I was going through my old LoRAs and decided to upload some that I didn't before. First on that list is Puniru from Puniru wa Kawaii Slime. I've not seen the show, but the character is cute, so I made it.

Why doesn't my output look like yours?

1. Check the negatives in my examples. Do you have those embeds?

2. I used !ADetailer because it's just a set and forget quality bump.

3. The images were genned at 512x768 then upscaled 1.5 times using 8x_NMKD-Superscale_150000_G with denoising 0.45. You can find it with a quick Google. Note that this upscaler is non-deterministic from my testing so your images may vary slightly. (Check the bottom of this post for info on the cover image.)

Settings and Training:

Trained on Anything v3
Genned on Anything v4.5
LoRA weight 0.8

What's up with the first image?

Since the upscaler is non-deterministic I could just keep trying until both of the hands were correct or I could do what actually happened and get one image with a good right hand and one with a good left hand. Since the upscaler is mostly consistent you can just layer them then erase a hand to get the proper result.

The prompt for the first image may not come through due to this so here is the positive and seed. The negative was the same in every image.

masterpiece, best_quality, 1girl, solo, puniru \(puniru wa kawaii slime\), puniru wa kawaii slime, gradient hair, bangs, green eyes, double bun, mini hat, long hair, jacket, skirt, simple background,

Seed: 330420430

Where have you been? Are you making LoRAs again?

Busy. Just finished my second degree, so I have time now while I send out job apps. I might make more LoRAs, but at the same time I feel this strange aversion when I look at the backlog from comments. Time will tell I guess. For now though I will be uploading 4-5 style LoRAs and a clothing LoRA. Like this character all are pretty old, but not bad quality wise.