The versions in this series are NOT progressive upgrades, each version is a different theme. You can view it as totally different models. 这个系列的版本并不是递进式升, 每个小版本都是独自的一种风格。你可以把每个版本当做是完全不同的模型。版本之间没有任何对比性。

2026-09-26

📸 Moody Amateur V5 — 轻熟随拍

Prosumer Shots

今天给大家带来的是 Moody Amateur 系列的第五个主题 ——「轻熟随拍」。

这个系列想做的，其实是一种很特别的感觉：

用最好的器材，去拍最“业余”的照片。

不是棚拍，不是精心设计好的商业摄影，也不是每一张都要有完美的构图。

而是那些生活里随手记录下来的瞬间——
街拍、车展、漫展、活动现场、朋友之间的随手拍……

就像一个喜欢摄影的业余玩家，拿着一套不错的专业器材，到处拍自己觉得有意思的人和画面。

这次的训练素材来源也比较丰富。

其中一部分来自一位不愿透露姓名的粉丝授权素材，一部分来自身边朋友的分享，一部分来自我自己这些年积累下来的存货，包括很多生活中比较琐碎、自然的照片；另外还有一部分来自公开可见的中国、韩国车展及活动摄影素材。

涉及真人的素材，大部分都经过了去身份化处理，最终模型学习的重点仍然是整体摄影风格、人物气质、光线和画面感觉，而不是还原某一个具体的人。

这次和以前 Moody 比较常见的“嫩妹感”也不太一样。

V5 的人物整体会更加轻熟、成熟一点。

不是刻意做成很强的熟女风，而是那种已经褪去一点少女感，更接近现实生活里二十多岁到三十岁女性的气质。

所以对于某些人来说……

嗯，我感觉这版应该会非常对胃口。😏

虽然好像每次发模型我都会说这句话，但还是想再说一次：

每一个模型对于没有团队的我来说，真的都花了我非常非常多的时间。

从最开始构思这一期到底要做什么主题，到慢慢收集素材、整理素材、取得授权、筛选训练集、反复测试，再到最后一点一点融合进整个模型里，这里面其实有非常多看不到的工作。

我平时毕竟还有自己的主职工作。

做模型从头到尾一直都只是因为我自己喜欢，也是一个占用了我每天非常大一部分私人时间的爱好。

所以我也真心希望大家能够尊重每一位创作者在作品背后的付出。

融合模型真的并不是简单地拿几个别人发布好的 LoRA 混在一起，就能够做出来的东西。

素材、方向、审美、训练、测试、取舍和最终融合，每一个环节其实都需要花非常多时间。

这一版从最开始准备到现在，前前后后差不多也用了 一个月。

希望最后做出来的东西，也依然能够让你们觉得：

“嗯，这还是那个熟悉的 Moody。”

谢谢大家一直以来的支持。❤️

希望你们会喜欢这次的——

Moody Amateur V5 · 轻熟随拍。

📸 Moody Amateur V5 — Prosumer Shots

Today I'm bringing you the fifth theme in the Moody Amateur Series — Prosumer Shots.

The idea behind this series is pretty simple:

Using the best gear to take the most amateur-looking photos possible.

Not studio photography.
Not carefully planned commercial shoots.
Not perfectly composed images every single time.

Instead, it's about those casual moments from everyday life — street photography, auto shows, conventions, events, and random snapshots taken just because something caught your eye.

Think of it as a photography hobbyist carrying professional-level gear around and simply shooting whatever feels interesting.

The dataset for this release comes from several different sources.

Some images were generously authorized by supporters who prefer to remain anonymous, some came from my own personal archive of everyday photography, and another portion was collected from publicly available Chinese and Korean auto show and event photography.

For images involving real people, most of the material has been de-identified. The goal is to learn the overall photographic style, lighting, atmosphere, and character presentation — not to reproduce any specific individual.

This release is also a little different from the younger, cuter look that you may normally associate with Moody.

V5 leans noticeably more mature.

Not aggressively “mature,” but more of a light-mature, grown-up feeling — women who have moved slightly beyond the typical youthful-girl aesthetic and feel closer to people you might actually encounter in everyday life.

So for some of you...

Yeah. I have a feeling this one is going to hit exactly the right spot. 😏

And although I feel like I say this with every release, it's still worth saying:

Every model takes an enormous amount of time to create especially for a one-man-army.

