This LoRA is trained on the Iron Bull, a Qunari mercenary from Bioware video game Dragon Age: Inquisition. The dataset is 57 pics, including in-game screenshots, offical concept arts, fanarts, and pictures of fan-made 3d model.
Base model is AbyssOrangeMix2-nsfw, but should work on other model too, especially anime-focused ones. I tested it on AOM2-7thV3_C_40, Snovelpc, Anything V3, and SD 1.5. It works well with style LoRA too.
The trigger word is "iron bull". For a look closer to the character, I recommend adding "iron bull, horns, grey skin, eyepatch" into the prompt.
Note that because I didn't use a unique token when training, when using this LoRA with general model like SD 1.5, you might want to add "cattle, ox, cow" into negative prompt, in order to avoid generating a literal bull.
Join us in the #men-only channel on Unstable Diffusion Discord server, for lively discussion of how to make good men with Stable Diffusion. You can find my style LoRAs, along with other men-focused LoRAs made by other amazing people, in the #LoRa thread of this channel.
Description
This version is trained on 57 images, and the base model is AbyssOrangeMix2-nsfw. It's not perfect, and I might train another version in the future. Other technical details are as follows:
dataset repeat: 2
base resolution: 512x512
epoch: 10
batch size: 4
network dimension: 128
learning scheduler: cosine_with_restarts (10 cycles)
unet learning rate: 1e-4
text encoder learning rate: 5e-5
FAQ
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.

