Skip to main content

How to Improve AI Model Quality

Training your AI Character model with high-quality inputs is essential for achieving the best results. Follow these tips to ensure a realistic and accurate representation of your character:

Key Guidelines for Training​

  1. Upload Clear Photos of the Character

    • Use clear, high-resolution photos of the character's face.
    • For the best results, upload 16 photos showcasing the character from different angles and highlighting unique features.

  2. Avoid Group Photos

    • Ensure the photos focus on a single character.
    • Avoid including multiple people in the image to keep the subject clear.

  3. Minimize Obstructions

    • Avoid photos where the face is heavily obstructed (e.g., wearing sunglasses, hats).
    • Make sure the facial features are fully visible and prominent.

Training Options​

A. Train with a Single Image​

Quickly create a character with just one photo:

  • Ideal for Quick Results
    Use a single clear photo of the face or upper body that best represents your character.
  • Simple Setup
    Perfect for users seeking a basic virtual model with minimal effort.

B. Train with Multiple Images​

Create a detailed and realistic character by uploading multiple photos:

  • Recommended for Best Results
    Include a mix of:
    • Close-ups of the face
    • Full-body shots
    • Photos with varied backgrounds, angles, outfits, and expressions
  • Rich Detail
    Provides a more comprehensive and dynamic virtual character.

By following these steps and choosing the right training method, you can maximize the quality and realism of your AI Character in Lover Snap.