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.

YouTube

@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 3Dghostmannequin at the beginning of your prompt.

Diffusers (Python)

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)