Very pleased with final LoRA but it was a labor of love fueled by stubborness. Hope you enjoy her as much as I do.It took SIX tries to get this right! Two Ostris AI Toolkit runs with mediocre results not much improved with training parameter changes. Three civitai runs with refinements in data set and tagging/captioning. Switched back to Ostris for final run (hence Omega...this would surely be the last run, right?!?).

Pixxie: Omega Edition- Anima

Step 2750-Quality "early" epoch with some added flexibility relative to the primary version.

Pixxie Omega_000002750.safetensors

Anima-Flexible

2250 Steps

3250 Steps

3750 Steps

Anima Version-Best!

2000 Steps

Pixxie Omega.safetensors

Pixxie Omega_000003000.safetensors

Pixxie Omega_000002500.safetensors

Pixxie Omega_000002250.safetensors

Pixxie Omega_000003750.safetensors

Pixxie Omega_000002000.safetensors

Pixxie Omega_000003500.safetensors

Pixxie Omega_000003250.safetensors

Files with SHA256: 1d7931f4800c81c16aedaa5d59aa8dba6833675f28b119140307669dbf43e047