Deep Dive Original

Deep Dive (2016) offers an endless zoom into a digital picture, revealing an infinitely detailed imaginary world that emerges from the original pixels. Instead of the blurring effect that normally results from zooming in and interpolating pixels, Deep Dive generates new features continuously. This occurs through the image refining itself according to information that is extracted from the image concerned. The data is collected through analysing the fundamental image construction on several scale levels, from coarse to fine. New image fragments are subsequently assembled during the synthesis on the basis of their position between other fragments. The system learns as it were how an image can ‘grow’, how it can arise from smaller components. The viewer can steer this process by following an interesting area with the mouse, and thus further developing and exploring it in detail. Initially on start up, you see the original image, unaltered and clearly figurative. Everything gradually becomes more and more abstract during the zooming, while some characteristics are preserved from the original, such as coherent forms, combinations of colour tones and textures. Each input image has unique image features and thus also a specific output ‘style’. At the bottom left of the screen the user sees a small camera icon. By clicking on it, he/she may download a new, randomly selected picture from the online WikiMedia archive, and then 'dive' into it.