Things Can Tell

The viewer should photograph the text with the images and upload them to an image-generating AI.

Afterword

This work is a study of image reproduction, as well as the transmission of thought and emotion. I removed parts of various photographic images I had taken and inserted textual descriptions into the deleted areas. In some cases, the text within the image even directly provides instructions for AI image generation. In this work, I hope that viewers will use AI to regenerate the missing parts of the images. This is one of the central themes of the project.

Text itself is an abstract medium of information, while photography, in comparison, is a more mechanical medium. A photograph can be obtained by recording traces of light on photosensitive material or an image sensor. Generally speaking, photographs carry a higher density of information than text. Therefore, when people read text, they often need to rely more on imagination and prior experience in order to construct an image. In other words, viewing a photograph is usually an act of comparison: the viewer compares what they see before them with things they have seen before. Reading text, on the other hand, is closer to an act of creation: the viewer extracts materials from their own experience and memory, and constructs an image or impression in their mind. In this sense, photography is a high-density yet relatively closed medium of information, while text is a low-density yet highly expandable medium of imagination.

Because the work does not directly display the entire content of the photographs, the areas containing text immediately draw the viewer’s attention. Text functions here as a kind of “cheating” device, greatly stimulating the viewer’s curiosity and imagination toward that specific area. Something that may originally have been only an ordinary detail within the photograph acquires special significance after being removed. The absence of the image is not merely a reduction of information; rather, it allows the information of the photograph to extend beyond its original visual content, into the viewer’s memory, imagination, and AI-generated results.In addition to controlling the amount of descript In this process, AI is not simply a generative tool, but a medium that shifts the viewer from “looking at a photograph” to “participating in the reproduction of the photograph.” The redrawing and filling-in of the image by AI becomes a way of connecting the artist and the viewer within the work. The artist describes the missing parts of the photograph through language, while the viewer, having never seen the original visual information of those missing parts, can only rely on the artist’s descriptions to reconstruct them.

This differs from the conventional idea of photographic clarity. I call this viewing condition, determined by the amount of textual information, “imaginable clarity.” It does not refer to the physical resolution of the photograph itself, but to the extent to which the viewer can construct an image, either in their mind or through AI, based on textual description. The more ambiguous the artist’s description is, the more possibilities for imagination the viewer obtains, while the actual resolution of the photograph remains unchanged. When the prompt is limited or ambiguous, the image generated by AI deviates more significantly from the original photograph. Here, “low clarity” does not mean that the photograph itself is blurred; rather, it means that the outline of the deleted area becomes more uncertain within language and imagination.

In addition to controlling the amount of descriptive information, the prompts given by the artist also include repetition, negation, misleading expressions, first-person pronouns, and other words that are atypical of conventional AI prompts. These words participate in the process of image reproduction and become interfering factors. However, this interference is not merely an error, but a way of reducing the mechanical nature of the generated image. Through this method, the artist attempts to reveal the difference between human imagination and AI generation.