Haley D.va is a merged LoRA created from two persons; D.va (by user sololo_xyz) and Haley (by user blueadeptjux … BTW nice video.). Both LoRAs are robust and give consistently good results. I have a small corral of reliable LoRAs and started to see if I could create robust variants by merging them. Some worked out and some did not. This is one that worked pretty well. As you will see from the examples (there's metadata ... it's not showing), this LoRA works pretty well with several trained models and plays nicely when stacked with other LoRAs. I think in this case it did inherit the ability to yield consistently good results.
Use weight 1. Produces both NSFW and SFW results.