Cheetara (Thundercats) (iA3 Prodigy Experiment)

guide here: https://rentry.co/ProdiAgy (still figuring out optimal settings)

this is only to serve as an example that yes 200kb loras can give good representations, it was done on 15 batch size * 2 gradient accumulation and it uses noise_offset just to test it out therefore versatility may not be very good due to the small dataset and high batch size + the use of noise_offset

its only 210kb and was made in about 15 min with 30 images, now imagine if everyone switched to this bad boy

all previews were made casually, no lbw, strength at 1, not much thinking done in prompting.

the results speak for themselves, at only 200kb