All models in SDXL version are presented here.

This is a character LoRA trained specifically to work seamlessly with the Z-Image-Turbo engine. It is optimized for ultra-fast generations, delivering sharp likeness, rich textures, and high responsiveness even at low step counts.

🛠️ Recommended Settings (Turbo Engine)

Since this is a Turbo-based architecture, please use the following baseline settings:

• Steps: 8-20

• CFG Scale: 1.0 - 3.0

• Sampler / Schedule Type: Use the recommended settings for your specific Z-Image-Turbo base model.

💡 Troubleshooting & Enhancing Likeness

⚠️ If the character likeness or details are not strong enough:

• Increase Generation Steps: Raise the step count up to 20-25. This gives the Turbo engine more room to bake in the character's specific features and refine details.

• Raise CFG Scale: Increase the CFG Scale higher than usual for Turbo models to give the LoRA more weight and strict adherence to the character profile.

🎨 How to Use

• Trigger Word: uvg5

• LoRA Weight: Recommended between 0.9 - 1.2 (adjust based on your prompt complexity).

All reference photos are artificially generated. Any resemblance to real people is coincidental.

If you liked the work, please like the model and share your work, it really helps in promotion.

Good luck to you generations!