CivArchive
    Lost Ark - South Vern Neria (Candaria Neria, 칸다리아 네리아) - v1.0
    NSFW
    Preview 140637180
    Preview 140637172
    Preview 140637179
    Preview 140637185
    Preview 140637184
    Preview 140637187
    Preview 140637189
    Preview 140637190
    Preview 140637191
    Preview 140637171

    South Vern Neria (Candaria Neria) from Lost Ark

    kanne_(la), 20s, mature female, large breasts, short hair, medium hair, curly hair, grey hair, blue eyes, hair ornament, fishnet, fishnet hair ornament, hair flower, see-through, cleavage, sideboob, sleeveless dress, lace trim, frills, black dress, short dress, fishnet thighhighs, black high heels,

    Description

    FAQ

    Comments (6)

    a22681cAug 24, 2026· 1 reaction
    CivitAI

    Are there any trigger words for it to replicate the in game style? Since your gallery images are all 2D anime based ones.

    VJS1
    Author
    Aug 24, 2026· 1 reaction

    This LoRA only contains character data, so you might need another LoRA to make in-game style.
    I recommend this LoRA, it worked decently when I used it.

    https://civitai.com/models/2541362/lost-ark-general-character-and-style

    VJS1
    Author
    Aug 24, 2026

    And you're right, I intentionally used 2D Anime style Checkpoint and LoRA.
    If you want to really imitate ingame LA style, you can check some other checkpoints as well.

    My training data only includes ingame screenshots, so it won't be hard to make ingame-like images if the setting is right

    a22681cAug 24, 2026

    @VJS1 I've tried for a long time in the past to replicate the ingame style of Lost Ark in private LORAs I've worked with before. Most of the time it sort of works but not entirely and it has a hard time with the anatomy and korean art style. I've used multiple checkpoints in both Pony, Illustrious, and even Noob. Let's just say it's tough to get to about 90% of a proper replication accuracy to the game's graphics, style, and anatomy.

    VJS1
    Author
    Aug 24, 2026· 1 reaction

    @a22681c I totally agree with you, since I also tried same as my own. In my experience, SDXL models could not perfectly imitate LA styles.

    There are many brand-new models such as Krea, Anima or Qwen, and I have high expectations and hopium for those models. It's a pity my GPU can't bear them and train LoRA on my own for those models.

    a22681cAug 24, 2026· 2 reactions

    @VJS1 I've been retraining my private SDXL loras onto Anima. Actually if you can make SDXL LORAs, then you can easily do the same with Anima. If anything since it's Qwen 3 based and a 2B model, it's got lower system requirements and you can crank out equal quality LORAs with same network dim/alpha in Anima with a smaller filesize.

    I did try resuming my lost ark based LORA work in Anima before Anima Base v1 came out, (preview 3 version), didn't work as good as my attempts on illustrious or pony or even NoobAI. Still, really great for 2D based stuff.

    LORA
    Illustrious
    by VJS1

    Details

    Downloads
    106
    Platform
    CivitAI
    Platform Status
    Available
    Created
    8/24/2026
    Updated
    10/11/2026
    Deleted
    -
    Trigger Words:
    kanne_(la)

    Files

    CandariaNeria.safetensors

    Mirrors

    CivitAI (1 mirrors)