@The_MrAI that’s the idea behind the phosphene formations. they start out as random lights and gradually take structure through time, and active perception: bringing your associations to the lights.
My phosphene blobs went through an image-to-image model and came back as a valley of cloud. Nine months later I flew into it: a waypoint still every 5 s keeps the video model from drifting. One GPU.
https://t.co/Kvwccf4bIA
#PhospheneDreams#dragon#generativeart#FluxAI#Wan22
@dogramaci Mostly by eye, at the waypoint stills: did the colors stay true to the original, did the detail hold, and does it lead somewhere? Stills are cheap to redo, so that's where I judge. The video just fills in the motion.
Quiet here since 2016. New handle, same person.
Just out of the workshop: nonlinear magnification, the fisheye math from my PhD, running live on the GPU in your browser. Open source, with demos you can poke at.
https://t.co/laexbblAZ4