Juggernaut Z Detailer & Style Enhancer (SVD Extracted LoRA)
Welcome! This LoRA is mathematically extracted from Juggernaut Z V1 by RunDiffusion using the low-rank SVD (Singular Value Decomposition) method against the original Z-Image base model.
It is designed primarily as a lightweight detailer, texture enhancer, and style injector for Z-Image Turbo by Tongyi. If you want to bring the rich textures, depth, and creative variety of Juggernaut into the lightning-fast Z-Image Turbo model, this is exactly what you need!
🌟 Features & What It Does
No Trigger Words Required: Works seamlessly with your existing prompts.
Detail & Texture Booster: Significantly enhances skin textures, fabric details, backgrounds, and overall image sharpness.
Stylistic Variety: Injects the signature Juggernaut aesthetic into the fast-paced Z-Image Turbo model.
💾 Choose Your Variant (4 Versions Available)
We are providing four distinct versions so you can choose the perfect fit for your workflow and hardware:
1. Rank 32 / Alpha 16 — The Reference Standard
* Uses a balanced `0.5` scaling ratio. This matches the original model spec exactly and provides a subtle, highly stable style enhancement.
2. Rank 32 / Alpha 32 — The Lightweight Puncher
* Uses a full `1.0` scaling ratio. Compact file size with a more prominent style and detail boost. Great for fast setups.
3. Rank 64 / Alpha 32 — The Balanced Choice
* Offers double the detail capacity (Rank 64) with a smooth `0.5` scaling ratio. Highly recommended for complex prompts and maximum stability.
4. Rank 64 / Alpha 64 — The Maximum Detailer
* Large parameter capacity applied with a full `1.0` scaling ratio. Perfect if you want to push details, textures, and Juggernaut's aesthetic to the absolute limit.
🔧 Recommended Settings for Z-Image Turbo
Because Z-Image Turbo operates under very low step counts and fast sampling, we recommend driving this LoRA a bit harder to fully bring out its detailing effect:
LoRA Weight (Strength): 1.0 to 2.0 (Adjust within this range to find your sweet spot. Start at 1.0 and increase for more intense details and style variety).
Sampling Steps: Use the standard settings recommended for your Z-Image Turbo workflow (typically 4–8 steps depending on the sampler).
CFG Scale: Follow typical Turbo settings (generally 1.0 to 2.0).
Note: For non-turbo Z-Image base models, you may want to start at a lower weight of 0.5 to 1.0.
💬 Share Your Creations!
If you find this LoRA helpful, please consider leaving a review and posting your generated images in the gallery below! Seeing your creations is the best part of sharing these models.
🤝 Credits & Acknowledgments
Tongyi (MAI): For the groundbreaking Z-Image and Z-Image-Turbo base architectures.
RunDiffusion: For the Juggernaut Z V1 model, from which these stylistic details were extracted.