Z's Blog

Welcome to my blog. This is a technical site focused on game development. My open-source projects on GitHub: github.com/hxhb and Open-Source projects.
The articles in the blog include a C++ series, Hot Update for UnrealEngine, Plugins and tools development, Game Resource Management series, and thoughts series, etc.

In my spare time, I research and develop technical solutions and preview the results: Showcase. If you are interested in any of the projects, feel free to email me at imzlp@foxmail.com.

UE as Lib

Dictionary-based Shader Compression

Remake PAK at Runtime

Multi-stage Resource Inspection

Splitting Game Resource

Human Pose Tracking in UE

In addition to this site, I also created a community-driven Chinese knowledge base site for Unreal Engine https://ue5wiki.com/, hosted on GitHub Pages. Contributions of technical content are welcome to help build the Chinese Unreal Engine technology community! Please see the submission rules: Wiki Content Writing Format and Rules .

Recently, I encountered an extremely bizarre bug involving two maps. One map, A, can be accessed with its PAK placed in the engine’s automatic mount directory, but it cannot be accessed from the hot update directory. The other map, B, behaves completely oppositely: it is abnormal in the automatic mount directory but works normally in the hot update directory.

At first glance, the issue appears entirely elusive, with two mutually exclusive behaviors occurring within the same logical framework. Moreover, the hot update mount and the automatic mount only differ in timing and priority, so this problem shouldn’t theoretically exist.

While the issue can ultimately be resolved on the business logic side, this behavior involves another very obscure path within the engine. Understanding why and how it works is crucial. Therefore, I analyzed the engine’s code based on this behavior, came to a reasonable conclusion, and devised a method to detect and mitigate this issue.

This article assumes that readers have some basic knowledge of UE hot updates; if in doubt, please refer to other articles in this blog’s hot update series for more information.

For personal knowledge management, I am very keen on creating my own local-first and open source self-hosted services. This way, I can avoid relying on any platform, have complete autonomy over my data, and be able to migrate services at any time. Options like Obsidian, Hexo blog, and the memos introduced in this article all meet these criteria.

The protagonist of this article, memos, is an open-source lightweight note-taking service that allows you to take notes in a way similar to posting on Weibo, supporting TAG marking and citations. It features an account system and permission management, allowing web access at any time, and notes can be public or private, offering great flexibility. I chose it because it complements Obsidian; Obsidian is still too heavy for my needs—it works well on PC but has a poor mobile experience. Thus, I need a lightweight, always-available note-taking service.

This article will introduce how to deploy a Memos service on a VPS using Docker, along with Nginx to bind the domain name, Certbot to issue and automatically update SSL certificates, and regular backups of the memos database, as well as some optimization configurations I made for the service.

In previous articles, I primarily introduced how HotPatcher can be applied in the hot update process. However, HotPatcher is not just capable of fulfilling the needs for hot updates; it is also a flexible resource packaging framework.

As the project development reaches a certain stage, a massive amount of resources needs to be packaged, and waiting for a complete package to be built can take a long time. If a real device encounters issues, the serial process of repair -> package -> verify can become exceedingly lengthy. There are often cases of script, data, or other resource errors that lead to package abnormalities.

In such cases, waiting hours to generate a complete package prevents smoke testing from proceeding. Even if you rebuild the package, there’s no guarantee that there won’t be other overlooked issues; otherwise, you have to start over again.

Based on this pain point, we urgently need a method to quickly fix resource issues within packages and validate them during the development phase. HotPatcher provides this functionality, allowing you to utilize it fully without waiting for a complete package.

This article will introduce how to utilize the editor features provided by HotPatcher to quickly package preset configurations and specific resources, accelerating the validation process of resources on real devices and significantly improving testing efficiency.

Since the last version release of HotPatcher over half a year ago, its support for UE4 has been commendable, but its support for UE5 has been lacking. There have been frequent inquiries in the group about adaptation.

The last Release version of HotPatcher was v81.0, released on 20230604, which only supports UE5.1.0 and does not support WP and Nanite. Subsequently, Epic released UE5.2 and 5.3, but the plugin has not been adapted. Although one can modify compilation errors themselves, functionality support in UE5 is somewhat in a beta state.

