When Software Sees (But Cameras Don't)
Automatic Pictures from Surrendered Control
July 2025 ‐ by Jan Giliam
What happens when I hijack my smartphone's camera software? The photographs below emerge from these moments of resistance. They were created when I moved my phone in unexpected ways, ignoring on-screen prompts and refusing to cooperate with the camera's intended use. When encountering contradictory inputs it cannot process, the system produces computational noise: pixels arranged in strips and blocks that match surrounding colours but deviate in form.
The philosopher Vilém Flusser taught me to see photography differently. He demonstrated that cameras see nothing at all. They are programmed machines following fixed instructions without awareness. For Flusser, the camera represents a specific type of 'apparatus', a technological system that programmes not just images but human behaviour itself, turning us into functionaries of predetermined possibilities.
Flusser showed that creative photographers attempt to "find the possibilities not yet discovered within" the programme. This is precisely what I do, but deliberately through inducing failure. When processing algorithms encounter data they cannot reconcile, they generate strips of colour, rectangular blocks, unexpected geometries. Not through creative intention, but through breakdown. The question becomes whether I am discovering genuine unprogrammed possibilities or simply accessing error-handling routines that developers anticipated.
André Breton fascinates me because he consciously diverged from conventional practice. As a medical student working under neurologist Joseph Babinski, he ultimately abandoned medicine because his vision of hysteria and automatism clashed with rational clinical approaches. Where colleagues saw pathology, Breton discovered poetry. He understood that apparent failures could yield unexpected insights. This serendipitous discovery process, where valuable insights emerge from apparent chaos, became central to his artistic method. Just as in my drawings and paintings, which I also create through automatic processes, it is serendipity that surprises.
The panoramic image above was made this summer. Motion blur creates ethereal streaks across a wooded landscape whilst stitching algorithms generate impossible patterns in typically straight pine trunks and branches. The system hallucinates because it lacks contextual understanding to recognise its errors. Among dozens of such panoramic images, perhaps a few demonstrate what I consider 'comfortable' noise. The comparison with Surrealist automatism proves both illuminating and problematic. The Surrealists sought access to the unconscious. I do not work with the machine's 'unconscious' but with the surprise of the unexpected. When I describe my method as 'automatism', I use the term metaphorically.
This raises questions about control and surrender. My practice involves deliberately overloading the system until it produces unintended outputs, yet I remain the curator, selecting from hundreds of failed images those few that transcend mere error. Is this automatism or strategic hacking? I work between these approaches, using the machine's limitations to generate visual surprises I could never consciously design. I treat software breakdown and algorithmic confusion as visual poetry. Look closely at these panoramic impossibilities. Within each seemingly realistic photograph lie moments where processing algorithms generate their own geometric logic. Spanning years from my iPhone 4S in 2011 to today's device, I have found unexpected beauty in forcing sophisticated systems beyond their intended parameters.
Click a thumbnail to view.
All images are sized for display, 100% unretouched.
Curated selection: 40 panoramas from 2011-2025