"The Amazing! The Stupendous! The Inimitable! Alec Azam the Magic Rabbit!"
Alec Azam, the charming little rascal from the Pixar short "Presto".
Trained on Illustrious XL on a dataset of 48 images for 15 epochs, mainly screenshots from the short.
Trigger word: "alec_azam"
Helper tags: "white fur, blue eyes, whiskers, pink nose, rabbit tail, chest tuft, big feet"
After some testing, the model's main issue seems to be that the ears are always perked while in the dataset they were more often than not curved forward and I'm not sure why this is. It may help to prompt "droopy ears" and/or "long ears" and "unaligned ears" to improve chances of at least one of the ears being properly floppy. These options can sometimes give Alec two sets of ears though, and aren't too reliable. I may have to retrain to get the ears to generate right.
I suggest using expressions and posture descriptions, it may have difficulty with the leg anatomy as the design in the short cheats it slightly between shots to be more anthro or less anthro depending on the scene.
This model is not to be used for commercial purposes and generations should never be sold.
