Norman Rockwell style LoRA for Pony V6 XL and Illustrious-XL v0.1.
For best results, prompt using Booru tags.
See training data for image list, tags and training config.
Description
The dataset was remade with a total of 197 images (see training data for image list). The only tag was "norman rockwell". A weight of 0.9 works well and upscaling gives better results. Only tested with base Illustrious-XL v0.1 model.
Most common bucket sizes: 960x1024 (44), 896x1088 (43), 896x1152 (40)
FAQ
Comments (10)
Out of interest, why did you do specific tags for the old version, and only "norman rockwell" as a tag for this one? Is that better?
No real reason. It was a test without detailed tagging that turned out well. Also, prompt context is limited, so fewer tokens wasted.
Edit: Tagging can damage the model's existing understanding of tags when finetuning. Additionally, it can cause the style only to trigger when certain common tags are present.
LORA made me want to create AI images again, beyond the interest in AI images that I had given up on for more than half a year and had never wanted to touch again. I would like to express my endless gratitude to the creator who created LoRA.
I'm so happy it inspired you! You made my day. I looked through your image gallery previously and liked your eye for detail and composition. As long as you're having fun, I'd love to see more in the future!
Thank you for creating a wonderful LORA.
Finally some naturalistic art. Gods save us from anime
Awesome style!
great style
Such a good lora to add to just about any gen. 10/10 A+
Wish there was a noob version .
Details
Files
Available On (4 platforms)
Same model published on other platforms. May have additional downloads or version variants.