We've been refining our digital art style through careful prompt crafting, exploring diverse illustration methods, and iterative image refinement. This ongoing journey draws inspiration from various artists and integrates elements of our unique artistic vision.
In our experimentation with illustration styles, we've explored techniques like training a Latent Diffusion model (like LoRA) on our own generated images. This involved meticulously crafting prompts to achieve a desired aesthetic, then curating the results to further refine the style. By incorporating artistic influences and our own creative approach, we were able to achieve a unique artistic direction.
Description
Training Setting: This version was trained under a different setting specifically designed to produce more realistic illustrations.
Impact: This setting allows the model to generate faces that appear closer to real photographs compared to previous versions.
Application: This version is particularly suitable for tasks requiring photo illustration character creation with abstract background