The Upscaling of the Early Cinema Image through Artificial Intelligence. A New Aesthetics between Continuity and Dissent in Image Theory

Abstract

An active community of users is manipulating the image of early cinema using Machine Learning-based software. This practice, encompassed in the field of Artificial Intelligence, has transformed the materiality of these images, generating a mutated image as a result. Through a qualitative research design, based on a hermeneutic approach, this research aims to delimit the nature of this new image by observing it from the theoretical parameters that have shaped the specificity of the original filmic image. Starting from three historical concerns of film image theory, we have studied the interferences that this mutated image produces in the field of aesthetics, in its signifying relationship with reality and in its own ontology. Surprisingly, the new mutation presents a meagre break with the theoretical tradition of the cinematic image, oscillating between a slight dissidence and a marked continuity of its postulates.