Coming up with the concept, finding and organizing material, obtaining permission, curating the dataset, testing, adjusting, and finally integrating everything into the finished model — there is a huge amount of work behind the scenes.

I still have a full-time job outside of this.

Making models has always been something I do because I genuinely enjoy it, but at this point it also takes up a very large portion of my personal time every day.

So I sincerely hope people can respect the work that creators put into their models.

Model merging isn't simply downloading a few public LoRAs and mixing them together.

The dataset, direction, aesthetics, training, testing, selection, balancing, and final integration all take time.

This release alone has been in preparation for almost a month.

Hopefully, after all of that work, you can still look at the result and think:

“Yep. This still feels like Moody.”

Thank you again for all your continued support. ❤️

I hope you enjoy:

Moody Amateur V5 — Prosumer Shots.

Added Support for INT4, new converters for all variants, plz report problems. 新增INT4版本，用了全新的转换器，如果有任何模型遇到问题请告知。 Do not steal my models, I WILL BE WATCHING!

2026-08-29

🌿 素人第四版-归璞存真 — Vol4. RAW STUDIO SHOT STYLE 📸

🇨🇳 中文

大家好！👋

这是这个模型的第 4 个系列，这一次，我想带给大家更加极致的 RAW Studio Photography 风格。📸✨

这一版本重新平衡了我其他的 photorealistic LoRA，在保留 RAW 摄影真实感的同时，为人物加入了一点点轻微、自然的美感增强。不会过度磨皮，也不会让人物看起来像塑料或 CG，希望能在“真实”与“好看”之间找到一个舒服的平衡。 ❤️

这一次的训练集主要来自大量 RAW 摄影素材，整体偏向艺术人体摄影风格。模特们拥有非常优秀的身材比例，同时模型也学习到了大量全新的摄影姿势、构图方式和人物表现。🧍‍♀️📷

当然，训练集中的所有素材都进行了去身份化处理，包括纹身、脸部特征等可能用于识别身份的信息，尽可能确保模型不会学习到具体人物的身份特征。🔒

另外，这一版本也受到一位朋友的启发。💡
我尝试在脸部保留一些适度的美颜和视觉优化，同时尽可能保留身体皮肤真实的纹理、细节和材质感。

希望我没有把美颜开得太过头 😂
也希望大家会喜欢这个版本！ ❤️

感谢一直以来的支持！🙏
Enjoy the RAW. Enjoy the realism. 📸✨

🇺🇸 English

Hello everyone! 👋

This is the 4th series of this model, bringing you what I hope is the ultimate RAW Studio Photography style. 📸✨

For this version, I rebalanced it with my other photorealistic LoRA to maintain a light and subtle touch of beauty enhancement on the human subjects, while keeping the overall RAW photography feeling intact. The goal was to find a comfortable balance between realism and beauty — without excessive skin smoothing or making the subjects look artificial or CGI-like. ❤️

This version was trained on a large collection of RAW photography, with a strong focus on artistic figure photography. The subjects feature excellent body proportions, and the model has also learned many new poses, compositions, and ways of presenting the human figure. 🧍‍♀️📷

Of course, the training materials were carefully de-identified, including tattoos, facial characteristics, and other potentially identifying features, to prevent the model from learning specific identities. 🔒

This version was also inspired by a friend of mine. 💡
I experimented with applying a moderate amount of facial beauty enhancement, while intentionally preserving the natural texture, details, and material quality of the skin on the body.

Hopefully I didn't push the beauty enhancement too far. 😂
I really hope you enjoy this version! ❤️

Thank you, as always, for your continued support! 🙏
Enjoy the RAW. Enjoy the realism. 📸✨

