand and interpret the information of this blog post, so its audiences could discover it through search or recommendation. As I don’t have a deep computational background, AI is a valuable assistant in my research journey.
Here’s a breakdown of the process:
- Identifying Key Entities: I first identified the core entities involved, such as the music video itself, the original film it’s based on, the director, and related works.
- Mapping to Schema.org: I then mapped these entities to relevant types defined in Schema.org, a vocabulary for structured data on the web. For example, I used
MusicVideoObjectfor the video,Moviefor the films, andPersonfor the director. - Adding Properties: I added specific properties to each entity to provide more detailed information. This included properties like
name,datePublished,genre,director, and equivalents forisBasedOn, andisSimilarTo. - Defining Relationships: I established relationships between entities using equivalent properties like
isBasedOnandisSimilarToto connect the music video to the original film, similar works, and some of my short films (to help also my work). - Embedding in JSON-LD: Finally, I should have embedded this structured data in the JSON-LD format within the
<head>section of the blog post, but because I don’t know yet how to access the <head> section, I embedded it under the post title with the help of a HTML box where I could copy-paste the code. This allows search engines and other machines to easily access and interpret the information. Before, I tested and validated my schema.
By translating this information into structured data, I’m theoretically increasing its chances to become more accessible and discoverable online. This may help people find the Shadowplay project through various searches, leading to a wider audience and greater engagement, leading them further to Meandre, Nights of Cabiria, and my shorts. Time will tell.
This experiment highlights the potential of creating a richer, more interconnected web of knowledge for our creative works to organically enhance their discoverability.
