Developing this LORA wasn't an easy task. Initially wanted to use the datasets from existing artwork which slightly altered from Grok but seems failed to get the accurate output so I decided to regenerate the image from the existing Illustrious LORA and then ported it into here.

Works well with most of ANIMA models. Requires prompt with weights eg. (prompts:2) on both positive and negative prompts respectively to produce more accurate outputs. For further adjustments eg. showing the character doing stuff, please use Grok or Seedream Image Editor instead.

For prompt guides, see image samples here for prompt examples.

Generation settings

• Sampler: er_sde, gradient_estimation, Euler_Ancestral

• Scheduler: Simple

• Steps: 30

• CFG: 7.0