2026-08-29

🌟 Amateur Series V3 — 素人系列第三版

🇨🇳 中文

首先，感谢一直以来支持我的每一个人 ❤️
你们的支持真的比你们想象中更加重要。

当然，这段时间发生的很多事情，也让我看清了不少丑陋的嘴脸。😮‍💨
但不管怎么样，为了那些一直支持我的人，我还是会继续创作，继续给大家带来更多有趣的作品。

同时，今后我也会定期制作一些支持者专属模型，作为对一直支持我的人的回馈。🎁

今天终于带来了 Amateur Series V3（素人系列第三版）！🔥

其实这个版本已经完成一段时间了，只是因为一些大家都知道的事情，一直拖到了现在。

期间我也纠结了很久，到底还要不要公开发布。毕竟现在有太多模型贼、不会感激的用户，以及一些典型的“升米恩，斗米仇”的人。🙂

不过最后我还是决定发布。

我也想借这个机会重新说明一下 Amateur Series 这个系列的设计思路：

这个系列的版本并不是递进式升级，并不是 V3 就一定比 V2 “更好”。
每一个版本都会有自己独立的主题和摄影风格。📸

• V1 — 千禧年代风格，约 1995–2005 年左右的摄影美学

• V2 — 约 2015 年前后的摄影风格

• V3 — 老司机们应该会非常熟悉的经典素人 JAV 风格 🇯🇵

所以每一版更像是不同的“口味”，而不是单纯的版本升级。

其实做这个系列真的非常不容易。😵‍💫
每次都需要准备大量的图片和图集，而这次最大的难点，就是在尽可能保留原本风格的同时，处理掉马赛克、封面字体以及人脸处理等问题。

这次也花了不少时间和精力，希望大家会喜欢！❤️

Enjoy Amateur Series V3! 🔥📸

🇺🇸 English

First of all, thank you to everyone who has supported me through everything. ❤️
Your support means more to me than you probably realize.

At the same time, everything that has happened recently has also made me see quite clearly some of the uglier sides of this community. 😮‍💨

But regardless of all that, I’ll keep creating for the people who genuinely support my work. I still want to bring you more interesting and fun creations.

I’ll also be making supporter-exclusive models from time to time as a way of giving something back to those who have continued to support me. 🎁

Today, I’m finally bringing you Amateur Series V3! 🔥

This version has actually been finished for quite some time, but due to some things that I think many of you already know about, its release kept getting delayed.

I also spent quite a while debating whether I should release it publicly at all.

With so many model thieves, ungrateful users, and people who seem to believe in the whole “give them an inch and they’ll take a mile” mentality, sometimes you really start wondering whether sharing your work publicly is worth it. 🙂

In the end, I decided to release it anyway.

I also want to take this opportunity to clarify the idea behind the Amateur Series once again.

The versions in this series are NOT progressive upgrades. V3 is not simply supposed to be “better” than V2.

Each version has its own unique theme and photographic aesthetic. 📸

• V1 — Y2K / Millennium-era photography, roughly inspired by 1995–2005

• V2 — Photography styles and aesthetics around 2015

• V3 — A classic Japanese amateur JAV aesthetic that longtime enthusiasts should be very familiar with 🇯🇵

So think of each version as a different flavor rather than a straightforward upgrade.

Honestly, making this series is not easy. 😵‍💫
Every release requires a huge amount of image collecting and dataset preparation.

For V3 in particular, the biggest challenge was preserving the original aesthetic while removing things such as censorship mosaics, cover typography, and unwanted facial processing.

A lot of time and effort went into this one.

I hope you enjoy it! ❤️🔥📸

Enjoy Amateur Series V3!

2026-08-25

Moody Amateur Mix V2 — Release Notes

🇨🇳 中文

🎉 Moody Amateur Mix V2 来了！

✨ 本次更新：

• 📸 提升整体画质表现

• 🎞️ 强化 Amateur Photography / 随手拍 风格，更自然、更有生活感

• 🔥 使用最新的 Moody Style & Anatomy LoRAs

• 👤 降低默认内置角色的脸部权重，解锁更多不同的脸型与人物类型

希望 V2 能让你的照片看起来更自然、更像真实生活中的随手拍 📷❤️

🇬🇧 English

🎉 Introducing Moody Amateur Mix V2!

✨ Quick Notes:

• 📸 Improved overall image quality

• 🎞️ Enhanced the amateur photography / casual photoshoot aesthetic for a more natural, candid feel

• 🔥 Uses the latest and greatest Moody Style & Anatomy LoRAs

• 👤 Softened the default built-in character face weight, unlocking a much wider variety of faces and character types

V2 is all about making your generations feel more natural, spontaneous, and like real-life amateur photography. 📷❤️

2026-08-15

🇬🇧 ENGLISH

🎞️ Moody Amateur Mix — New Model Release

Hello! After going back and forth for a while, I finally decided to make a new model called Moody Amateur Mix. I’ve actually been preparing this one for quite a while.

