Instructions to use WarmBloodAban/NewChinese_IPfigurine with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use WarmBloodAban/NewChinese_IPfigurine with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("Qwen/Qwen-Image-2512", dtype=torch.bfloat16, device_map="cuda") pipe.load_lora_weights("WarmBloodAban/NewChinese_IPfigurine") prompt = "-" image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- Draw Things
- DiffusionBee
NewChinese_IPfigurine

- Prompt
- -
Model description
NewChinese_IPfigurine: Porcelain & Ink Aesthetics LoRA NewChinese_IPfigurine (瓷影墨韵) is a specialized LoRA model based on the Qwen-Image-2512 architecture. It is designed to define a new visual language for "New Chinese Fantasy" by merging traditional Eastern aesthetics with high-precision modern figurine modeling.
01 Design Concept Unlike traditional freehand brushwork, this model focuses on the intricate textures of physical collectibles. Key features include:
Refined Textures: Delicate skin textures paired with highly realistic silk, brocade, and hair rendering.
Ethereal Lighting: Uses soft diffuse reflection to avoid "industrial plastic" looks, creating a more sophisticated, jade-like or porcelain glow.
Fantasy Symbols: Deeply integrates classical motifs such as cloud patterns (Yunwen), ancient armor, and immortal/samurai elements.
02 Training Methodology: Dual-Fusion Strategy To solve the loss of detail during the "figurine-ization" of traditional styles, we utilized a unique Synthetic Dataset approach. We fused three distinct LoRA effects to generate training data:
Anime-to-Real Characters: For realistic physical material properties.
Anything-to-3D Figure: For iconic figurine proportions.
3D Donghua Style: For the specific "Chinese Animation" aesthetic. This strategy ensures the model captures both the "non-human" stylized proportions of an IP character and the realistic physical details of premium collectibles.
03 Technical Specs Base Model: Qwen-Image-2512
Training Logic: 5-Dimensional Semantic Tagging (Palette, Texture, Structure, Lighting, Subject).
Format: Safetensors
04 Usage Trigger Concept: New Chinese IP figurine style, porcelain texture, intricate silk patterns.
Recommended Weight: 0.8 - 1.0
If you have questions, feedback, or want to collaborate on AI video workflows, feel free to reach out: 商务合作WeChat (微信): aigctyd QQ Group (QQ社群): 1058747239 Email: a592991299@gmail.com
Download model
Download them in the Files & versions tab.
- Downloads last month
- 6
Model tree for WarmBloodAban/NewChinese_IPfigurine
Base model
Qwen/Qwen-Image-2512