feat: LOD 网格分级——meshopt 解析期简化 + 按距离/级联选级

vendored meshoptimizer 1.2(submodule 环境故障改直接 vendor)。Mesh
多 LOD:各级索引连续拼进同一 index buffer(共享顶点,LodRange 区间
选择),解析期 meshopt_simplify 生成链(LOD0=原始,每级目标减半,
相对误差 1e-2,小网格复用上级);都在 worker 线程,加载零额外卡顿。

SceneRenderer 按 (mesh, LOD) 聚合 batch(lookup key = 指针<<8|lod):
主视图按屏占比近似 radius/dist 选级(阈值 10/20 倍半径);阴影级联
按级联序降级(越远越简)。Draw 按 batch.lod 选索引区间。

A/B 对比(100 背包 10×10,RTX 5060 Laptop,vsync off):
- 纯背包:LOD 开 93-95 fps / 关 90-107 fps —— 无显著差异
- 加 SunTemple:LOD 开 52-56 / 关 51-55 —— 无显著差异
结论:当前场景瓶颈不在顶点数(5060 顶点吞吐富余),LOD 基础设施
保留(几何复杂资产/低端 GPU/更大场景生效),帧率问题转向 Tracy
GPU 时间线定位真实瓶颈。

坑:LodRange 记录"LOD 不足复用最后一级"时 first 不得继续累加
(buffer 无重复数据,累加必越界——VUID-08798 抓的)。
This commit is contained in:
2026-09-24 20:53:19 +08:00
parent 34b593f734
commit 5bdfb53566
125 changed files with 67608 additions and 29 deletions
+63
View File
@@ -0,0 +1,63 @@
Thanks for deciding to contribute to meshoptimizer! These guidelines will try to help make the process painless and efficient.
## Questions
If you have a question regarding the library usage, please [create a GitHub discussion](https://github.com/zeux/meshoptimizer/discussions/new/choose).
Some questions just need answers, but it's nice to keep them for future reference in case other people want to know the same thing.
Some questions help improve the library interface or documentation by inspiring future changes.
## Bugs
If the library doesn't compile on your system, compiles with warnings, doesn't seem to run correctly for your input data or if anything else is amiss, please [open a GitHub issue](https://github.com/zeux/meshoptimizer/issues/new).
It helps if you note the version of the library this issue happens in, the version of your compiler for compilation issues, and a reproduction case for runtime bugs.
Of course, feel free to [create a pull request](https://help.github.com/articles/about-pull-requests/) to fix the bug yourself.
## Features
New algorithms and improvements to existing algorithms are always welcome; please open an issue describing an improvement you'd like to see or make before opening a pull request.
This will give us a chance to discuss the idea before implementing it - some algorithms may not be easy to integrate into existing programs, may not be robust to arbitrary meshes or may be expensive to run or implement/maintain, so a discussion helps make sure these don't block the algorithm development.
## Code style
Contributions to this project are expected to follow the existing code style.
`.clang-format` file mostly defines syntactic styling rules (you can run `make format` to format the code accordingly).
As for naming conventions, this library uses `snake_case` for variables, `lowerCamelCase` for functions, `UpperCamelCase` for types, `kCamelCase` for global constants and `SCARY_CASE` for macros. All public functions/types must additionally have an extra `meshopt_` prefix to avoid symbol conflicts.
## Dependencies
Please note that this library uses C89 interface for all APIs and a C++98 implementation - C++11 features can not be used.
This choice is made to maximize compatibility to make sure that any toolchain, including legacy proprietary gaming console toolchains, can compile this code.
Additionally, the library code has zero external dependencies, does not depend on STL and does not use RTTI or exceptions.
This, again, maximizes compatibility and makes sure the library can be used in environments where STL use is discouraged or prohibited, as well as maximizing runtime performance and minimizing compilation times.
The demo program uses STL since it serves as an example of usage and as a test harness, not as production-ready code.
## Testing
All pull requests will run through a continuous integration pipeline using GitHub Actions that will run the built-in unit tests and integration tests on Windows, macOS and Linux with gcc, clang and msvc compilers.
You can run the tests yourself using `make test` or building the demo program with `cmake -DBUILD_DEMO=ON` and running it.
Unit tests can be found in `demo/tests.cpp` and functional tests - in `demo/main.cpp`; when making code changes please try to make sure they are covered by an existing test or add a new test accordingly.
## Documentation
Documentation for this library resides in the `meshoptimizer.h` header, with examples as part of a usage manual available in `README.md`.
Changes to documentation are always welcome and should use issues/pull requests as outlined above; please note that `README.md` only contains documentation for stable algorithms, as experimental algorithms may change the interface without concern for backwards compatibility.
## Sensitive communication
If you prefer to not disclose the issues or information relevant to the issue such as reproduction case to the public, you can always contact the author via e-mail (arseny.kapoulkine@gmail.com).
## Contributor agreement
Any code you submit will become part of the repository and be distributed under the [meshoptimizer license](https://github.com/zeux/meshoptimizer/blob/master/LICENSE.md). By submitting code to the project you agree that the code is your work and that you can give it to the project.
You also agree by submitting your code that you grant all transferrable rights to the code to the project maintainer, including for example re-licensing the code, modifying the code, and distributing it in source or binary forms. Specifically, this includes a requirement that you assign copyright to the project maintainer.
## AI
AI-generated issues or pull requests (including pull request descriptions), will be closed. Repeat submissions may result in a ban.