Riccardo Silano
artist designer media archaeologist
My artistic practice investigates generative AI video systems through a media archaeological lens, focusing on the unexplored boundaries where these technologies reveal their unique material properties. Rather than using these tools as intended to create conventional content, I push them to their limits—creating deliberate stress points that expose their underlying visual textures, interpretative tendencies, and distinctive ways of resolving visual contradictions.
This research operates at the intersection of practical experimentation and theoretical inquiry, documenting the specific "visual grain" of different AI video systems while they're still emerging and developing methodologies for manipulating these systems to reveal their hidden capabilities.
"Injection" Techniques and System Manipulation
A significant focus of my research involves developing methods to influence video generation models through initial video input. Similar to code injection in computer security, these techniques leverage the high-bandwidth information channel of video input to "hypnotize" models into producing specific kinds of outputs.
This approach allows for creating sophisticated visual experiences that go beyond conventional usage, revealing emergent capabilities that wouldn't manifest through standard prompting. My current technical stack combines AI-coded visual generators (01 pro) for initial clip creation with Runway Turbo for primary transformation and Sora Remix for final interpretation.
Emergent Storytelling Methodology
I've developed a distinctive approach to narrative that inverts traditional filmmaking methods. Rather than starting with a script that dictates the creation of specific images, I use unstable prompting techniques to generate diverse visuals, then discover potential narratives within this material retrospectively.
This method creates space for unexpected connections and stories that couldn't have been predetermined, allowing the AI systems themselves to contribute to the narrative development process. The approach resembles archaeological excavation—uncovering narratives rather than constructing them.
Visual Textures and System-Specific Signatures
My work has identified how different AI video models develop unique visual signatures when pushed beyond their stable parameters. Each system disintegrates distinctively—Kling shows crosses and diagonals, Pika manifests bubble-like patterns, while Runway exhibits horizontal and vertical grid structures.
These signatures represent a kind of "digital materiality" specific to each system, creating visual timbres that may become increasingly valued as these technologies evolve toward standardization and photorealism. Drawing parallels with Brianino's observation about analog equipment, as technical imperfections become avoidable, they paradoxically become desirable for their expressive possibilities.
This aspect of my research is still a work in progress yet to be fully published, though these distinctive signatures can already be observed in completed projects like my "Evolution/Extinction Games" series, where the specific visual grain of early Kling AI implementations becomes part of the aesthetic language of the work.
My research serves several valuable functions:
This work contributes to understanding the material and expressive qualities of AI video generation at a crucial moment in its development, documenting how these systems construct visual meaning and how their limitations can become sources of distinctive artistic expression.
I'm systematically documenting these techniques and findings in both practical demonstrations and theoretical frameworks. My process includes detailed analysis of how each system responds to different types of manipulation, creating a taxonomy of visual textures and behaviors that offers insights into how these technologies function at their limits.
My most recent developments focus on refining injection techniques that can consistently evoke specific visual responses from different systems, creating a more predictable methodology for producing distinctive aesthetic results while maintaining the sense of exploration and discovery that drives this research.