This model was the result of experimentation with reproducing custom signatures. The LoRA is not very precise and can be hit or miss. It is best used to generate random cursive signatures either in bulk or single or multiple lines.

Experiment and try different strengths. 1.00 strength is good enough to generate signatures. Prompting can be either just the trigger word or short key words.

Testing Results:
- Wide Aspect Ratios work best for single line signature generations.
- Portrait A:Rs typically generate multiple signatures, one per line or even multi-column
- Can sometimes generate portraits with a signature beneath
- Output can sometimes generate noisy backgrounds. When this happens, change the sampler to "ddim" or "dpmpp_2m" combined with "sgm_uniform, "beta" or "simple" schedulers.

Example training text:
- A signature of Richard Dangardt, written in cursive, is displayed on a white background. The signature is in black and white, giving it a classic and elegant appearance. The cursive style of the signature adds a touch of sophistication and personalization to the image.
- A black and white logo of the name "Ben Tomas" written in cursive style. The letters are connected and flowing, giving a sense of elegance and sophistication. The font is large and bold, making the name stand out. The background is white, allowing the logo to be the focal point of the image.
- A signature, handwritten, cursive, elegant, in black ink, with the name "Charles Bukowski" written on a white background.
- YFG-Signatures (just rhe trigger word)
- YFG Signatures text, "John Smith"

Please post your resulting generations below and share your experience with the model.

Cheers!