Anime-Otaku-Qwen-Image

Prompt

Anime, a girl with blue hair holding a sword

Prompt

Anime, a samurai standing on a cliff during sunset

Prompt

Anime, a wizard boy reading a floating book in a library

Prompt

Anime, a fantasy castle surrounded by clouds and flying ships

Model description

Anime-Otaku-Qwen-Image

Anime-Otaku-Qwen-Image is a LoRA fine-tuned adapter for the Qwen-Image model, trained on the Anime Gen v2 dataset. It specializes in generating high-quality anime-style images.

---

🚀 Features

• Fine-tuned on xingjianleng/anime_gen_v2 dataset.
• Generates anime characters, scenes, and illustrations in a consistent style.
• LoRA adapter (~rank 16) — lightweight and easy to use with the base Qwen-Image model.
• Compatible with Hugging Face Diffusers pipelines.

---

🖌️ Usage

1. Install dependencies (if not installed):

pip install diffusers transformers accelerate safetensors torch

2. Load base Qwen-Image and LoRA adapter:

from diffusers import DiffusionPipeline
import torch

pipe = DiffusionPipeline.from_pretrained(
    "Qwen/Qwen-Image",
    torch_dtype=torch.bfloat16
).to("cuda")

# Load Anime-Otaku LoRA adapter
pipe.load_lora_weights("suayptalha/Anime-Otaku-Qwen-Image")

# Recommended prompt format
prompt = "Anime, a girl with blue hair holding a sword"

# Generate an anime image (1024x1024 recommended)
image = pipe(prompt, height=1024, width=1024, num_inference_steps=40, guidance_scale=7.5).images[0]

# Save output
image.save("output.png")

Make sure your prompt starts with "Anime, " to match the training data format.

---

⚠️ Notes

• Recommended inference resolution: 1024×1024. You can reduce to 512×512 if GPU memory is limited.
• Use torch_dtype=torch.bfloat16 or torch.float16 for efficient VRAM usage.
• Enable attention slicing to reduce memory usage:

pipe.enable_attention_slicing()

---

Trigger words

You should use Anime to trigger the image generation.

Download model

Download them in the Files & versions tab.