关于过去，现在，将来 / About the past, present and future

关于归档 / About the archive

老实说，我低估了归档整理的难度。除开工作原因，关于个人的文档是从来没做过仔细的整理和归档，我预想中的归档也就文件一拉然后打包，最后上传作品，一天就能完成，但实际上手后才发现完全不一样。我需要对过去创作所产生的文件进行归类和梳理，整理过程中我才发现我过去的创作流程以及原始文档的管理非常混乱，以至于完全无法梳理和归档；2024年重启时我会有意识地去梳理和改进我的创作流程，相对来说比较容易整理。

最终我花了好几天的时间才完成了这些工作，并对之前发布过的作品（19-22年）全部导出至最高的分辨率。所打包的作品不仅包含我过去已发布作品的最大分辨率版本，还会包含由于一些原因没有发布的作品，以及一些探索和实验性质的创作过程记录图。当然这些未发布的作品基本集中在19-22年这个时间段，24年所打包的作品则没有变化。AiLab的24年打包作品除了作品+WiP外会加上创作中使用的草稿以及24年初被封禁时发布的两个作品。

以上是关于作品归档，接下来，是我个人经历、思考以及未来计划

To be honest, I underestimated the difficulty of archiving. Except for my job, I've never carefully organized and filed my personal documents. I expected that archiving would be a matter of dragging and dropping files, packing them up, and uploading my work in a day, but I actually realized that it's completely different. I needed to categorize and sort through the files I had created in the past, and during the process I realized that my past creative process and management of the original files was so confusing that it was impossible to sort and archive them; for the 2024 reboot I was going to make a conscious effort to sort and improve my creative process, which would be relatively easy to organize.

In the end it took me several days to accomplish all of this and export all of my previously published work (years 2019-22) to the highest resolution. The packaged works not only contain the highest resolution versions of my past published works, but also include works that for some reason were not published, as well as some works of the creative process that were exploratory and experimental. These unpublished works are of course concentrated in the years 2019-22, while the works packaged in 2024 unchanged; AiLab's 2024 packages will include the released works + WiPs, also with the unpublished drafts used in the creation of the works, and the two works released in early 2024 during the time of the suspension.

The above is about the archiving of works, and next, my personal experience, thoughts, and future plans.

过去发生了什么 / What happened in the past?

之前我会经常停止更新，是因为我的身体出了很大的问题，有很长的一段时间只能卧床修养，无法自由活动，连坚持坐着半小时都不行，更不要说进行创作了。

为了恢复身体，我告别了很多以前喜欢的事物；需要经常去医院做各种治疗，也做了几个手术；也多次做身体检查尝试找出原因，但始终找不到原因。在一段时间里，我既要面对找不出原因的身体病痛，还要兼顾我的日常工作，令我身心俱疲。

幸好，这些小小的改变慢慢累积了起来后，我的身体正在逐渐恢复，至少日常活动已经没有问题；为了改善我的精力和专注力，我需要定期进行一定强度的运动，也比之前好了不少。可惜的是在我打算重启创作活动时，我运营将近8年的Pixiv账号被封禁。接下来的事情大家都知道了，开创了新账号，而为了避免风险将AiLab的内容进行分开更新，虽然过程有点小波折，但总算是恢复了更新，并持续到现在。

Before I would often stop updating because I had a big health problem, and for a long time I could only stay in bed and recuperate, unable to move freely, and I couldn't even insist on sitting for half an hour, let alone creating.

In order to recover my body, I said goodbye to many things I used to favor; I needed to go to the hospital frequently for various treatments and had several surgeries; I also had many body checkups to try to find out the cause, but I could never find it. For a period of time, I was physically and mentally exhausted as I had to deal with my unexplained physical ailments as well as my daily work.

Luckily, as these small changes slowly accumulated, my body was gradually recovering, and at least my daily activities were no longer a problem; I needed to exercise regularly at a certain level of intensity in order to improve my energy and concentration, which was also much better than before. Unfortunately at the time I was planning to restart my creative activities, my Pixiv account, which I had been running for almost 8 years, was banned. The next thing you all know is that a new account was created, and the process of updating the AiLab content separately to avoid risk was a bit of a bumpy ride, but it finally resumed and continues to this day.

