Hair Style: Two face framing tendrils / sidebangs

Hello! This is my first TI, an attempt to teach the diffuser how to generate the [two?] face-framing tendrils hair style.

This is an early release that may not be followed up upon, depending on popularity/usage/issues.

Keyword for each individual file is the "Trigger word" in the file details on the upper right hand part of this screen.

Version 0.8

Just one file/version this time, seems to work okay on both realistic and anime checkpoints

See the "About This Version" section for details.

I've attached some sample images with simple example prompts, as well. Note that Anime images may require "bare forehead" in positive prompt and/or "(bangs)" in negative prompt if you don't want the "standard anime bangs" to accompany the sidebangs/tfft.

Has the best results when paired with some sort of "pulled back" hairstyle specified in the prompt (e.g. ponytail, braid, bun).

Known issues:

• Similar issues to version 0.7, but less severe.

  • Affects facial details and body shape of subject.

  • May still struggle with certain hair colors and/or non-caucasian women (e.g. black and indian women)

  • May affect coloration (preferring bland/desaturated colors)

  • May affect backgrounds (preferring to simplify backgrounds)

Version 0.7

"Photo" version is a more trained version of the "Illustration" inversion, and has a heavier application strength. If the photo version is used for illustration/anime-style checkpoints, it may start to override racial identity and facial features without further prompting. "Strong"/"Nike" version mightier than both; I selected the training strength that came JUST shy of starting to burn images on some models. About the heaviest application I can manage, for the truly stubborn checkpoints.

V07 Known issues:

• May require a higher-strength version and/or a higher prompt weight for better adherence, depending on the model.

• Strongly affects facial structure and body shape.

• Weights images towards NSFW, bland colors, and simple backgrounds.

• Still struggles with some hair colors and races, despite a more diverse training image set than version 0.5.

• If it applies too strongly, it can start to weight the hair color towards black or red. This should (should..) be fairly easy to override without much extra prompting.

If you have any issues or feedback, please comment below!

Huge thank you to BelieveDiffusion's textual inversion training tutorial located at https://github.com/BelieveDiffusion/tutorials/blob/main/consistent_character_embedding/README.md!

Thanks!