Tried to catch our candy thief this year by rigging an alarm to the bowl. It didn't work like I hoped 😆.
Video edited in just a few minutes of my own time with @claudeai. Just handed it the porch cam vids and a few photos, @ElevenLabs API for voicing the noir detective, and @ChatGPT image for the case-file images.
@j3rah_ The speed is dependent on the model API, but usually anywhere from 10-30s per photo. I set it up so they queue in the background rather than waiting on each one to finish before you can take another photo. When it finishes you get a little toast you can tap to view.
I built a camera that reimagines every photo with Muse Image 📷
Pick a style on the touchscreen, press the shutter, and the restyled shot appears on screen. 30 styles, from Tiny Clay World to Mall Portrait, 1988.
Raspberry Pi inside, 3D-printed case, public repo 🧵
@costacreates Took a couple prototypes over a few evenings to settle on a design that wasn't too clunky but housed all the pieces I wanted.
The final version takes a few hours to print and probably less than an hour of assembly + software setup.
Want one? Print the case, follow the build guide, then run one install command on the Pi and paste your Meta Model API key.
It runs standalone: no server and no account with me. The key is stored only on the camera.