CivArchive
    The Redheaded Libertarian - 1.0
    NSFW
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined

    This TI attempts to capture the likeness of "The Redheaded Libertarian" on Twitter. TRHL is an artist, model, and political commentator recognized for her intelligence, sharp wit, eponymous red hair, pronounced cheekbones, green eyes (central heterochromia) and seemingly age-defying youthful appearance.

    Used with the SD1.5 base model, it will create a reasonable likeness of the source. Used with any other model, it will add "TRHL" features, typically pronounced cheekbones and a demure look.

    The TI was using A1111 with 3400 steps.

    Please note! Only the base SD1.5 model seems to capture the likeness. If you're using a different model, you may want to inpaint using the base SD1.5 model if reproducing the general likeness is important.

    For whatever reason, using the base SD1.5 model, this TI wants to make the persona older by adding excessive crow's feet around the eyes. You'll need to negative prompt wrinkles, old, elderly, etc. to get the image closer to real life. If using another model (e.g. Deliberate) you can remove these negative prompts, as the TI will typically produce a younger female.

    Please note, most of these images have had some level of inpainting or face restoration. The base prompts will get you 90% of the way there, but you'll need to work a bit to get better hair, add freckles, etc.

    The initial images are a combination of SD1.5 base for the likeness with Deliberate for the details. The last few images were rendered with different models, and you can see what a huge difference that produces in the output. The models used are noted in the prompt details.

    Prompts are formatted for Invoke. If you're using A1111, you'll need to modify accordingly.

    Description

    Trained using SD1.5 with A1111. 3400 steps. This TI was trained on images from the TRHL Twitter feed, mainly selfies. It lends itself to these same sorts of images. You'll get the most realistic results using the base SD1.5 model for the face. I used Deliberate to inpaint and get better hair and face restoration to get the eyes to pop.

    FAQ

    Comments (4)

    thefoodmageFeb 27, 2023· 2 reactions
    CivitAI

    Oh gosh, I wonder how this is going to go down lol

    SteveWarner
    Author
    Feb 27, 2023

    Hopefully it will interesting. There are numerous TI's on this site of real people. Erin Moriarty, Orlando Bloom, Robert Downy Jr., Brie Larson, even Ivanka Trump. It raises an interesting question about the right to control one's own likeness, especially in the social media age where the cult of personality drives people to post pictures of themselves with reckless abandon in the pursuit of likes and subscribers. These images are being uploaded to the servers of big tech sites. Do we know what they do with that data? Every photo I post to Facebook or upload to Google is tagged with my facial attributes. What does posting images to public forums do to our privacy? Does it further it or does it add one more brick to the wall of control that's slowly being built around us? No one raises and eyebrow when a TI of Anne Hathaway goes up. Are celebrities images fair game, but social media influencers and private citizens not? Where do we draw the line? So yeah. Hopefully it will be interesting, as the tech is developing at a faster pace than the ethics around it.

    thefoodmageFeb 27, 2023· 3 reactions

    @SteveWarner This is why I have not uploaded my face to the internet for literal years. People are weird! Funny because I feel like every woman uploaded on here will be sued in URPM or other models quite soon after they're uploaded. That seems to be what 75%+ of users do with celebs and AI.

    SteveWarner
    Author
    Feb 27, 2023

    @TheFoodMage That's smart of you. Those that post images of themselves tend to do so for the dopamine hit that comes from getting likes on social media, but it comes with a price.

    The development of this TI was fairly innocuous (if you can believe that). If you look at my other TIs, they failed to truly capture the subject, one being a style (80's Memphis) and the other being an object (F-15 Fighter Jet). I figured I might be doing something wrong in the training, so I tried training a person (TRHL) and it was an absolute failure, which told me that my process was indeed broken. I had to re-learn how to mkae a TI. This iteration represents the first success at training a TI.

    TRHL has really strong feminine features, so adding this embed to any prompt is going to push any image in that direction. I've added a few images using different models to show how it renders for those not going after a specific likeness.

    TextualInversion
    SD 1.5

    Details

    Downloads
    463
    Platform
    CivitAI
    Platform Status
    Deleted
    Created
    2/27/2023
    Updated
    5/13/2026
    Deleted
    5/23/2025
    Trigger Words:
    rhlib-3400

    Files

    rhlib-3400.pt

    Mirrors

    HuggingFace (1 mirrors)
    CivitAI (1 mirrors)

    Available On (1 platform)

    Same model published on other platforms. May have additional downloads or version variants.