Character LoRA trained on 21 images fails across models—training setup questioned
IntrovertedOtaku03 · reddit · 2026-10-24
A Reddit user training a character LoRA with Kohyass 26.0.0 on Wai Illustrious SDXL 1.60 (21 images, 2500+ steps) reports the LoRA failing on some models, losing hair/clothing features, and producing NSFW outputs despite negative prompts. They ask for advice on base model choice and training config.
More from Multimodal
- Stunning AI-Generated Sci-Fi Scene: Futuristic Car Rides Toward a Floating Tower — ciguleva · 2026-10-24
- Community fine-tune fixes Qwen-Image 2.1 VAE 8-pixel grid artifact — Crazy-Repeat-2006 · 2026-10-24
- Tencent's Hy-Nomas video world model ported to native ComfyUI, runs on 12GB VRAM — LatentSpacer · 2026-10-24
- Video models replace NeRF/3DGS pipelines: single-frame novel view synthesis, open-source soon — jonstephens85 · 2026-10-24
- Recreating PS2-Era Harry Potter With FastH3-V2 in 4 Steps — aziib · 2026-10-24
- A Full AI Music Video Made Locally on DGX Spark With MiniMax H3 — Ornery-Block-3522 · 2026-10-24