Sign In

Ruby (RWBY)

111
1.2k
8.5k
20
Updated: Oct 6, 2024
characteranime3dfemalerwbywoman
Verified:
SafeTensor
Type
LoRA
Stats
708
7,524
Reviews
Published
Sep 16, 2023
Base Model
SD 1.5
Trigger Words
Ruby_RWBY
Hash
AutoV2
E4B69CA65B

The goal of this Lora is to be able to make Ruby in her original 3D art style.


Please checkout my other RWBY Loras: https://civitai.com/collections/115875


I experimented with different training techniques and posted a my research in this article https://civitai.com/articles/2234


There are 4 versions of this Lora

Two versions trained on AnyLora Checkpoint and two on Anime Checkpoint

Accurate

  1. AnyLora_MoreTagsAbsorbed

  2. AnyLora_LessTagsAbsorbed

Flexible

  1. Anime_MoreTagsAbsorbed

  2. Anime_LessTagsAbsorbed

*These require more prompts to be accurate: that's the point

MoreTags Absorbed these tags: 1girl, short hair, black hair, solo, cape, grey eyes, red cape, corset, thighhighs, torn clothes, torn thighhighs, skirt, zettai ryouiki, dress, boots, parody, scarf, rose, torn cape, black thighhighs, anime coloring, flower, black footwear, gloves, belt, long sleeves, hood, crescent rose, hair over one eye,

LessTags Absorbed these tags: 1girl, short hair, black hair, solo, grey eyes

Given that my focus was creating the 3D art style her outfit is not very flexible but the LessTagsAbsorbed versions do have some flexibility.

Here are examples with the default outfit (no extra prompts)

As you can see AnyLora is better at capturing the original 3D art style.

And Examples where the prompt is denim shorts, white crop top,