Zeniji Mix QWEN: East Asian Beauty Generator
A Qwen-image model series for creating beautiful portraits with natural, contemporary aesthetics. Currently focused on East Asian features, with plans to expand globally.
About This Model Series
Designed to generate photorealistic portraits with a focus on cute, youthful beauty
Currently specialized in East Asian female portraits with clear skin and soft features
Works well for various scenes - from elegant studio shots to casual street photography
Future Development:
Expanding to diverse ethnicities:
More East Asian variations (Korean, Japanese, Chinese, etc.)
Latino/Latina features
European/Caucasian features
Middle Eastern/Arabic features
And more...
All versions will maintain a doll-like, East Asian aesthetic:
Soft, delicate features regardless of ethnicity
Youthful, cute appearance with clear skin
Refined beauty standards inspired by contemporary K-beauty and Asian aesthetics
Multiple versions with different faces and identities
Version: V1.0 - Fixed Identity (Experimental)
⚠️ This Version:
Experimental release - still testing and improving
Single fixed face - Always generates the same person
This happened during training (overfitting on a small dataset)
Great for character consistency, but can't change the face in this version
Built-in speed boost - Includes merged Lightning LoRA for fast generation (8-step)
SFW only - Does not include NSFW content
Recommended Settings
Quick Generation (4 steps):
CFG: 1.0
Steps: 4
Sampler: euler (or seed2)
Scheduler: simple, beta, or normalUpscale (2 steps):
CFG: 1.0
Steps: 2
Denoise: 0.25Total: 6 steps for a finished image - Very fast!
Best Use Cases for V1.0
✅ Good for:
Creating a consistent character across multiple images
Fast iteration and testing
Fashion editorials with the same model
Lifestyle photography series
⚠️ Limitations:
Always generates the same face
SFW content only
Tips
Perfect if you want the same character in different outfits/scenes
Try different prompts for clothing, poses, and backgrounds
Future versions will offer different faces and more variety
Description
Version: V1.0 - Hyper-Focused Identity
This V1.0 version is built for strong facial consistency and a highly reliable identity output.
Fixed Face Identity: This version is strongly trained to lock the face identity, meaning the model will consistently produce the same recognizable face.
Lightning Fast 8-Step Merged: This model includes a powerful Lightning (8-Step) LoRA merged directly into the file. This feature significantly speeds up the generation process while maintaining high visual fidelity.
Optimal Generation Workflow for V1.0:
To achieve sharp, fantastic images quickly, use this simple two-stage process:
Initial Fast Generation:
CFG Scale: 1.0
Steps: 4 Steps (Get your composition fast)
Sampler: euler (or recommend seed2 also)
Scheduler: simple, beta, normal (all works)
Upscale & Refine:
CFG Scale: 1.0
Steps: 2 Steps (Minimal steps for a crisp final look)
The Result: A sharp, awesome image with minimal effort.
FAQ
Looks like we don't have an active mirror for this file right now.
CivArchive is a community-maintained index — we catalog mirrors that volunteers upload to HuggingFace, torrents, and other public hosts. Looks like no one has uploaded a copy of this file yet.
Some files do get recovered over time through contributions. If you're looking for this one, feel free to ask in Discord, or help preserve it if you have a copy.