stabilityai/stable-diffusion · SD2.x LoRA Fine-Tuning Guide Request for Local Windows Laptop

Hey everyone!

I need some help with fine-tuning Stable Diffusion 2.x using LoRA on my local Windows laptop for a specific project. I know there’s a lot of debate around SD 2.x in terms of quality and usability, but I need to work with it since it’s a requirement for my project. If anyone has a step-by-step guide or resources that are up-to-date and work well in 2024, I’d really appreciate it!

Here’s my setup:

OS: Windows 11

RAM: 24 GB

Processor: Intel Core i5 (13th gen)

GPU: NVIDIA RTX 3050 (6 GB VRAM)

My goal is to generate high-quality, full-body fashion portraits, so any advice on dataset preparation, prompt tuning, and model setup would be incredibly helpful. I’ll be working in VS Code..

In captions csv, there are two columns: filename and caption. These are having the images name and their caption.

Thanks a ton in advance! Any help is appreciated 🙏