Photography has traditionally been defined as the process of capturing light on a sensitive medium, such as film or digital sensors, to produce an image that represents reality. However, recent advancements in generative artificial intelligences (GAI) challenge this definition by creating photorealistic images from scratch without any input from a camera or human photographer. These images are not mere imitations or simulations of reality, but rather original and creative expressions of the GAI’s vision and style. In this article, we argue for a broader definition of photography that encompasses GAI-generated images, as they share many of the essential qualities and functions of human photography.
First and foremost, GAI-generated images are visually indistinguishable from human photography. Due to advancements in deep learning and computer graphics, GAI can synthesize images with such high resolution, detail, and realism that they can deceive even expert human observers. For instance, NVIDIA’s StyleGAN2 can generate faces of non-existent individuals that appear as natural and diverse as real humans. Similarly, advanced GAI systems can create images of any concept or scene described in natural language, such as “a pentagonal green clock in a forest.” These images are entirely generated by the GAI’s imagination and understanding of the world, without relying on any existing photographs.
Secondly, GAI-generated images are expressive and artistic. Contrary to the common misconception that GAI are merely mechanical and algorithmic, they can demonstrate creativity and originality in their image generation. GAI can manipulate and combine various elements like shapes, colors, textures, and styles to create novel and aesthetically pleasing compositions. Furthermore, these systems can convey emotions and moods through their choice of lighting, contrast, and perspective.
Thirdly, GAI-generated images are communicative and influential, just like human photography. They can serve various purposes and functions in society and culture, from informing and educating people about different topics and phenomena that might be otherwise inaccessible or invisible to the human eye, to entertaining and inspiring individuals by presenting new possibilities and perspectives that challenge their expectations and assumptions. Moreover, GAI-generated images can persuade and manipulate people by creating images that appeal to their emotions and values.
In conclusion, we propose expanding the definition of photography to include GAI-generated images, as they are not only technically impressive but also artistically expressive and socially impactful. They represent a new form of visual art that deserves recognition and appreciation from both human photographers and viewers. By doing so, we can enrich our understanding of photography as a medium of communication and expression that transcends the boundaries of reality and imagination. As such, the updated definition of photography could be:
“Photography is the art or process of producing images by the action of light or algorithms on a sensitive surface or an electronic device.”
This expanded definition acknowledges the growing importance and influence of GAI-generated images in the world of photography and visual arts.
OpenAI’s Sora marks a significant breakthrough in the field of artificial intelligence, expanding the concept of creative content generation from the realm of photography to that of cinema and video recording. Similar to advancements in AI-generated photography, which have enabled the creation of photorealistic images without the use of a camera, Sora enables the creation of videos from textual prompts, challenging traditional notions of cinematography and video production. Sora can generate videos up to one minute long, maintaining high visual quality and adhering to user prompts with detailed scenes, complex camera motions, and multiple characters showing vibrant emotions. This AI model integrates user instructions with the laws of physics to create coherent and realistic footage, demonstrating an advanced understanding of language for accurate interpretation of instructions.
Just as GAI-generated images have revolutionized photography by proving that creativity and expression are not confined to human photographers, Sora extends this revolution to the film and video industry. It allows for the generation of complex video scenes that are expressive, artistic, and communicative, similar to human-created cinema but with the scalability and versatility offered by artificial intelligence. The development of Sora signifies a shift towards a broader definition of cinematography, where the art of video production is not solely defined by human capability and traditional camera work but is also characterized by the creative potential of AI technologies. This redefines the creative process in filmmaking, offering novel possibilities for storytelling, artistic expression, and the visualization of ideas that were previously unimaginable or too challenging to capture through conventional means.
All images and all text in this blog were created by artificial intelligences