I managed to make a lora that even the sloppiest images are soulful

I would gladly train a Qwen-Image version of this if someone points me to a flawless training tutorial specifically for Qwen