Auto-Generate Lighting ScenesOn-App AI
### Feature Introduction
The AI Generative Lighting Scene module leverages On-App AI on-device capabilities, combined with room space and lighting device capabilities, to automatically generate usable and executable lighting atmosphere solutions for users. The system recommends 9 lighting scenes and supports replacement, preview, and save functions, enabling quick creation of lighting environments without professional configuration.
This module is suitable for various lighting scene needs including home lighting, hotel apartments, commercial spaces, etc., improving the efficiency and experience of creating lighting atmospheres for users.
### Feature Description
**1. Auto-Generate Lighting Scenes**
- The system intelligently generates 9 candidate lighting scenes based on room information and lighting capabilities, each scene includes:
- Scene name
- Color/brightness/color temperature scheme
- Lighting control actions (action list)
Users can directly click to preview or save.
**2. Replace Lighting Solutions**
- Supports replacing individual scenes
- Supports batch replacement of unselected scenes
- AI will regenerate lighting configurations
Used to obtain lighting effects that better match preferences.
**3. Lighting Scene Preview**
- Click on a card to preview in real-time:
- Lights change immediately
- Does not affect device persistent configuration
- Can revert upon exit
Enhances experience certainty.
**4. Multi-Select Save to Cloud Scenes**
- Users can select favorite lighting solutions
- Once saved, they can be invoked and executed in the App long-term.
> Non-administrator accounts will be blocked from save operations
### Interaction Flow
```mermaid
sequenceDiagram
participant U as User
participant UI as Frontend Page
participant Room as Room Data Service
participant AI as AI Lighting Engine
participant Dev as Lighting Device
participant Cloud as Cloud Scene Library
U->>UI: Enter AI lighting page
UI->>Room: Get home/room information
Room-->>UI: Return roomId
UI->>Cloud: Get available lighting style data
Cloud-->>UI: Return style list
UI->>UI: Sample generate 9 candidates
UI->>AI: predictLightScenes(9 items)
AI-->>UI: Return lighting action configuration
UI->>U: Display recommended lighting scenes
U->>UI: Click preview
UI->>Dev: Send lighting actions
Dev-->>U: Real-time lighting effect
U->>UI: Click refresh (single/all)
UI->>AI: Regenerate action configuration
AI-->>UI: Return solutions
UI->>U: Display new lighting cards
U->>UI: Select multiple scenes
U->>UI: Click save
UI->>Cloud: Batch saveScene
Cloud-->>UI: Return success/failure
UI->>U: Show status
```
### Dynamic Dataset
```mermaid
flowchart LR
subgraph BD[Business Definition]
A[Lighting Style Library] --> B[Room Attributes]
B --> C[Device Capability Model]
end
subgraph AI[AI Lighting Generation Model]
C --> D[Style Matching]
D --> E[Lighting Parameter Inference]
E --> F[Device Action Generation]
end
subgraph DS[Scene Dataset]
F --> G["Lighting Scene Dataset\n(Name+Parameters+Action)"]
end
```
**Lighting Scene Data Generation Process Description:**
- Lighting Style Library: Preset atmosphere styles (e.g., reading, cinema, party)
- Room Attributes: Region ID / light group information
- Device Capabilities: Brightness / color temperature / RGB / controllable attributes
- Style Matching: Filter applicable lighting strategies
- Parameter Inference: Generate target lighting effects
- Action Generation: Output device control instructions
### Module Working Principle
The module core is based on the following logic:
#### 1. Get room information when entering the page
```ts
const homeId = await getAppHomeId();
const { roomId } = await getRoomList();
```
#### 2. Get lighting scene candidate set
```ts
const list = await getSceneDataList(homeId, roomId);
```
#### 3. Perform AI lighting prediction
```ts
predictLightScenes({
roomId,
generateSceneStyles: list,
sceneType,
})
```
#### 4. Support single card refresh / batch refresh
```ts
refresh(item)
refreshAll()
```
#### 5. Preview lighting effects
```ts
previewScene({
actions,
parentRegionId: roomId
})
```
#### 6. Batch save
```ts
await Promise.all(selectedList.map(saveScene));
```