* Re-train version is trained on cleaner versions of the original lora's dataset images and finally removed the annoying halo/antennae. Works much better than V1.
<lora:Darla2:1>
* Version 1 works well but the re-train is a lot better.
<lora:DarlaD:1>
(Only for V1): Add "antennae" and/or "halo" to neg prompt if necessary.
Description
Re-trained on cleaned up versions of the original lora's dataset and removed the halo/antennae.
FAQ
Comments (2)
Can we please get your prompts from the sample images?
Sure! It's nothing special actually but here "score_9, score_8_up, score_7_up, BREAK Darla2, 1girl, blonde hair, blonde ringlets, blunt bangs, blue eyes, pink bow, shortsleeved pink dress, pink tutu, light pink socks, flat pink shoes, female, female focus, solo, solo focus, screenshot, official art, parody, cartoon footage, official design, screencap, cartoon screenshot, official screenshot, 2d, cartoon, source_cartoon, full body, smiling at viewer,"
It's my generic prompt for most example images I make.
As for the negative:
"score_4, score_5, score_6, logo, text, signature, watermark, anime, source_anime, closeup,"