Jun Park — Last Reader Standing | Character LoRA (SDXL 1.0)
A character-focused LoRA that recreates Jun Park, the main hero of the Last Reader Standing universe.
What it captures
Face & build – Korean male, mid-20s, messy jet-black hair, deep-brown eyes.
Signature outfit (
JunPark Outfit) – long navy trench-coat, black waist-coat, open white shirt, black compass-rose pendant.Casual looks (
JunPark Casual Outfit) – linen shirt, yukata, hooded cloak, etc.Key props – glowing antique grimoire, golden arcane tattoos/energy, spectral guardian spirits.
Trigger words
JunPark (LRS)
JunParkLRSAdd either (or both) at the start of your prompt.
Optional outfit tags: JunPark Outfit, JunPark Casual Outfit.
Recommended settings
Setting Value Base checkpoint SDXL 1.0 LoRA weight 0.7 – 0.9 CFG 5 – 7 Clip-skip 1 img2img denoise 0.55 – 0.65
Prompt examples
JunPark (LRS), JunPark Outfit, masterpiece, best quality, highly detailed,
korean man channeling golden arcane energy, glowing grimoire, dramatic library lightingJunPark (LRS), JunPark Casual Outfit, masterpiece, ultra-detailed,
sunset beach, linen shirt, calm expression, holding ancient book emitting lightTraining details
Dataset : 26 images @ 1024 × 1024 with minimal captions.
Steps / Epochs : 520 steps (10 epochs).
Network dim / alpha : 32 / 16 (~55 MB).
LR : 5 e-4 (UNet) / 5 e-5 (Text Encoder) — cosine scheduler, 3 cycles.
Other: min-SNR γ = 5, noise offset = 0.1, smart buckets ON.
License / usage
Released under Own Rights: free to use for personal and non-commercial artworks.
Commercial use or hateful / defaming content is strictly prohibited.
Have fun bringing Jun Park to life! 🚀
Description
Version 1.0 — Initial release (SDXL)
What’s new / changed
Fresh training from scratch on 26 high-resolution images (1024²) covering Jun’s signature outfit (≈ 80 %) and casual looks (≈ 20 %).
Base checkpoint: SDXL 1.0.
Kohya pipeline:
network dim 32 / alpha 16 (≈ 55 MB)
520 steps over 10 epochs
UNet LR 5 × 10⁻⁴ · Text Encoder LR 5 × 10⁻⁵
cosine LR scheduler (3 restarts) · min-SNR γ = 5 · noise offset = 0.1
smart buckets ON, clip-skip 1, batch 4
Trigger words:
JunPark (LRS)orJunParkLRSAdd
JunPark OutfitorJunPark Casual Outfitfor clothing control.
Recommended weight: 0.7 – 0.9 (txt2img) · denoise 0.55 – 0.65 (img2img).
Compatible with any SDXL-based checkpoint.
Known quirks
Very close-up portraits may soften pendant detail; add “sharp pendant” or lower CFG.
Extreme dynamic poses can introduce minor hand artifacts at strength > 0.9.
Roadmap
v1.1 (planned) — dim 64 / alpha 32 build with ~50 image dataset for sharper close-ups and broader lighting variety.
Enjoy bringing Jun Park to life—feedback is welcome! 🚀
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.