Repository navigation
fix(director): correct optical coverage and complete camera feedback - #131
Conversation
Correct depth quantization, calibrate Fast defocus to the actual splat kernel, and align Operator, focus feedback and timeline edits with the observed reference. Add controlled optical diagnostics and retain completed frames by rendered state. Validation: Director 36 tests, typecheck and build; current browser and four-scene optical evidence recorded. Refs: MF-106, MF-62
Record source, current optical and dynamic output validation; update release mirrors without changing resource objects. Refs: MF-106, MF-62
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
Bugbot couldn't run - usage limit reachedBugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit. A user or team admin can review and increase usage limits in the Cursor dashboard. (requestId: ad0e287f-8d20-40f9-894f-86eb7d58bf2d) |
Bugbot couldn't run - usage limit reachedBugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit. A user or team admin can review and increase usage limits in the Cursor dashboard. (requestId: 45dec201-0f23-4508-bfa4-9bfde69f874f) |
Bugbot couldn't run - usage limit reachedBugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit. A user or team admin can review and increase usage limits in the Cursor dashboard. (requestId: f1dfdf35-6c0f-4be1-b009-95aaa9e8509c) |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 8194cdb3a4
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| : state.entryBlend < 1 | ||
| ? [state.entryBlend, state.shot.transition] | ||
| : state.exitBlend < 1 | ||
| ? [state.exitBlend, state.shot.exitTransition] |
There was a problem hiding this comment.
When a paused first shot has matching entrance and exit transition settings, identical endpoint poses, and symmetric blend values, these branches produce the same cache key even though compose() passes different entering values. Push and zoom transitions therefore restore the cached entrance frame while previewing the exit, placing or scaling the image in the wrong direction. Include the transition phase (entry versus exit) in the key.
Useful? React with 👍 / 👎.
| : best, | ||
| 0, | ||
| ); | ||
| if (workingPose) result.keys[selected].pose = structuredClone(workingPose); |
There was a problem hiding this comment.
When the playhead is between camera positions, workingPose is the interpolated pose returned by viewPose, so clicking “Add position” first replaces the nearest existing saved position here and only then inserts another position. This silently changes an endpoint or another previously saved camera even though the user invoked Add rather than the separate Save action; use workingPose for the new position without overwriting the selected key.
Useful? React with 👍 / 👎.
远景范围曾把非常接近的前后高斯量化成同一深度档，导致透明遮挡错误；Fast 按理想高斯校准又使松手后的圆孔径焦外尺寸明显变化。摄影路径改用 Float32 深度键与固定点序的上游 GPU radix 排序，并按实际截断高斯径向矩校准扩散、保留面积透明度补偿。受控单点等效半径差由 -8.27% 减至 +0.07%；原生对照不改动。
同时按 10 月 9 日实际来源修正 Operator padding/状态、内联读数/滑杆、对焦标记输入生命周期、兴趣区域和 Compose 时序、胶片标签/机位增删。时长弹层先提交再卸载，Enter 提交/Escape 取消并返回机位焦点；缓存忽略已解析的控件标签和未生效镜头元数据，仍按实际摄影/转场状态失效。
8194cdb3a4c8d0eb0f57c116c952d549b2207855的 CI 37884934299所选Viewer/Director、数据、治理、文档、CodeQL及summary成功；其余组件未选中。公开范围不增加 Studio、工程库、Frame 或在线服务；模型/环境/settings/资源对象/schema 不变。原始资产缺面、扫描残影与真实材料参数不足单独解释，未运行实际手机多指及 Safari/Edge，不把实验数值一致称为完整物理恢复。中文图片、过程视频、参数与哈希索引见唯一验证记录，中文原生附件与原因分析已发布，7图/2视频匿名GET字节哈希一致；正式部署结果随后回填。
版本为兼容 PATCH 5.21.7；先验证完整站点预览，再合并、回填实际产品 SHA 和正式 build/tag、部署并观察15分钟。回退目标为5.21.6成功部署
6ac7ec0fa8239794c0ab5d12。Refs: MF-106, MF-62
完整站点预览
6ac86d1c8db6a7638d84d9b1ready，8条redirect/9条header、8项文件比对/300项路由引用通过；真实缇宝/环境、390亮色PHOTO/VIDEO与桌面Operator已复验。四场景5种输入各5轮共100轮程序化调度压力测试，全有新画面及最终状态、无错误，960×540；与原生DOM输入证据分开。治理初次缺PR trace及公共镜像分别失败后如实修正，未绕过检查。远端自动Codex对e1e4c3e未报行内问题，Copilot/Cursor因额度未运行；不称独立人工审查。