👕 FLUX.2-Klein-9B Ghost Mannequin
The ultimate LoRA for instantly creating professional 3D invisible mannequin effects for e-commerce apparel.
Say goodbye to expensive and time-consuming product photography editing. This specialized LoRA transforms flat or mannequin-worn clothing into clean, floating 3D presentations—the industry standard for high-end fashion e-commerce.
🚀 MASTER E-COMMERCE AI WORKFLOWS
Want to automate your product photography pipeline? Subscribe to the channel for upcoming in-depth tutorials on creating perfect ghost mannequin effects, ComfyUI batch processing, and building commercial AI pipelines for fashion brands.
@JettHuangAI — Hit 🔔 to level up your commercial AI workflows!
✨ Why Choose This LoRA?
- Production-Ready Quality: Trained on 50 diverse, meticulously curated pairs of apparel images to ensure versatility across different garment types and lighting conditions (6,000 training steps).
- Rank 256 for Ultimate Detail: Fashion photography requires perfect texture reproduction. Trained at an exceptionally high Rank 256, this model preserves the micro-details of fabrics, precise color accuracy, and crisp edges.
- Practical & Commercial Focus: Eliminates the artificial or flat look common in AI generation, delivering strictly high-end, studio-grade product visualization.
🖼️ Visual Showcase
(View the high-resolution samples below showcasing the clean 3D volume and texture retention).
| Product Sample 1 | Product Sample 2 |
|---|---|
| Product Sample 3 | Product Sample 4 |
|---|---|
🛠️ Usage & Cfg Settings
Trigger Word: 3Dghostmannequin
Simply use the trigger word in your prompt alongside standard high-quality product photography descriptors (e.g., 3Dghostmannequin, plain white background, studio lighting, detailed fabric...).
📋 Recommended Settings
1. Standard Base Model (High-Detail Inference)
- Inference Steps:
20 – 30 - Guidance Scale (CFG):
2.0 – 4.0 - LoRA Strength:
0.8 – 1.0(Adjust based on garment complexity)
2. Distilled / Lightning Model (Ultra-Fast Inference)
- Inference Steps:
4 steps(Highly optimized for rapid generation) - Guidance Scale (CFG):
1.0(Strictly required for distilled/schnell base models) - LoRA Strength:
0.85 – 1.0
💻 Integration Guide
ComfyUI Setup
(Sample custom workflow updating soon...)
- Download the model from the Files & versions tab.
- Place it in your
ComfyUI/models/loras/directory. - Add the LoraLoader node into your workflow and use the trigger word
3Dghostmannequinat the beginning of your prompt.
Diffusers (Python)
from diffusers import FluxPipeline
import torch
pipe = FluxPipeline.from_pretrained(
"black-forest-labs/FLUX.2-klein-base-9B",
torch_dtype=torch.bfloat16
).to("cuda")
pipe.load_lora_weights("nhathoangfoto/FLUX.2-klein-ghost-mannequin", adapter_name="ghostmannequin")
pipe.set_adapters(["ghostmannequin"], adapter_weights=[0.85])
# For Distilled Model usage:
# result = pipe(prompt="3Dghostmannequin, highly detailed...", num_inference_steps=4, guidance_scale=1.0)