This model is designed to recreate the iconic Elemental-Bending style of "Avatar: The Last Airbender" and "The Legend of Korra." It captures the unique visual elements and dynamic action scenes characteristic of these animated series, making it ideal for generating images that evoke the spirit of elemental bending. Avatar-Mode is the newest addition and will stay exclusive to Flux. It is trained on all 4 Datasets at once and can generate multiple elements in the same picture.

Base Model: Flux/SDXL/Pony

Recommended Settings:

• For Flux:

  • Steps: 20 - 30

  • CFG: 2.0 - 3.5

• For SDXL/Pony:

  • Steps: 30 - 60

  • CFG: 5.0 - 7.0

Additional Information: The model was trained on a dataset of 25-35 images, which were generated using Dalle3 and then upscaled with Krea AI. Therefore the dataset is completely synthetic and can only resemble the characteristics of the bending, while not resemble the nations etc correctly.

Note: For best results, use the recommended settings for each base model. The model performs particularly well in generating dynamic action scenes.