CivArchive
    Pokemon Sprite Animation Video LoRA - WAN 1.3B

    Pokémon Sprite Video LoRA

    This LoRA is trained on animated, Gen 5-style Pokémon sprites for video generation. It specializes in creating pixel-art-style motion and battle animations and is based on the Wan_t2v_1.3B model.

    Tips

    • Standard t2v: You can use this LoRA with standard T2V sampling. We recommend setting the output resolution to 768x768 for optimal results.

      • Note: The model does not use a trained text encoder, so it cannot understand Pokémon names. However, it can still generate pixel sprite animations of common creatures like cats, dogs, and dragons.

    • VACE i2v (recommend): The model able to use to create image-to-video animation by using VACE. The example workflow can be obtain in the sample image or you can find an example workflow here : Kijai's Example VACE Workflow

      • To get started, we recommend generating a Pokémon sprite image first and then feeding it into this model for animation. Sprite generation resources can be found here: NoobAI, SDXL, SD1

    Description

    This model is trained on WAN 1.3B

    FAQ

    Comments (6)

    vgb96May 19, 2025· 21 reactions
    CivitAI

    Yeah, what we need isn’t porn, but this kind of Lola. If you want porn, go watch it on Pornhub

    gambikules858May 22, 2025

    for what ?

    damiangJun 26, 2025· 4 reactions

    Pornhub is boring compared to the unique niche stuff you can do with AI

    DecodeMay 22, 2025· 1 reaction
    CivitAI

    very interesting! what kind of prompts can used to describe the animations?

    titansteng
    Author
    May 22, 2025

    A pixelate ....... pokemon on a white background.

    AndroidXLMay 25, 2025· 1 reaction
    CivitAI

    nicely done

    LORA
    Wan Video 1.3B t2v

    Details

    Downloads
    472
    Platform
    CivitAI
    Platform Status
    Available
    Created
    5/18/2025
    Updated
    5/20/2026
    Deleted
    -

    Files

    pkspani_w13B_v1.safetensors

    Mirrors

    HuggingFace (1 mirrors)
    TensorFiles (1 mirrors)