Neta Lumina Prompt Book
输入“/”快速插入内容
📖Neta Lumina Prompt Book
Neta Lumina Prompt Book
用户6250用户62502025年7月23日修改
Introduction
Neta Lumina uses LLM (Gemma2) as text encoder, allowing it to interpret a wide range of prompt styles—Danbooru tags, natural language, or any multilingual blend (Chinese, Japanese, English recommended).
This guide will help both new and experienced users write effective prompts to create high-quality images with Neta Lumina.
Showcases
代码块
一幅充满戏剧张力的动漫插画，描绘了一位短发黑发的年轻女性在昏暗环境中点燃香烟的瞬间。角色身着正式西装，打着领带，构图风格带有浓厚的黑色电影（Noir）气质。强烈的光影对比在她的脸部与双手上投下鲜明阴影，使画面更具张力。画面为近景肖像构图，聚焦于她那目光炯炯的眼神与成熟内敛的气质。场景采用情绪化氛围光，搭配打火机火焰带来的温暖高光，营造出低调却富有情绪的环境。角色在点烟时双手动作细腻，显得格外专注，展现出一种私密又沉思的片刻。整体配色以深黑、灰色为主，辅以打火机火光所投下的橘黄色高光，形成冷暖对比。画风专业，结合黑色电影美学与动漫人物设计，背景深沉，烘托出神秘氛围，同时突出面部表情与手势的细节。画风强调写实的光影效果，烟雾与阴影的刻画极为精致。角色设定展现出一种成熟、严肃的个性，暗示其可能为侦探或职场精英角色。构图整体营造出一种静谧的孤独与深思的氛围。
代码块
lineart, greyscale, ,1girl, blonde hair, long hair, gothic lolita, black lace, frills, ribbons, cake, eating, open mouth, elegant, teasing expression, silver fork, gloves, vintage interior, romantic lighting, dark aesthetic, detailed outfit, digital art
代码块
1boy, solo, male, handsome, young man, messy black hair, sharp gray eyes, white background, ornate black military jacket, gothic jacket, silver embroidery, black raven, crow, gloved hands, multiple earrings, silver cross necklace, exposed chest, gothic aesthetic, vampire, cold expression, side profile, dramatic lighting, high detail, black butler, kuroshitsuji
代码块
pixel art, 16 bit,氛围感动漫插画，描绘一位长棕发、红眼睛的年轻女性，慵懒地倚靠在复古棕色皮沙发上。她头上有两只醒目的红色恶魔角，增添了超自然与叛逆气息。画面为中景，展现她的上半身和双腿，身体后仰，表情自信且略带挑逗。她身穿红色字母短款上衣和黑色高腰短裤，配以白色腰带。光线温暖而富有戏剧性，柔和高光与深色阴影突显皮革和皮肤的质感。背景为深色绘画风，点缀微妙金色斑点，营造出私密、情绪化的氛围。画风半写实，笔触表现力强，突出角色的态度与氛围，整体给人酷感自信与神秘感。
代码块
1girl, solo, long hair, black hair, red eyes, geometric pattern, toy gun, pink gun, white gun, "BOOM" text, cherry blossoms, black outfit, strappy outfit, choker, bell, tattoo, cherry blossom tattoo, shoulder tattoo, multicolored background, pink background, blue background, white background, black background, pop art, graffiti, halftone, surprised expression, innocent expression, neon colors, modern anime style, street art, red flower, hair ornament, punk aesthetic, urban fantasy
代码块
Emotional anime-style illustration of two men embracing tightly in a moment of deep vulnerability and comfort. Close-up composition with a shallow depth of field, framed in cinematic widescreen format. The man facing the viewer has dark, slightly wavy hair and teary eyes filled with relief and sorrow, his expression trembling and flushed. He clutches the other man firmly, burying his face in the crook of his shoulder. The other man has curly light brown hair, only partially visible from behind. Both wear dark formal coats, suggesting a serious or solemn context. Floating white paper-like particles fall gently around them, adding to the poignant and atmospheric mood. Lighting is soft and desaturated, emphasizing the emotional intensity and intimacy of the scene. The illustration conveys themes of reunion, forgiveness, or unspoken grief, rendered in delicate digital linework with muted tones and cinematic storytelling sensibility.
One Minute Cheat Sheet
Use what you need: character‑only or environment‑only prompts are fine. You may also check LLM as prompting copilot for a quick experiencing.
代码块
You are an assistant designed to generate anime images based on textual prompts. <Prompt Start>
+ Character
+ Art‑Style
+ Character Appearance
+ Clothing
+ Expression & Action
+ Camera / Positioning
+ Lighting & Effects
+ Scene Atmosphere
+ Quality Tag
+ Natural‑language enrichment
Basics
Fixed Prefix (Recommended)
This prefix can improve image quality. Please add it at the beginning of each prompt
代码块
You are an assistant designed to generate anime images based on textual prompts. <Prompt Start>
1girl, long hair, beautiful detailed eyes, sitting under cherry blossom tree
Alternatives
During training, we experimented with various prompt styles. You can use these alternative prompt types to explore different image generation effects. Please note that these versions have significantly less training data than the standard version, which may lead to inconsistent results.
Danbooru Tags: use this option when your prompt primarily consists of Danbooru-style tags.
代码块
You are an assistant designed to generate anime images with the highest degree of image-text alignment based on danbooru tags. <Prompt Start>
Natural Language: use this option when your prompt is a detailed, natural language description.
代码块
You are an assistant designed to generate high-quality images with the highest degree of image-text alignment based on textual prompts. <Prompt Start>