Downloads · 30 days
39
6% of all-time downloads
JunkieMonkey69/BarbaraPalvin_ZimageTurbo
BarbaraPalvin_ZimageTurbo is a text-to-image model from JunkieMonkey69. Use it when you need an image from a text prompt. It is set up for diffusers. The card lists the license as apache-2.0.
Downloads · 30 days
39
6% of all-time downloads
All-time downloads
632
Public
Repo size
185 MB
Likes
2
Public
Click a slice to open those files.
.safetensors170 MB · 92%
From the Hugging Face model README
🧩 LoRA Name: BarbaraPalvin_Actress
> Base Model: Tongyi-MAI / Z-Image-Turbo > LoRA Type: Character / Model > Trigger Word(s): Barbara Palvin > Recommended Weight: 0.6 – 0.85 (sweet spot ≈ 0.75)
✨ Overview
> This LoRA is trained to reproduce the facial identity, proportions, and high-fashion presence of Barbara Palvin, optimized specifically for ZimageTurbo. It emphasizes strong likeness fidelity while remaining flexible across editorial, paparazzi-style flash, lifestyle selfies, and cinematic fashion setups.
The model balances photorealistic skin rendering with glamour-oriented styling, making it well suited for magazine editorials, luxury fashion scenes, modern mirror selfies, and dramatic outdoor compositions.
🎨 What This LoRA Is Good At
> High likeness accuracy with consistent facial proportions > Editorial and high-fashion aesthetics > Realistic skin texture with visible pores and natural highlights > Strong performance with direct flash, paparazzi, and glam lighting > Works well in portrait, mid-shot, and full-body compositions
⚙️ Usage Notes
> Best results with portrait, editorial, and fashion-oriented framing > Prompting “Barbara Palvin” is sufficient; avoid stacking strong character LoRAs > Performs especially well with flash photography, cinematic lighting, and luxury styling > Guidance scale ≤ 1 is recommended for native ZimageTurbo behavior
🧪 Example Prompt Barbara Palvin, high-fashion editorial portrait, direct flash, cinematic lighting, realistic skin texture, confident pose
📐 Training Details
> Dataset Size: Captioned image set > Training Resolutions: 512 / 768 > Total Steps: 4,500 > Batch Size: 1 > Content vs Style: Balanced > Noise Scheduler: FlowMatch > Optimizer: AdamW 8-bit > Learning Rate: 1e-4 > Precision: BF16 > Text Encoder Training: Disabled > UNet Training: Enabled > Quantization: QFloat8 (model + text encoder) > Architecture: ZImage Turbo LoRA (LoKR full-rank enabled)
⚠️ Limitations / Notes
> Not intended for anime or heavily stylized cartoon outputs > Extreme facial distortion or exaggerated lenses may reduce likeness > Best used alone or with light stylistic LoRAs only
You should use Barbara Palvin to trigger the image generation.
Download them in the Files & versions tab.