StellarCreateOptimization

Stellar Create Optimization 1.0.4

Added

• Added comprehensive Display Board rendering optimizations on Forge 1.20.1 and NeoForge 1.21.1.

Changed

• Small and Large Water Wheels on NeoForge now cache their complete original geometry instead of using reduced-detail proxy models.
• Water Wheel persistent VBO rendering no longer requires CreateBetterFPS; unsupported models and rendering states continue using Create’s original renderer.

Fixed

• Fixed a NeoForge 1.21.1 crash when CreateBetterFPS and Sodium were installed without Iris. The Iris-specific buffer Mixin now loads only when the required rendering stack is present.
• Removed the hard runtime dependency on Iris’ internal NormI8 class by adding an internal packed-normal implementation.
• Fixed signed packed-normal and tangent decoding, including preservation of the tangent W handedness used by normal mapping.
• Fixed the Iris entity vertex stride from the incorrect 56-byte layout to the actual 54-byte layout used by Iris 1.8.12.
• Fixed possible corruption of Iris entity IDs, midpoint UVs, normals, and tangent attributes caused by writing vertices with the wrong stride.
• Fixed persistent VBO geometry occasionally not being submitted when every visible object came from cache and Iris produced no temporary buffer segment.
• Fixed linked-block and Factory Gauge outline positioning. The renderer now includes the event’s model-view matrix, preventing outlines from floating above blocks or appearing near the camera when the view rotates.
• Fixed Display Board fallback rendering so non-controller sections do not repeat controller text and font lookup work.
• Fixed Display Board cogwheel RenderType reuse and cache invalidation across resource reloads and world changes.
• Unsupported shader stages, custom models, subclasses, dynamic transforms, incompatible vertex formats, and optional-mod API mismatches continue using their original rendering paths.

Performance

• Water Wheels: 100–115 FPS after optimization, compared with 0–2 FPS before optimization — at least approximately 4,900% faster.
• Mechanical Belt: 20–25 FPS after optimization, compared with 3–5 FPS before optimization — at least approximately 300% faster.
• Shafts: 35–38 FPS after optimization, compared with 3–4 FPS before optimization — at least approximately 775% faster.
• Blaze Burners: 36–40 FPS after optimization, compared with 0 FPS before optimization — at least approximately 3,500% faster, conservatively calculated using 1 FPS for the stalled baseline.
• Millstones: 45–50 FPS after optimization, compared with 0–1 FPS before optimization — at least approximately 4,400% faster.
• Display Boards: 57–65 FPS after optimization, compared with 0–1 FPS before optimization — at least approximately 5,600% faster.

Results were measured in high-load test scenes. Actual performance depends on hardware, shader pack, renderer version, rendering distance, configuration, and factory size.

Stellar Create Optimization 1.0.4

新增

• 为 Forge 1.20.1 和 NeoForge 1.21.1 新增完整的翻牌显示器渲染优化。

调整

• NeoForge 的大小水车改为缓存完整原始网格，不再使用降低细节的代理模型。
• 水车持久化 VBO 不再强制依赖 CreateBetterFPS；不支持的模型和渲染状态仍使用 Create 原始渲染器。

修复

• 修复 NeoForge 1.21.1 在安装 CreateBetterFPS 和 Sodium、但没有安装 Iris 时发生的启动崩溃。Iris 专用缓冲 Mixin 现在只会在所需渲染栈完整时加载。
• 新增模组内部的打包法线实现，移除对 Iris 内部 NormI8 类的运行时依赖。
• 修复有符号打包法线和切线的解码，并正确保留法线贴图使用的 tangent W 正负手性。
• 将 Iris 实体顶点步长从错误的 56 字节修正为 Iris 1.8.12 实际使用的 54 字节。
• 修复错误顶点步长可能导致的 Iris 实体 ID、中点 UV、法线和切线属性错位或损坏。
• 修复所有可见对象都命中持久缓存时，Iris 因未生成临时缓冲区段而不提交 VBO 队列的问题。
• 修复物流连接方块和工厂仪表轮廓错位。绘制时现在会包含事件提供的模型视图矩阵，避免转动视角后轮廓悬在方块上方或出现在镜头附近。
• 修复显示板非控制器区段重复执行控制器文字渲染和字体查询的问题。
• 修复显示板内部齿轮 RenderType 复用，以及资源重载和切换世界后的缓存失效问题。
• 不支持的光影阶段、自定义模型、子类、动态变换、顶点格式和可选模组 API 会继续安全回退至原始渲染路径。

性能表现

• 水车：优化后 100–115 FPS，优化前 0–2 FPS，至少提升约 4,900%。
• 传送带：优化后 20–25 FPS，优化前 3–5 FPS，至少提升约 300%。
• 传动杆：优化后 35–38 FPS，优化前 3–4 FPS，至少提升约 775%。
• 烈焰人燃烧室：优化后 36–40 FPS，优化前显示为 0 FPS；按 1 FPS 保守折算，至少提升约 3,500%。
• 石磨：优化后 45–50 FPS，优化前 0–1 FPS，至少提升约 4,400%。
• 翻牌显示器：优化后 57–65 FPS，优化前 0–1 FPS，至少提升约 5,600%。

以上数据来自高负载测试场景。实际性能会受到硬件、光影包、渲染器版本、视距、配置和工厂规模影响。