All you need is a linux box :) I use kubuntu.com 24.04.3LTS with updates.
it runs locally, and expected run time for 4 AI models and 2 questions (description, keywords) is about 9.5 days for 22K 1080p+ photos. But that's on a box that cost me 4 grand back in 2023. Yeah, just the box. Already had the monitors. I got 64GB RAM and 2TB nvme storage, on this board : https://www.amazon.nl/ASUS-CROSSHAIR-X670E-moederbord-geheug...
yeah, virtual box should work, but here's a tip : make sure all of your scripts (i recommend asking grok.com for scripts, and asking him again to fine tune it) are (A) resumable and (B) save their overall results often (to a real disk).
Image-to-text takes "forever".
In this case, I already let grok make the script that starts this thread, resumable and often-saving.
i'm a patient of psychiatry, due to a persistent sleep deprivation issue leading to occasional confused speech coming from my mouth.
and what i've noticed for myself and for other patients, is that psycho-analysis does indeed work better than anti-psychotics. jeez, the side-effects alone, are horrible.
today i've been busy with improvements to my vividMenu. with the aid of
@grok
over at https://grok.com/c/a8ef9122-7dd1-469f-b7ac-def6297078c2?rid=... i was able to finally tackle most of the problems that had been plagueing it for years already, and which would reoccur with each small change introduced.
AI mutually-assisted coding is the way forward for all coders. Those that lag behind will probably see a lot of their jobs terminated. The time to learn how to do this is now, at http://x.ai or http://grok.com
