First version trained on almost 400 images.










This LoRa is trained on the cover designs of vintage matchbox art. I’ve always been drawn to their unique illustrations and typography. The dataset is a collection of carefully selected images, and I plan to do additional training as I discover more original designs.
Prompting isn’t difficult but benefits from detailed descriptions of the design elements to achieve better results. For example, describe the central figure, color scheme, ornaments, textures, and finally the text. You can check some of my example images for reference; of course, these prompts can still be refined further.
If you enjoy it, share your art and let me know what you think!
First version trained on almost 400 images.
1.転載モデルは学習・共有目的のみで使用し、著作権は原作者に帰属します
2.モデルの認証は公式チャンネルでご連絡ください。クリエイターの権利保護に努めています クリックして認証