接下来的计划 / Next Step：TKO_MK2

先说一下我在TKO_MK2作品的创作流程。

创作流程趋于稳定后，基本是这样的：

• 根据题材找天空盒、场景图片或者根据好的场景素材确定主题

• 放入人物并搭配服装

• 对人物姿势进行调整，打光

• 调整人物模型和服装穿脱制作不同版本，为制作差分做前期准备

• 人物和服装分开输出素材

• 其中人物的皮肤材质将会根据高光、法線貼圖分别输出素材，方便后期在Photoshop混合

• 进入后期流程，人物皮肤材质处理，添加服装和阴影，调整整体光影结构，最后和背景融合并进行整体调色处理

目前的创作流程中，最花费时间的环节是寻找合适的背景，不是所有的天空盒和场景图片都适合用于创作，它需要同时拥有强烈的光影以及强烈的主题性。今年重启后能够大量更新的原因很大一部分是因为都是全都为重制作品，这意味着这些作品的前期准备工作已经做得差不多了，后期更新的新作品包括所有的前期工作都需要重新开始准备，所以后期更新会花费了更多的时间，这也是后续更新变慢的原因。

虽然TKO_MK2的热度以及关注数量完全不及AiLab，流程也比较枯燥和麻烦，但我还是非常享受创作的过程，除了让我更熟练Photoshop的各种用法，枯燥的流程也能使我浮躁的内心平静下来，能让我更沉浸地去把控作品的各种细节。

老实说，作为一个既使用Ai也使用实时渲染创作的人，我对Ai的心情十分复杂，一方面我可以让Ai让我目前的作品变得跟现实一样，但另一方面这也让我这几年对风格和流程的探索以及所做出的努力也跟着消失了，所以我一直对TKO_MK2的作品介入Ai持谨慎态度，而且不谨慎的后果大家也看到了。所以TKO_MK2这边我仍会纯人工进行创作，我也会继续提升自己的技能，比如最近的更新我都运用了电子绘画中的各种技巧来改善画面表现力。

由于VaM这个工具创作限制比较大，今年我可能会尝试学习使用Daz或者Blender以获得更好的效果，但因为要兼顾AiLab的更新，为了挤出时间学习新技能和兼顾现实生活，今年TKO_MK2这边仍会将重心放在重制过往已发布和未发布的作品上。

Let's start with my creative process for TKO_MK2 works.

After the creation process has stabilized, it's basically like this:

• According to the theme, find the skybox, scene pictures or determine the theme based on good scene materials.

• put in the characters and match the costumes

• Adjust the pose of the characters and lighten.

• Adjust the character model and clothing wear off to make different versions, for the production of differential preparation

• Separate output materials for characters and costumes

• The character's skin material will be based on the highlights, normal mapping output material, to facilitate the later in the Photoshop mixing

• into the post-process, character skin material processing, add clothing and shadows, adjust the overall light and shadow structure, and finally blend with the background and the overall color adjustment

The most time-consuming part of the current creative process is finding the right background, not all skyboxes and scene images are suitable for creation, it needs to have both strong light and shadow as well as a strong cinematic nature. The reason for the large number of updates after this year's reboot is that they are all remakes, which means that the prep work for these works is pretty much done, and new works for later updates include all the prep work that needs to be done again, so the later updates will take more time, which is also the reason for the slower subsequent updates.

Although TKO_MK2 is not as popular as AiLab, and the process is quite boring and troublesome, I still enjoy the process of creation very much. In addition to making me more skillful in the various uses of Photoshop, the boring process can also calm my restless heart, and allow me to be more immersed in controlling the various details of the work.

Honestly, as a person who uses both Ai and real-time rendering to create works, I have very mixed feelings about Ai, on one hand I can let Ai make my current works become like reality, but on the other hand it also makes my exploration of style and process and the efforts I made in the past few years disappear along with it, so I've always been wary of intervening in Ai for TKO_MK2's works and the the consequences of not being cautious have been seen by everyone. So I'm still going to work purely manually on TKO_MK2, and I'm going to continue to improve my skills, for example, in the recent updates I've utilized various techniques from digital painting to improve the expressiveness of the images.

