Trained with v1-5-pruned-emaonly. Checkpoint for sample images was consistentFactor_v3Light using ControlNet for posing.