Image Sending

After enabling this feature, you can send images to characters. The AI model will recognize and describe the image content to facilitate conversation.

Operation Path

1. Open the left sidebar.
2. Click the "More" option, find and select "Voice" in the menu.
3. After entering the settings interface, select the "Chat Settings" tab.
4. In the chat settings, find and enter the "Image Sending" submenu.

Core Configuration Steps

1. Enable the Feature: Set the "Image Sending" function switch to the "On" state.

2. Configure Image Description API (Key Step):

   ◦ You need to select a model API that supports multimodal recognition.

   ◦ API Requirement: The model behind this API must have the ability to understand image content (i.e., a "multimodal model"), such as GPT-4V, Claude-3, Gemini, etc.

3. Configure Prompts (Recommended Default):

   ◦ Image Description Generation Prompt: Used to guide the model on how to describe the images you send.

   ◦ Image Description Injection Prompt: Used to guide the system on how to integrate the image description into the conversation context with the character.

   ◦ Friendly Reminder: If you are unfamiliar with writing prompts, it is recommended to directly use the system's default prompts, which usually yield good results.

After completing the above configurations, you can send images to characters in the chat interface.