Train

The Train tool in Amaro allows you to create custom AI models by training them on your own assets. This feature is ideal for generating images in a specific style or creating highly personalized outputs based on a subject.

How It Works

Training a custom model involves uploading a dataset of images, defining a trigger word, and fine-tuning complexity settings. The AI will learn from these inputs to generate high-quality, tailored outputs.

Steps to Train a Custom Model

1. Model Information

• Custom Model Name – Enter a unique name for your model as it will appear in your model library.
• Description – Provide a short description (max 100 characters) to help remember the purpose of your model.

2. Training Configuration

• Trigger Word – Choose a unique, non-real word that will activate your model’s concept in prompts (e.g., "IPHNE" or "CYBRPNK"). This word links your dataset to AI-generated outputs.
• Asset Complexity – Adjust the complexity slider based on how detailed the subject or style is.
  • Basic: Works well for simple subjects.
  • Complex: Required for intricate details like fine textures or text-based elements.

3. Upload & Caption Training Assets

• Upload between 12 and 20 images in JPEG or PNG format and caption them.
• Autocaption Images (optional):
  • Use the checkboxes to the left of the images to select images for autocaptioning.
  • Clicking Autocaption Images will automatically generate captions that include the trigger word.
  • Modify captions if needed for better accuracy (e.g., placing the trigger word directly before the subject descriptor for clarity).

4. Start Training

• Click Finish to begin training.
• Training takes approximately 20 minutes.

Using Your Trained Model

Once training is complete, the model will appear in your Custom Models library. You can now use it in Generate to generate new images. For details on how to generate images with your model, refer to the Generate Documentation.

Best Practices

• Choose high-quality images that represent your subject or style consistently.
• Use diverse angles and lighting to improve model performance.
• Refine autocaptions manually for optimal results.
• Keep the trigger word placement intentional in captions for better AI comprehension.

Training custom models in Amaro empowers you to create unique, style-specific content with AI, tailored exactly to your vision.