This is a character SDXL 1.0 LoRA training model,it takes around 3700 total steps,8 epochs to complete finish,run pure python with kohya script on google Colab.

The training LoRA model location:https://huggingface.co/frank-chieng/michelleyeoh/tree/main/output

you can upscale in comfyUI with 4xUltraSharp upscaling model nodes,it will give more details on the character,i've updated the sdxl 1.0 base and refine model comfy workflow of Michelle Yeoh LoRA json file on huggingface link as below:https://huggingface.co/frank-chieng/michelleyeoh/blob/main/sdxl1.0base_lora_upscale.json