Japanese HimeLook LoRA – Qwen Edition (v1.0)
Trained on 1024x1536 images to reproduce a refined, elegant “Hime-style” look.
Optimized for portraits and upper-body generations.
🔹 Trigger word: "hime"
🔹 Recommended weight: 0.7 – 1.0
⚠️ This character does not exist (fictional).
⚠️ Facial appearance may vary when combined with other LoRAs or workflows.
⚠️ Facial features are highly prompt-dependent. Please adjust via prompt (e.g., bangs, blush, lip gloss, gaze, angle).
🟢 Tested on Qwen with good results, but still in testing phase, so generations may fail at times.
🟢 In my case, Sampler = res_2s_ode, Scheduler = bong_tangent, Steps = 12 → higher chance of stable results. Please try adjusting within 8–12 steps.
⚠️ Even when using 8-step LoRA, Euler, Simple, or 8 steps may cause collapse (at least in my ComfyUI “nunchaku” build, reason unknown), so not recommended.
📂 Workflow settings are embedded in the sample images for reference.
❌ Selling or redistributing the LoRA file itself is prohibited.
Please note: The example prompt shown was originally written for Wan.
For best results, please use a Qwen-optimized prompt instead.