# ChatCanvas demonstrations

[English](#english) | [简体中文](#简体中文)

## English

The README includes three edited demonstrations made from user-supplied screen recordings and a screenshot. These show actual input, canvas drawing and UI layout editing. They do not show model response speed, generated answers, or a benchmark.

| Clip | What is shown | Duration |
| --- | --- | --- |
| [Compose](https://github.com/Tryanstm/dsh-chatcanvas/blob/main/media/demos/compose.mp4) | Formatting screenshot, math notation refinement and a short typed instruction | About 21 seconds |
| [Canvas](https://github.com/Tryanstm/dsh-chatcanvas/blob/main/media/demos/canvas.mp4) | Inserting a canvas and sketching a layout | About 26.5 seconds |
| [UI editor](https://github.com/Tryanstm/dsh-chatcanvas/blob/main/media/demos/ui-editor.mp4) | Selecting, moving and arranging elements | About 24 seconds |

Editing includes cropping to the relevant controls, removing pauses and repeated input, accelerating selected operations, interpolating motion to 60 fps, and brief transitions. The formatting still is identified as a screenshot. Clips are silent, include bilingual captions, and exclude subsequent AI replies. Crops omit surrounding conversation, account details and unrelated overlays. No interaction or generated result has been recreated.

MP4 files use H.264 with progressive download enabled. The READMEs use GitHub native video attachments and include separate MP4 download links. Poster images are also stored for reuse; playback controls depend on GitHub and the browser. Videos live under `media/demos/` and are excluded from the plugin installation package. The READMEs use online media URLs so installed documentation can still link to them.

For future demos, provide a focused recording of one feature: start with a clean example, perform the action, then hold the result briefly. Music playback, MIDI/audio import and export, standalone math plots and other features need their own recordings; the current videos do not demonstrate them. Recordings may predate the latest release, so minor visual details can differ.

## 简体中文

README 中的三段演示由用户提供的真实录屏和截图剪辑而成，展示输入、画布手绘和界面布局编辑，不展示模型回复速度、生成的答案或性能测试。

| 片段 | 展示内容 | 时长 |
| --- | --- | --- |
| [输入与表达](https://github.com/Tryanstm/dsh-chatcanvas/blob/main/media/demos/compose.mp4) | 格式截图、数学表达整理、输入补充说明 | 约 21 秒 |
| [画布手绘](https://github.com/Tryanstm/dsh-chatcanvas/blob/main/media/demos/canvas.mp4) | 插入画布并勾画布局 | 约 26.5 秒 |
| [界面编辑](https://github.com/Tryanstm/dsh-chatcanvas/blob/main/media/demos/ui-editor.mp4) | 选择、移动和排列元素 | 约 24 秒 |

剪辑包括聚焦操作区域、删除停顿与重复输入、局部加速、60 帧运动插值和短转场。静态格式示例已标注为截图。视频无声，配有中英文说明，均不包含后续 AI 回复；裁切范围也避开了周围会话、账号信息和无关悬浮层。没有重新制作交互过程或生成结果。

视频采用支持边下载边播放的 H.264 MP4。README 使用 GitHub 原生视频附件，并提供独立的 MP4 下载链接。封面图也一并保存，便于复用；具体播放控件取决于 GitHub 和浏览器。视频放在 `media/demos/`，不进入插件安装包；README 使用在线素材链接，安装包中的文档也可以打开这些演示。

后续可以继续提供每项功能各自的录屏：从干净的示例开始，完成一次操作，最后让结果停留片刻。音乐试听、MIDI／音频导入导出、独立数学图像等功能仍需对应素材，当前视频没有演示这些内容。录屏可能早于最新版，细节与当前版本可能略有不同。