Chat System
Master Tiny Land's powerful chat features for the best conversational experience.
Overview
Tiny Land's chat system is designed for immersive, dynamic conversations with AI characters. With features like swipes, branching, and regeneration, you have complete control over your interactions.
Basic Chat Controls
Sending Messages
- Type your message in the input box at the bottom
- Press Enter to send (Shift+Enter for new line)
- Or click the Send button
- Wait for the AI to generate a response (usually 2-5 seconds)
Message Input Tips
- Use *asterisks* for actions: *smiles*, *looks around*
- Use quotes for speech: "Hello, how are you?"
- Describe your character's emotions and context
- Be descriptive to get richer responses
Swipe Between Responses
One of Tiny Land's most powerful features - multiple responses for every message.
How Swipes Work:
- When a character responds, you can generate alternative versions
- Use the â â arrows next to any message to swipe
- Characters with multiple greetings start with swipeable options
- Each swipe is a different interpretation of the same prompt
When to Use Swipes:
- The response doesn't match your expectations
- You want to explore different conversation paths
- Looking for a better quality or longer response
- Character behavior feels off
Regenerate Responses
Don't like a response? Generate a new one!
How to Regenerate:
- Click the đ Regenerate button on any character message
- The AI will create a completely new response
- Previous response is saved as a swipe option
- Regenerate as many times as you want
Regenerate vs Swipe:
- Swipe: Browse existing alternative responses
- Regenerate: Generate a brand new response
Edit Messages
Change any message in the conversation - yours or the character's.
Editing Your Messages:
- Click the âī¸ Edit button on your message
- Modify the text
- Hit Save
- The character will respond to your edited message
Editing Character Messages:
- Click Edit on any character response
- Adjust the text to your liking
- Continue the conversation from that point
- Useful for guiding character behavior
Why Edit?
- Fix typos in your own messages
- Clarify your intent
- Adjust character responses to maintain consistency
- Correct factual errors
- Steer the conversation in a specific direction
Conversation Branching
Every edit creates a new branch - explore multiple conversation paths!
How Branching Works:
- When you edit a message, you create an alternative timeline
- Each branch maintains its own history
- Switch between branches using the conversation tree
- All branches are saved automatically
Use Cases for Branching:
- Explore "what if" scenarios
- Try different conversation approaches
- Test character consistency
- Keep multiple storylines active
Message Management
Delete Messages
- Click the đī¸ Delete button on any message
- Deletes that message and all following messages
- Use carefully - deletion is permanent
Copy Messages
- Click the đ Copy button to copy message text
- Useful for saving favorite responses
- Share interesting conversations with others
Message Ratings
- Use đ đ to rate character responses
- Helps improve character quality
- Your feedback makes characters better over time
Chat Settings
Customize your chat experience with advanced settings.
Model Selection
Choose which AI model to use for this chat:
- GPT-5: Best overall quality and creativity
- Claude Opus 4.5: Excellent for detailed roleplay
- Gemini 2.5 Pro: Great context understanding
- See all 30+ models: Model List
Temperature (Creativity)
- 0.0-0.5: Focused, consistent, predictable
- 0.6-1.0: Balanced (default: 0.9)
- 1.1-2.0: Creative, varied, surprising
Higher temperature = more creative and random responses.
Context Length
- How many previous messages to include
- Short (10-20): Fast, recent context only
- Medium (30-50): Balanced (recommended)
- Long (60-100): Full conversation history
Longer context = better memory but slower generation.
Streaming
- On: See words appear in real-time (default)
- Off: Wait for complete response
Advanced Features
Stop Generation
Click âšī¸ Stop during generation to halt the response early. Useful if the response is going in the wrong direction.
Retry with Different Model
If a response isn't satisfactory, try switching models and regenerating. Different models have different strengths.
Export Chat
Export your conversation as:
- Plain text
- Markdown
- JSON (for backups)
Chat Best Practices
- Be Descriptive: "walks slowly toward the window, gazing out at the rain" beats "walks"
- Set the Scene: Include environmental details and context
- Use Actions: *asterisks* for actions add immersion
- Maintain Consistency: Remember character details they've mentioned
- Give Context: If jumping topics, provide transition
- Experiment with Temperature: Try different values for different conversation types
- Use Swipes: Generate multiple responses and pick the best