pip install -U diffusers transformers accelerate
import torch
from diffusers import DiffusionPipeline
# switch to "mps" for apple devices
pipe = DiffusionPipeline.from_pretrained("Silvelter/Sathariel", dtype=torch.bfloat16, device_map="cuda")
prompt = "masterpiece, best quality, 1girl, fox ears, looking at viewer"
image = pipe(prompt).images[0]Sathariel
Sathariel is a latent diffusion model tailored for anime enthusiasts. Similar to other anime-styled Stable Diffusion models, it supports danbooru tags to generate images.
To develop this model, I conducted two finetunes based on the merge of MidnightMixerV2 and MidnightMapleV1, then blended these finetunes together.
At present, the repository contains several versions of the model. These include SatharielV1 as well as alternative versions like SatharielV1A, SatharielV1B, or SatharielV1C. The differences among them are minimal, hence they were released as alternatives.
SatharielV1C_DX is a merge of SatharielV1C with a model by DutchaAlex, offering enhanced color rendering.
Although originally intended to emulate the style of artist Mito Nagishiro, the attempt fell short. For a better outcome, I recommend using NAIV3 instead, accompanied by the "by nagishiro mito" tag.
The "other" directory houses older mixes, models, experiments, etc.
Big Thanks to
- Downloads last month
- -
# Gated model: Login with a HF token with gated access permission hf auth login