CivArchive
    Harrlogos XL - Finally, custom text generation in SD! - Harrlogos
    NSFW
    Preview 3165043
    Preview 3165042
    Preview 3165041
    Preview 3165044

    🚀HarrlogosXL - Bringing Custom Text Generation to SDXL!🚀

    Teaching Stable Diffsuion to spell, one LoRA at a time!

    Harrlogos is an SDXL LoRA trained on my own art and graphic design work, unlocking the ability to generate custom text.

    It does not require any control net, any img2img, or characters from fonts on the canvas.

    All you need is the LoRA, and you're ready to go!

    Versatile, Responsive, Powerful

    The model comes with a plethora of styles, modifiers, and extra content built in, but that's only where it begins.

    One of the greatest aspects of the model is how well and how seemlessly it works with almost any checkpoint, other LoRA, or even things like Stable Video Diffusion!

    Using Harrlogos:

    Harrlogos works with individual terms, separated by commas, in a specific order. You do not need all of these in every prompt, but this is the general format:

    1. YOURTEXT text logo

    2. Text Color: blue, teal, gold, rainbow, red, orange, white, cyan, purple, green, yellow, grey, silver, black

    3. Accent Color

    4. Background color

    5. Style Modifiers: dripping, colorful, graffiti, tattoo, anime, pixel art, 8-bit, 16-bit, 32-bit, metal, metallic, spikey, stone, splattered, comic book, 80s, neon, 3D

    6. Accent Modifiers: smoke, fire, flames, tentacles, hell, glow, horns, wings, halo, roots, embossed, blood, digital, ice, frozen, japanese, chrome, pastel, robotic, hearts, cute, egyptian, viking

    7. Additional Content: cat, sword, owl, cat ears, cthulhu, sun, roses, clouds, space, stars, skeletons, demons, fog, trees, moon, skulls, bones, planet, earth, cherry blossom, pentagram, crosses, lightning, bolts, crown, circle, moth

    Example of a prompt with the resulting image:

    HarroweD text logo, white, grey, red, spikey, splattered, dripping, blood, hell, crown

    Example Image




    Helpful Tips/FAQ:

    • Harrlogos is trained on a 1024x1024 data set, so it works best at resolutions: 1024x1024, 1024x768, 768x1024, 768x768

    • You will not get the correct spelling of every word, every time. Some text may take multiple tries to get right

    • The model works best with single words, multiple words are very achievable but it may be more difficult

    • Try weighting the text in the prompt, ex: ("Harrowed":1.5) text logo, red, black....

    • There's two schools of thought in achieving the perfect Harrlogo - A. Brute force spam through seeds B. Find a good starting point and lock the seed in, then slightly asjust the other paramters up or down, which will create slight changes to the image.

    • With a fixed seed, raising or lowering the CFG, Steps, or even resolution will modify the image, potentially fixing things or correcting spelling.

    • In the negative prompt, consider things like: Double letters, repeating letters, malformed letters, and more than X letters. If your renders keep coming out with a specific double letter, also try putting that in the negatives.

    This is a work in progress!

    Support us on Ko-fi

    Follow us on TikTok

    Follow us on Instagram

    Description

    Mix and match these keywords to create a logo -
    1. HarroweD text logo
    2. white, black, red, teal and cyan, purple, pink, grey, silver, metal, blood
    3. dripping, splattered, moon, skulls, demon, horns, pentagram, cross, hell, spiky

    FAQ

    Comments (13)

    ZactarOct 26, 2023
    CivitAI

    does it need controlnet?

    DazMakeArtOct 26, 2023· 2 reactions

    ControlNet will always be helpful in text-based generations but if you play around enough with this one, you'll find out that it doesn't require it. For example, try "zactar text logo, cyberpunk style, neon, blue, orange, stylized" Go a few runs with randomized seed and I'm sure you'll get a good result.

    HarroweD
    Author
    Oct 26, 2023· 3 reactions

    Nope, no control nets necessary!

    4tyzapOct 26, 2023· 1 reaction
    CivitAI

    dreamed about something for SD working similar with IDEOGRAM concerning to generation of a readable text without any controlnet. seems that this one makes a step in this direction. really cool if any.

    HarroweD
    Author
    Oct 26, 2023· 1 reaction

    Hey so this was honestly just an experiment, it's trained on only images that say "HarroweD, so honestly I'm incredibly impressed by the range it can handle already. The next version I'm working on right now has a much broader data set and is 4x larger, so hopefully that will be what you're looking for!

    4tyzapOct 27, 2023· 1 reaction

    @HarroweD just noted that yours is for SDXL. any chance for 1.5 how do you think?

    HarroweD
    Author
    Oct 27, 2023· 1 reaction

    @4tyzap As soon as the LoRA trainer comes back online I have a massive update coming for this one, but as far as a 1.5 version, I'd have to do some testing to see if it even works. SDXL handles text much better than 1.5

    AaliShineNov 5, 2023· 1 reaction

    @4tyzap nah! man this is totally crap shit i tried 15 time and all time result are shitt what i gave em texts always its not generate these text even single word always show wierd artifact type words not similar what i want plus its not even close to ideogram or dalle3 text generation which is sad from heart now deleting this crap if creator make it more cooler then ideo or dalle3 text generation in future then i will definitely redownload it also promote it too but still i cant see wholw civitia have that type of model -_- fusshh! why A1111 far backword then dalle3 text effect??
    Plus its not show into my A1111 untl i check LoRA incompatible shows all type check box why is my A1111 is old not have SDXL power how to make it ???

    herrXOct 26, 2023· 7 reactions
    CivitAI

    Works well. Depends a bit on the model. Had success with JuggernautXL. About 40% success rate with spelling. Also CFG-Scale has a big influence. Sweet spot for me was around 6. Less and the successrate goes down, more and the images miss dynamic. Works great with Img2Img (success rate 80% for me)

    HarroweD
    Author
    Oct 26, 2023· 2 reactions

    Hey thanks for the info, and for the rating, I really appreciate that! This LoRA was an experiment more or less to see if I could get SDXL to gen custom text, and now that I know, I'm training a MUCH more capable version of it, which will be up soon!

    marduk191Oct 27, 2023· 1 reaction

    @herrX Quality review with useable data. Thanks so much for the information. And certainly to @HarroweD for the lora :)

    RifzNov 5, 2023

    Hi HerrX could you give a bit more detail on how you use it with i2i? what sort of image do you start with, a black and white font image?

    simartem07Dec 22, 2023

    @herrX can you kindly put a link with a screenshot of basic usage in img2img ? Really appreciated!

    LORA
    SDXL 1.0

    Details

    Downloads
    631
    Platform
    CivitAI
    Platform Status
    Available
    Created
    10/25/2023
    Updated
    5/17/2026
    Deleted
    -
    Trigger Words:
    HarroweD text logo
    white
    black
    red
    teal
    cyan
    Purple
    Pink
    grey
    silver
    metal
    blood
    dripping
    splattered
    ice
    frozen
    moon
    skulls
    demon
    horns
    pentagram
    cross
    hell
    spiky

    Files

    198234_training_data.zip

    Mirrors

    CivitAI (1 mirrors)

    Harrlogos.safetensors

    Mirrors

    CivitAI (1 mirrors)

    Available On (1 platform)

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