How to Train Character LoRA with 8GB VRAM
Dark_Sytze · reddit · 2026-07-19
An early Stable Diffusion user wants to get back into generative imaging, primarily to generate consistent appearances and styles for a specific game character while freely changing poses, clothing, and other details.
They note that they currently prefer the results from Krea 2, but the prompting experience differs from SD1.5/Illustrious. They are also asking if this character setup is suitable for training a LoRA, and what model to choose and realistic approach to take given their 8GB VRAM (3070) constraint.
More from Multimodal
- Magnific teases October 28 release: 'We're making it move' — cuenca · 2026-10-27
- Midjourney --sref demo: retro Americana Batman art series — gizakdag · 2026-10-27
- AI video is a new medium, not cinema or animation — and it's perfect for depicting dreams — Kyrannio · 2026-10-27
- Dad turns his toddler's 31-year-old quote into a song via SUNO kids' choir — NathanpmYoung · 2026-10-27
- Band's local-AI music video hits 200k views: 900 generations kept 64, 3-second clips beat face drift — Hwang0500 · 2026-10-27
- Adobe Firefly video editor walkthrough: one-sentence Quick Cut turns 32s into 20s — LudovicCreator · 2026-10-26