to noise or not to noise
to noise or not to noise
I have been thinking about noise. In the context of large language models, noise can’t be recognized. Fun fact: you can create a diffusion model by destroying data—by adding noise.
I have been exploring this concept freely and playfully. The free falling of feeling things the way we want—is that even possible? Can I choose and pick up a feeling through images? How can I trick the viewer, or myself?
The exercise is fun: it’s like playing poetry with images and screenshots. If I type the word "yesterday" or "today" in my phone, photos related to "yesterday" or "today" will appear. But "not every yesterday should remain" in memory". I create noise through erasure, marking the moments where I hold agency.
The question of what lies beneath the noise—and how humans live amid the latency of those noises—can itself form a pattern. I generated a sample by trying to denoise.
Initial drafts for study - for when words feel heavy, writing is not possible.
