🎨 HUNLUAN

迷幻表现主义与抽象数字艺术结合的特征，核心特点为：

色彩表现：以高饱和度的霓虹色系（红、紫、蓝、绿、黄）为主，色彩碰撞强烈且晕染融合，带有迷幻的渐变与光斑效果，有视觉冲击力。

造型手法：人物轮廓偏向写实基底，但面部、肢体被抽象化的线条、色块和几何纹理解构，叠加了多重复象、扭曲变形的视觉效果（如多眼、面部拆分），兼具超现实主义的荒诞感。

质感与氛围：画面带有数码绘画的肌理感，部分区域有类似故障艺术（Glitch Art）的像素偏移、色彩错位，整体营造出混沌、诡谲又充满赛博朋克式的迷幻氛围，情绪上偏向压抑、躁动的心理具象化表达。

艺术融合：结合了表现主义的情感宣泄、抽象艺术的形式解构，以及数字艺术的技术质感，属于当代先锋数字艺术中 “情绪可视化” 的创作风格。

模型与权重

• 使用通用Anything_v50.safetensors训练，适配通用 checkpoint模型，推荐权重：0.8 – 1.1

• 搭配写实风模型会有神奇效果

触发词指南

全局触发词：

HUNLUAN

搭配推荐提示词效果更好

• 彩瞳： colored sclera, multicolored eyes

• 抽象： abstract

出图尺寸

本模型基于 425 张图训练，原始尺寸为：

• 512×768

推荐直出图尺寸 512×768，避免崩图。

高清放大

推荐：

• 放大算法：Latent

• 重绘幅度 ≤ 0.6

• 建议启用 ControlNet Tile

另可尝试在 Extras 中使用纯放大算法（不带重绘），有时反而保留得更自然。

使用声明

CUC AI STUDIO 出品
欢迎大家使用这款 LoRA 自由创作！
本模型仅供个人学习和非商业创作使用。

基于互联网图片训练模型，如侵权请及时联系微信：VECTOR1914

🎨 HUNLUAN

A distinctive style that blends psychedelic expressionism with abstract digital art, featuring the following core characteristics:

Color Expression

Dominant in high-saturation neon colors (red, purple, blue, green, yellow), with intense color collisions and soft color blending. It incorporates psychedelic gradient and light spot effects, delivering powerful visual impact.

Stylization Approach

Characters are built on a realistic outline foundation, while their faces and limbs are deconstructed with abstract lines, color blocks, and geometric textures. Overlaid with visual effects like multiple image repetitions and distorted transformations (e.g., multi-eyes, facial fragmentation), it exudes the absurdity of surrealism.

Texture and Atmosphere

The artwork has the texture of digital painting, with pixel offset and color misalignment effects similar to Glitch Art in certain areas. Overall, it creates a chaotic, eerie and cyberpunk-style psychedelic atmosphere, serving as a visualized expression of repressed and restless emotions.

Artistic Fusion

Integrating the emotional release of expressionism, the formal deconstruction of abstract art, and the technical texture of digital art, it falls into the "emotion visualization" genre of contemporary avant-garde digital art.

Model & Weight

Attention!!!

Both the training process and sample images are generated based on the base model Anything_v50.safetensors.

• Trained on the general-purpose Anything_v50.safetensors; compatible with common checkpoint models

• Recommended weight range: 0.8 – 1.1

• Pairing it with realistic-style models yields surprising results, and the model has a certain degree of versatility.

Trigger Word Guide

Global Trigger Word:

HUNLUAN

For better results, combine it with the following recommended prompts:

• Colored sclera: colored sclera, multicolored eyes

• Abstract style: abstract

Image Output Size

This model is trained on 425 images with the original resolution:

512×768

Recommended direct output resolution: 512×768 to avoid image distortion.

High-Resolution Upscaling

Recommendations:

• Upscaling algorithm: Latent

• Denoising strength ≤ 0.6

• It is advisable to enable ControlNet Tile

Alternatively, try using pure upscaling algorithms (without redrawing) in the Extras tab—this sometimes preserves the original style more naturally.

Usage Statement

Produced by CUC AI STUDIO

You are welcome to use this LoRA for free creative projects!

This model is intended solely for personal study and non-commercial creative use.

This model is trained on images sourced from the internet. In case of copyright infringement, please contact us promptly via WeChat: VECTOR1914