Taking advantage of the Qingming holiday, I worked intensively for a few days to comprehensively optimize and adapt HotPatcher to be compatible with the latest versions of UE4.21 to UE5. A new Release version has been published: v82.0. The latter part of this article will discuss the adaptation of HotPatcher in UE5, precautions, and the update log.

Additionally, in 2020, I wrote an article: Some Thoughts and Ideas on Open Source, and over the past two years, I’ve had some new thoughts on maintaining open source projects. Let’s summarize together.

During the New Year’s holiday, my iPhone 12 dropped and caused some malfunctions, making it barely usable. However, recently the battery also swelled and it easily shut down automatically, rendering it completely unusable. After switching to a backup phone, it was set aside for a while.

I recently thought of it again, as it could still power on normally and most of its main functions could still operate, so I speculated that it was only partially faulty and not completely damaged. It would be a pity to completely scrap it, and it seems I can only swap it for a stainless steel basin.

I took it apart to analyze the cause of the faults and attempted to buy parts for repair. This article documents the disassembly and resolution process.

This year, I plan to celebrate the Spring Festival in Shenzhen and the next stop for my Greater Bay Area trip is Macau. Traveling from Shenzhen to Macau is relatively convenient, so I am making arrangements for the trip. I’ve done some research and itinerary planning that can serve as a reference for those visiting Macau for the first time.

In the previous article Design and Implementation of Resource Self-Correction in UE, I introduced the implementation plan of filtering using the asset inspection of ResScannerUE, followed by automated processing.

Normally, if you want to check a certain property within a resource, you need to load the asset and retrieve the object:

However, if you want to check the properties of many resources in bulk, loading resources one by one in this way takes a long time, especially in the absence of DDC, where loading resources triggers the construction of DDC cache, which is time-consuming and has a high performance overhead.

As the size of the assets increases, scanning all resources in a project completely can be extraordinarily time-consuming. Therefore, I wonder if it’s possible to implement a method that does not require loading resources but can still retrieve properties within resources.

After research, I found a clever way to achieve this by combining reflection, asset serialization, and the features of AssetRegistry. Moreover, this is a non-intrusive approach that does not require modifying any existing resource type code.

This article will introduce the implementation principles and analyze the related logic in the engine, as well as present the practical application of this mechanism in the project.

In a previous article, I introduced the implementation and efficiency optimization of multi-stage and automated resource checks using ResScannerUE in UE.

• Multistage Automated Resource Check Solutions in UE
• Automated Resource Check Practices Based on ResScannerUE
• Resource Compliance Check Tool ResScannerUE in UE

This already allows for very flexible configuration of resource checks and timely alerts, with very powerful rule expression capabilities. However, they are still limited to the stage of discovering problems for reminders; although there are constraints to prevent submission, manual processing is still required. Especially in cases with a large amount of existing resources, the manpower involved in manual processing is very time-consuming.

Based on this situation, I designed a mechanism for automating resource corrections that can automatically fix resources after rules are established, ensuring that all scanned resources are set to the correct state.

This completes the final piece of the puzzle in my resource check solution and can be applied throughout the scanning process as needed. This article will introduce its design concept and implementation mechanism, as well as some practical application scenarios in projects.

If I had to choose one recent experience in working with UE that felt the most meaningless and torturous, it would definitely be trying to find a resource in the ContentBrowser based on a long string path.

The inability to quickly jump directly to a directory or resource makes me feel like I’m wasting my life every time I have to navigate deeply nested folders one step at a time.

To address this pain point, I wrote a small tool that significantly alleviates this manual anxiety. I named it j2 (jump to), which is a minimalist ContentBrowser jump tool, open-source on GitHub: hxhb/JumpTo.

This article will introduce the usage of this project, the thought process behind conceptualizing the solution, and the step-by-step code implementation. Although the functionality itself is a very simple plugin, the discovery, analysis, and solution process for the actual pain points is worth documenting.