The same Gwain Saga/GeoEXE dataset that was used for the Illustrious version that's now used for this Anima version. This is my first time creating an Anima LoRa through CivitAI's trainer. The images were trained using taglist rather than natural language captions, which makes this LoRa very much experimental. Regardless, the results turned out relatively decent.

Recommended Settings (Main configuration I used for the output images)
LoRa Weight: 0.9 - 1.1
Base Checkpoint: RDBT | Anima [version b1 v0.39.b]
Steps: 25
CFG: 4
Sampler: er_sde
Tags: @gwainsaga, gwainsaga