When we take a picture, we capture a moment in time within a certain frame. Similarly, before the advent of photography, we relied on oil paintings to tell stories through the artists' hands.
What could possibly happen beyond the canvas? Can we use AI to explore the potential past(s) of GLAM (Galleries, Libraries, Archives, and Museums) archives and examine the areas that are missing? What might we discover by focusing on the parts omitted from the paintings? AI predicts the most probable scenarios, but what can we learn from these missing pieces?
One of the methods to explore that is through outpainting. The 'outpainting' in digital imaging is a technique used to extend the boundaries of an image. It involves creating additional content beyond the original edges of an image, effectively expanding the canvas. This method is particularly useful for tasks like extending landscapes, filling in missing areas of panoramic photos, or creating a wider field of view from a limited original image.
Outpainting uses AI algorithms or manual digital artistry to predict or invent what the scenery might look like beyond its original confines. This requires an understanding of the content and context of the existing image to ensure a seamless and realistic extension. The process can be challenging, especially when maintaining the style, lighting, and perspective of the original image.
But why is it predicting other women in the picture where it should just paint more trees? Who are these people that do not exist and what does it tell about us? What could be the untold stories that AI is telling?
<aside> š§ What could be the paintings actually worth exploring from a historian POV? With interesting plots / stories / secret? What would they say about the results? What kind of stories do results say?
</aside>
Kornel SpĆ”nyik ā DievÄa v kroji
