This LORA was trained with only 51 images of Henrietta, so yes, it has its flaws. But in my opinion, this one is better than the Seri one, and I still recommend using the Zankuro Style LoCon for better results using the default prompt below.

I trained the LORA with NAI, and created the images using MeinaMixV11.

V2

For better results use the default prompt: 1girl, solo, henrietta (zankuro), skull hair ornament, twintails, red eyes, brown hair, long hair, flat chest, white bikini

You can also change the "white bikini" with "red dress" to get a different "variation", but right now it's very inconsistent for what I've seen.