A planned character runtime SDK and standalone player for SpriteForge character packs. It aims to turn expressions, animation cues and speech playback signals into a coherent character performance, with shared rendering and resource management that applications can embed.
Status: placeholder and design draft. This repository currently contains documentation only. There is no installable SDK, runnable player or stable API.
SpriteForge creates and exports character assets and behavior graphs. Today, additional performance rules live in Amadeus: speaking routes, transition holds, silence mouth overlays, post-speech exits and runtime loading.
Amadeus-SpriteForge-Animator is intended to give those reusable responsibilities one home, so the authoring preview, a standalone player, Amadeus and other applications can use the same runtime and improve it independently.
- Behavior-graph execution and character performance rules: expressions, transitions, loops, holds, speech completion and interruption.
- Mouth presentation: silence closure, frame-specific anchors, overlays and closed-mouth settling after speech.
- Animation delivery from parsed TXT or other scripted cues, with an optional text-format adapter and synchronization to actual audio playback.
- Sprite rendering, KTX2 loading, prefetching, texture residency budgets and explicit resource cleanup.
- A standalone player for replaying and inspecting performances, plus thin adapters for embedding the SDK in host applications.
Hosts supply semantic intent, audio playback state and the selected presentation owner. SpriteForge owns asset production, editing and export. Animator owns how the selected performance plays out. The initial direction is an embeddable JS/TS runtime with a PixiJS renderer; process or network integration can be added when a real caller needs it.
See PLAN.md for the draft boundary, migration stages, acceptance criteria and unresolved decisions. The plan is a proposal, not a release schedule or compatibility promise.
Amadeus-SpriteForge-Animator 计划提供 SpriteForge 角色运行时 SDK 与独立播放器， 把表情、动作、解析后的 TXT 演出事件和语音播放信号组织成连贯的角色表演。
当前为占位仓库与设计草案，只有文档；SDK、播放器和稳定 API 尚未实现。
目前，角色行为图之外的说话路线、过渡停留、静音闭口、说完后的收尾与资源加载等 逻辑仍在 Amadeus。这个项目希望把可复用的角色演出逻辑独立出来，让 SpriteForge 预览、独立播放器、Amadeus 和其他应用共享同一套实现，单独修改、验证和优化。
计划涵盖行为图与导演规则、嘴部覆盖和逐帧锚点、演出事件与音频同步，以及 KTX2、 纹理缓存、预加载、资源预算和释放。宿主决定表达内容、音频播放状态与呈现来源； SpriteForge 负责素材创作和导出；Animator 负责角色怎样演出来。
详细边界、迁移阶段和第一版验收条件见 计划草案。