This is my sixth style lora so go easy on me with this one.
Normally here I'd list the pros and cons between v1 and v2, but imo, v2 is a flat upgrade. I've remade the v1 sample images in v2 for comparison, plus a few bonus ones.
I was very specific about the Meesh style I was aiming for, which is 2017 and onwards. That's also why, regretfully, no characters are officially supported like my other style loras, as the v1 I had trained with had a 2x larger dataset but it started diluting the style. There are plenty of great Meesh character loras out that that can be used in combination with this lora if you want to nail characters.
All example images were created with only the trigger word "m33sh". Feel free to add "meesh" to enhance it a bit, especially when using a NAI model.
Description
Complete overhaul. Three times the dataset size, new tagging, and new training parameters, which include a significant increase in the number of training steps.

