So I’m back on the DALL-E bullshit to gin up cover images for my writing exercises. I got it running on my own hardware, at least for the inference / image generation side of things. Also throwing the results through some imagemagick fuckery. Feeling less squicky about the whole thing, which could be learning or brain damage.
Like, I’m pretty pleased at the janky 90s web and shareware aesthetic I seem to be coaxing out of this thing.
The model’s doing the heavy lifting, but I feel a little more involved after iterating through multiple phrasings of text input and running several image results through imagemagick options before landing on something satisfying. Maybe I’m just filing the serial numbers off stolen clip art, but it feels interesting
@maya heh yeah, could be I understand better or have suffered a concussion from the head desking
@mykie now that’s a name I’ve not heard in a long time. I think my uncle owned a Yugo