I wanted to create something more realistic, raw, and nostalgic — something that doesn’t look heavily photoshopped or overly polished. Something that feels like going back to the golden era of the 2010s. 📸✨

Unlike my other models, where a large portion of the training dataset was AI-generated due to ownership and legal concerns, this model was created with a big portion of images from photos taken with my iPhone 4S and iPhone 3GS (yes, they still turn on — good ol’ Apple 😂🍎), along with some old photos of real people.

You might even be able to generate me from 16 years ago, since I was clearly in a few of the training set images. 😏 So... let’s play Where’s Waldo? 😂

🔓 This model is uncensored, just like my other models, and I hope it brings back some good memories.

⚠️ Known Issues

• It is primarily trained for people photography, so it’s not particularly good at generating complex scenery or environments.

• The liquid/wet skin issue is still there, but it’s MUCH better thanks to @Loraholic . He was kind enough to provide me with some private LoRAs that helped significantly with this issue. Many thanks to him! ❤️

💬 If you like the model, please leave me some comments or suggestions. If there’s enough interest and likes, I’ll consider continuing to update it.

This model took me a very long time to prepare and caption. A huge portion of the process was done manually and involved a lot of repetitive work. Please understand that I’m a ONE MAN army — unlike many creators who may have more resources or hardware, I’m still just one guy working with my good old 3080 Ti. 😂

Anyway, good luck and have fun generating! ❤️📸

Some helpful prompts:
A casual amateur snapshot taken with a cheap early digital camera using on-camera flash, in the style of early 2000s internet photography, strong frontal light creating harsh shadows and overexposed skin highlights.

Make sure you describe the race (Asian|Caucasian|Latina|Black).

━━━━━━━━━━━━━━━━━━

🇨🇳 中文

🎞️ Moody Amateur Mix — 全新模型发布

大家好！经过一番犹豫不决，我最终还是决定制作一个全新的模型，叫做 Moody Amateur Mix。其实这个模型我已经准备了相当长一段时间。

我一直想做一个更加真实、原始、自然，同时又带有怀旧感的模型——不追求过度 Photoshop，也不追求那种过分精致、商业化的照片质感，而是希望找回一些属于2010 年代黄金时期的感觉。📸✨

和我之前的其他模型不同，因为版权、所有权以及法律方面的顾虑，我以前很多训练数据都是使用 AI 生成的。而这一次，这个模型的训练素材很大一部分来自我自己的 iPhone 4S 和 iPhone 3GS（没错，它们居然还能开机，Good ol’ Apple 😂🍎），以及一些以前保存下来的真人老照片。

甚至你可能可以生成出16 年前的我，因为训练集里确实有几张照片里出现了我本人。😏 所以……我们来玩一下 “大家来找我” 吧 😂

🔓 和我之前的模型一样，这个模型也是**无审查（Uncensored）**的，希望它能够带给你一些美好的回忆。

⚠️ 已知问题

• 这个模型主要针对人物摄影进行训练，所以在复杂场景和环境生成方面并不是特别擅长。

• Liquid / 湿皮肤的问题依然存在，但在 @Loraholic 的帮助下已经好很多了。他提供了一些私人的 LoRA 来帮助解决这个问题，非常感谢他！❤️

💬 如果你喜欢这个模型，欢迎留下评论或者建议。如果大家喜欢的人比较多、点赞也足够，我会考虑继续更新这个系列。

这个模型从数据整理到 Caption 花了我非常非常长的时间。其中很大一部分都是人工完成的，而且包含大量重复、枯燥的工作。也希望大家理解，我真的是一个ONE MAN army——不像 Civitai 上很多其他创作者可能拥有更多资源或者更强的硬件，我基本就是一个人，抱着我的老伙计 3080 Ti 一路干过来的。😂

好了，祝大家生成愉快！希望这个模型能带你找回一些属于那个年代的回忆。❤️📸

可以尝试的提示词:
照片是用一台廉价的早期数码相机拍摄的，使用了机顶闪光灯，强烈的正面光线，阴影生硬，皮肤高光过曝，背景略暗，白平衡偏暖且不稳定，色彩鲜艳但略有偏差，动态范围有限，阴影部分可见传感器噪点，轻微的数码锐化光晕，轻微的JPEG压缩，曝光不足。这是一张随意的业余摄影作品，带有2000年代初日本网络摄影的风格，属于随性抓拍。