My first attempt at making a Lora, done through the aid of this wonderful guide: <a target="_blank" rel="ugc" href="https://civarchive.com/models/22530/guide-make-your-own-loras-easy-and-free">https://civarchive.com/models/22530/guide-make-your-own-loras-easy-and-free</a>As a massive fan of both the Nanoha series and the Numbers, I've been wanting to do something with their uniform for the longest time.I have also made use of this Lyco: <a target="_blank" rel="ugc" href="https://civarchive.com/models/32108/numbers-uniform-magical-girl-lyrical-nanoha">https://civarchive.com/models/32108/numbers-uniform-magical-girl-lyrical-nanoha</a>I however found that it didn't always produce the results I was looking for. Thus I decided to see if I could make something myself. It's by no means perfect, but I have found myself able to more consistently get the look I'm trying to go for with this.Recommended parameters:Weight: I myself adhere to somewhere between 0.9 and 1.1. All depending on other prompts, lora's and models.Base trigger: numbersuitAdditional:yellow eyes, glowing eyes, intricate iris, eyeball, eye focus - For the most iconic Number look, feel free to experiment with leaving out everything aside from 'yellow eyes', but I found what I list to work best for me.two-tone bodysuit:0.9, blue bodysuit:0.9, light blue bodysuit:0.9 - It can be a bit fickle with coloration, on bare bones prompts I find that this isn't necessary, bit helps when going a bit further down into use of other lora's.No specific advise on model or VAE as I've found it to work decently with most of what I've tested, I can only advise experimenting. Though I do generally stick to Clip Skip 2.I generally also recommend taking a gaze at the added images for further prompts I have used to get the results shown.Training data:Trained with AnyLora on about 71 images.

Files with SHA256: 5c25215a3530d12f64ad7419031a7d69dbda46e267cb52c8e4a98c097177659a