Due to the limitations of VaM as a tool, I may try to learn to use Daz or Blender this year to get better results, but since I have to take care of AiLab updates, I'll still focus on remastering published and unpublished works this year for TKO_MK2 in order to make time to learn new skills and take care of real life.

接下来的计划 / Next Step：TkoAiLab

Ai的发展速度实在是太快了，每几个月就会有新的模型出现，很难说有什么具体的计划，好在创作流程和作品风格趋于稳定，也算是能够应对快速更新迭代的模型同时保持自己的创作风格。

很多人经常会问我用的什么模型进行生成，首先我使用的都是Civital上能够下载的热门模型，且避免使用各种画风lora；其次目前的作品我会使用2到3个大模型，目前的制作流程大概如下：

• 根据主题进行纯提示词图片生成测试，并不断优化提示词和构图

• 风格大概确定后在VaM里进行人物摆放、姿势调整以及摄像机角度输出素材

• 进Photoshop绘制大概的服装以及简单的背景形状

• 将处理好的图片加入到Controlnet，再次使用文生图加上不同风格的模型进行混合生成，并不断加入各种首饰、衣服等细节关键词

• 选择一张最满意的图片作为基底，在此基础上在Photoshop制作各种差分版本，衣服、首饰以及各种差分部件也会使用绿色背景单独导出

• 将制作好的图片以及差分部件放大到5K分辨率

• 放大后将脸、身体、以及包含衣服和首饰的身体部位分别截图，在Photoshop处理过后进行图生图分别来改善细节、修复肢体问题、进一步加入各种差分版本等等

• 全部完成后会调整整体光影结构，整体调色处理，最后导出不同的差分版本

大家可能看出我没有更新过插入场景，原因是比起单人，多人的场景复杂度会更高，尤其是插入，我花了很多的时间来进行测试，而且我更希望这不是一个单独的事件，而是会有一定的故事背景以及铺垫，我估计大家已经看出来上司系列是一个连续的时间线，我保证后续会有大家想看到的内容。

虽然这些用于铺垫的内容能够制作的差分版本不会很多，但付出的精力是一样的。之后的更新也将会是这样的模式，如果有个人觉得效果好的图片生成视频也会一同更新。

The development speed of Ai is really fast, new models appear every few months, it's hard to say what specific plan, but the good thing is that the creation process and style of work tend to be stable, it is also able to cope with the rapid updating and iteration of models while maintaining my own creative style.

Many people often ask me what models I use to generate, first of all, I use all the popular models that can be downloaded from Civital, and avoid using all kinds of painting style lora; secondly, I will use 2 to 3 models for my current works, and the current production process is as follows:

• Test t2i according to the theme, and continuously optimize the prompt and composition.

• After determining the style, I'll do the character placement, pose adjustment and camera angle output in VaM.

• Photoshop to draw a rough costume and a simple background shape.

• Add the processed images to Controlnet, and again use t2i plus different styles of models for mixing and generating, and continue to add a variety of accessories, clothing, and other details keywords.

• choose the most satisfactory picture as a base in Photoshop to create differential versions, clothing, accessories, and differential parts will also use the green background exported separately

• Enlarge the image and differential parts to 5K resolution.

• After enlarging, the face, body, and body parts including clothes and jewelry will be screenshot separately, and processed in Photoshop to i2i to improve the details, repair the limb problems, and further add various differential versions, etc.

• All completed will adjust the overall light and shadow, the overall color adjustment, and finally exported different differential versions

As you can probably tell I haven't updated the insertion scene, the reason is that the complexity of the multiplayer scene will be higher than the single player, especially the insertion, which I spent a lot of time testing, and I would prefer that this is not a stand alone event, but rather there will be a certain background to the story as well as the buildup, and I'm guessing that you can already tell that the superior series is a continuous timeline, and I promise that there will be content that you will want to see in the future.

Although there won't be many differential versions that can be made of this content used for buildup, the effort put into it is the same. Future updates will follow the same pattern, and if there are any image-generated videos that I personally feel work well, they will be updated along with them.

很感谢大家能够看到这里，正值快到农历新年，提前祝大家新年快乐，也希望自己能够努力更新给大家带来更好的作品！

Thank you very much for being able to read this, just as it's almost Chinese New Year, I wish you all a happy new year in advance, and I hope I can work hard to update and bring you better works!