This is mostly Pony checkpoints and LoRAs with some Illustrious mixed in.

For positive prompts you can add: “masterpiece, best quality, amazing quality”.

You can also try: “ornate, detailed” in the positive.

For the negative try “bad quality, worst quality, worst detail, sketch, censor,”.

You can enter the score prompts: “score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up,” in the positive prompt, but it won't have much effect and will make it more likely that “score” will be written somewhere in the generated image. I leave it in because I think I can tell a difference.

To enhance image quality, try adding the Hyper-SDXL-8 Steps CFG LoRA at 50% strength and keeping the rest of the settings the same as below.

Resolution: at least 1024px x 1024px (or other supported SDXL resolutions).

Clip Skip: 2 or -2 depending on the software (“CLIP Set Last Layer” node in ComfyUI)

VAE: sdxl_vae.safetensors is baked in.

Sampler: DPM++ 2M or Euler A

Scheduler: Karras

Steps: 20 - 30

CFG: 6 or 7