All Collections
Style Studio: Create Art Styles
What If I Don't Have Enough Source Images to Create a Style?
What If I Don't Have Enough Source Images to Create a Style?

A guide to show how to multiply your training assets using Layer

Updated over a week ago

In order for AI to learn your art style effectively you need high quality input images. The best practice at Layer is to upload 25 images that are formatted the same way according to our image upload guidelines. However we understand you may not have 25 assets per use case to begin with as that would be a lot of assets.

  • In these cases we recommend creating styles by uploading the bear minimum of 6 assets and try forging new assets out of the style you created with 6 images.

  • As AI learning will be limited with 6 images only it's likely that you'll need to forge quite a bit to get an output that's decent enough to be put back in the training.

  • It's expected that you may even need to make some edits on these images to make them ready for training.

  • Once you are happy with the AI generated new input assets, you can put them back to training and re-generate your style.

  • You can repeat the process until you think your style is in good shape.

This is a tested and validated way of enriching your training set. One thing to really pay attention to is to make sure everything else about the images (format & captions) are perfect given you are working with a really small set of images. You can refer to the styles troubleshooting guide if you have any questions.

Did this answer your question?