//=time() ?>
From lexical constraint to syntactic control: CLIP titled this `rooted reflected worries` with one rule to guide it. https://t.co/6vUAepRWkC
This version allows the latents to change so you get something more seamless (like right) -- the next will be pixel-wise (like left).
LatentReVisions adds the ability to edit and re-guide different areas of an image using CLIP and the Taming Transformers VQGAN. Planning on putting an experimental notebook out for patrons soon.
Thinking of this fantastic article: https://t.co/niCB4aYgsX and forcing the title to rhyme "end."
Here is `fatal skull prose visits bend
tuscan painting underthecomprehend` https://t.co/LbR0DU3XQe
Playing a language game where I slowly restrict CLIP's vocabulary as it attempts to describe an art piece, renaming the work in the process.
Here is `surreal atrophora`
From painterly babies to a man on fire -- the ability to create a photo from natural language and a selfie is a *lot* of fun.
Latent3Visions is out on Patreon -- featuring a much more friendly interface (no need to even look at code 😉).
https://t.co/RtkPMl9Mp7
And you can now guide generation with images! Could even turn yourself into a cat person. If you want. I guess.
Hoping to release my text + image notebook tonight to patrons. Really excited to see what people come up with! It's another CLIP + Taming Transformers system.
Here is `Hell is Paradise` + Mona Lisa + Brzezinski