What is style training?
Styles training is a process where an existing AI model, which has already been trained on a large dataset of images, is fine-tuned using a smaller, specific set of images. A dataset is a collection of images from which an AI learns a style. This helps the model to generate outputs that align more closely with a particular style, aesthetic, or characteristic visible in that smaller dataset.
Train your First Style
1. Gather Training Data
Gather 5-60 images of a specific style or object. The more images, the better.
Crop Images 1:1. We recommend this to avoid distorted images after training.
Avoid inappropriate images and follow our Content Policy
2. Upload your Dataset
After identifying your dataset, navigate to the 'Styles' tab, followed by 'My Styles,' and then 'Create Style.' Now you can upload your images, with the added convenience of being able to crop them using the built-in feature on starryaI! After uploading, you can assign a name to your style and select a relevant category to the dataset. Training a style uses 50 credits.
Here’s an example of an image that was created using this style
Prompt: Portrait of a handsome man
3. Start the training process
Typically, training requires 45 minutes. During peak server activity, this process may take up to 2 hours. On a rare occasion, if after 2 hours the style has failed to train, the credits are automatically refunded.
Once the style has been trained, you will receive a notification confirming its completion. After successful training, you can use the created style for your unique artistic endeavors and even share it with other users by publishing it for their use.