CivArchive
    DC 1980's Sword and Sorcery Movie - v1.0 [FluxKleinBase9B]
    NSFW
    Preview 121239987
    Preview 121239990
    Preview 121239978
    Preview 121239981
    Preview 121239991
    Preview 121239986
    Preview 121240006
    Preview 121239980
    Preview 121239982
    Preview 121239994
    Preview 121239995
    Preview 121239985
    Preview 121239976
    Preview 121260338
    Preview 121260367
    Preview 121260366
    Preview 121260370

    THIS LORA WAS NOT TRAINED TO CAPTURE THE FACE OF ANYONE IN PARTICULAR ONLY TO GET THE 1980's SWORD and SORCERY MOVIE STYLE and AESTHETIC

    Trigger Word DCSWORDNSOR.

    Big Negative - Faces are absolute dogshit sometimes i will try sort out and prune the dataset a bit better and maybe change training parameters as i only used standard ostris ai-toolkit settings - If you have any idea what i could do to correct the faces in groups of people feel free to leave a comment with suggestions i would really appreciate it, Thankyou.

    This lora was trained with Ostris Ai-Toolkit with Runpod on 17 different 80's movies screenshots mainly Conan the Babarian - Other films include Xena, Beastmaster, Ator, Clash of the Titans, Deathstalker, DragonSlayer, Excalibur, Sinbad, Hercules (not really 80's), Ironmaster, Kull, Red Sonja, The Barbarians, The sword and the sorcerer, Willow and Zardos. I Also added a few of my high quality ai generated images + A couple of Images of Lawrence of Arabia and Kingdom of Heaven as it matches the aesthetic i was going for.

    Trained with Ostris AI-Toolkit for 10,0000 steps Rank 32 Tagged with detailed captions consisting of 100-150 words with Gemini3FlashPreview (233 Images Total)

    All images have Prompts, generation details Workflow for ComfyUi just click the image you want, then save, then drop into ComfyUI or Open the image with notepad on pc and you can search all the metadata there. My workflow has multiple Upscalers to choose from Seedvr2, Flash VSR, SDXL TILED CONTROLNET, Ultimate SD Upscale and a DetailDaemon Upscaler and an Qwen 3 llm to describe images if needed

    If you use this in a merge, to train a model or upload to another site generator please credit me by directing to my civitai page https://civarchive.com/user/dkpc69

    I write all my training details in post in the hope it helps someone learn roughly what i am doing for training my aesthetic style lora's, That is why you'll see different settings here and there with my uploaded lora's, For people wanting to start training it can be a good rough guide but I am still learning everyday, So don't take my settings as bible as i hardly know what i am up to either lol.

    http://Ko-fi.com/dkpc69

    Description

    FAQ

    LORA
    Flux.2 Klein 9B-base

    Details

    Downloads
    329
    Platform
    CivitAI
    Platform Status
    Available
    Created
    2/16/2026
    Updated
    4/27/2026
    Deleted
    -
    Trigger Words:
    DCSWORDNSOR.