记录一下SD几个大模型产生图像的参数

本文列举了几个用于AI图像生成的模型配置,包括HongKongDollLikeness结合magmix_v10和chilloutmix_NiPrunedFp32Fix,以及MagMix_s1。这些配置涉及到图像质量、艺术风格、细节处理和抗噪等多个方面,旨在创造高质量的CGI艺术作品。

摘要生成于 C知道 ,由 DeepSeek-R1 满血版支持, 前往体验 >

1、HongKongDollLikeness +magmix_v10

hkgirl, beautiful girl hiking, yoga pants, t-shirt
Negative prompt: illustration, 3d, sepia, painting, cartoons, sketch, (worst quality:2), (low quality:2), (normal quality:2), lowres, bad anatomy, bad hands, normal quality, ((monochrome)), ((grayscale:1.2)), futanari, full-package_futanari, penis_from_girl, newhalf, collapsed eyeshadow, multiple eyebrows, vaginas in breasts,holes on breasts, fleckles, stretched nipples, gigantic penis, nipples on buttocks, analog, analogphoto, signatre, logo,2 faces, soft focus, hands, multiple belly buttons, EasyNegative, paintings, sketches, (worst quality:2), (low quality:2), (normal quality:2), lowres, ((monochrome)), ((grayscale)), skin spots, acnes, skin blemishes, age spot, glans,extra fingers,fewer fingers,strange fingers,bad hand,signature, watermark, username, blurry, bad feet,bad leg, duplicate, extra limb, ugly, disgusting, poorly drawn hands, missing limb, floating limbs, disconnected limbs, malformed hands, blurry,mutated hands and fingers,(extra legs), nsfw, naked
Size: 512x768, Seed: 1433226744, Model: magmix_v10, Steps: 20, Sampler: DPM++ SDE Karras, CFG scale: 5, Clip skip: 2, Model hash: febdce0731, AddNet Enabled: True, AddNet Model 1: HongKongDollLikeness_v15(766394cedb20), AddNet Module 1: LoRA, Face restoration: CodeFormer, AddNet Weight A 1: 0.25, AddNet Weight B 1: 0.25

 

2、 HongKongDollLikeness +chilloutmix_NiPrunedFp32Fix

hkgirl, beautiful girl having dessert in restaurant
Negative prompt: illustration, 3d, sepia, painting, cartoons, sketch, (worst quality:2), (low quality:2), (normal quality:2), lowres, bad anatomy, bad hands, normal quality, ((monochrome)), ((grayscale:1.2)), futanari, full-package_futanari, penis_from_girl, newhalf, collapsed eyeshadow, multiple eyebrows, vaginas in breasts,holes on breasts, fleckles, stretched nipples, gigantic penis, nipples on buttocks, analog, analogphoto, signatre, logo,2 faces, soft focus, hands, multiple belly buttons, EasyNegative, paintings, sketches, (worst quality:2), (low quality:2), (normal quality:2), lowres, ((monochrome)), ((grayscale)), skin spots, acnes, skin blemishes, age spot, glans,extra fingers,fewer fingers,strange fingers,bad hand,signature, watermark, username, blurry, bad feet,bad leg, duplicate, extra limb, ugly, disgusting, poorly drawn hands, missing limb, floating limbs, disconnected limbs, malformed hands, blurry,mutated hands and fingers,(extra legs), nsfw, naked
Size: 512x768, Seed: 780227355, Model: chilloutmix_NiPrunedFp32Fix, Steps: 20, Sampler: DPM++ SDE Karras, CFG scale: 7, Clip skip: 2, Model hash: fc2511737a, AddNet Enabled: True, AddNet Model 1: HongKongDollLikeness_v15(766394cedb20), AddNet Module 1: LoRA, Face restoration: CodeFormer, AddNet Weight A 1: 0.25, AddNet Weight B 1: 0.25

3、MagMix_s1

(MASTERPIECE:1.2),(CGI ART:1.3),(REALISTIC:1.5),(POST PROCESSING:1.3),(SHARP FOCUS:1.3),8K,1 girl, ((Portrait)), ((turtleneck)), (((shiny skin))),garden, (high shadows detail) , (ILLUSTRATION:1.2), (ulzzang-6500:1.0)
Negative prompt: paintings, sketches,(worst quality:2), (low quality:2), (normal quality:2), lowres, normal quality, ((monochrome)), ((grayscale)), skin spots, acnes, skin blemishes, age spot, double navel, mutad arms, hused arms
Steps: 25, Sampler: DPM++ SDE Karras, CFG scale: 8 

### 大规模机器学习模型的主要类型和分类 大规模机器学习模型可以根据不同的标准进行分类,主要包括按架构、用途和技术特点等方面。 #### 按照架构划分 - **浅层模型**:尽管不是严格意义上的“大模型”,但在某些特定场景下仍然具有广泛应用价值。这类模型通常指那些结构相对简单、层数较少的人工神经网络或其他类型的统计学习器[^3]。 - **深层模型(Deep Learning Models)**:这是当前最为流行的大规模机器学习模型类别,其核心在于多层感知机的设计理念被扩展到了更深更复杂的层次。典型代表有卷积神经网络(CNNs),循环神经网络(RNNs)及其变体LSTM/GRU等,在图像识别、语音处理等领域取得了巨大成功[^1]。 #### 根据用途区分 - **通用型大模型**:旨在解决广泛的任务需求而设计出来的预训练模型,比如BERT (Bidirectional Encoder Representations from Transformers) 和 GPT系列(Generative Pre-trained Transformer),它们能够适应多种下游任务如文本生成、问答系统构建等,并且只需少量微调即可达到良好效果[^2]。 - **专用型大模型**:针对某一具体应用场景定制开发的大型AI解决方案,例如用于医疗影像诊断的支持向量机(SVM)-CNN混合框架;或是专为推荐系统打造的基于矩阵分解加深度强化学习机制的复合型算法体系等等。 #### 技术特性角度 - **自监督学习模型**:该类模型能够在未标注的数据集上自动发现潜在规律并完成特征提取工作,进而辅助其他任务的学习过程。这不仅降低了人工标记成本,还提高了系统的泛化能力和鲁棒性。 - **迁移学习模型**:利用已有的源域知识来帮助目标域的新任务快速收敛至较优解空间内的一种有效策略。对于资源有限的小样本情况特别有用处,因为可以直接借用成熟的大规模预训练成果来进行针对性调整优化。 ```python # 示例代码展示如何加载一个预训练好的BERT模型 from transformers import BertTokenizer, BertModel tokenizer = BertTokenizer.from_pretrained('bert-base-uncased') model = BertModel.from_pretrained('bert-base-uncased') text = "Replace me by any text you'd like." encoded_input = tokenizer(text, return_tensors='pt') output = model(**encoded_input) ```
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包

打赏作者

caridle

你的鼓励将是我创作的最大动力

¥1 ¥2 ¥4 ¥6 ¥10 ¥20
扫码支付:¥1
获取中
扫码支付

您的余额不足,请更换扫码支付或充值

打赏作者

实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值