CivArchive
    POV Lap Pillow XL - V1
    NSFW
    Preview 19599060
    Preview 19599705
    Preview 19600022
    Preview 19600317
    Preview 19600707
    Preview 19601068

    Generates a POV pose with the girl looking down at the viewer from close range, as if the viewer was lying with their head on her lap. Designed for anime-style images and trained on PonyXL.

    Trigger words: pov lap pillow
    Suggested weight: ~1
    Other suggested prompt words: ceiling, sky, etc
    Potential negative prompt words: legs, thighs, etc

    Suggestions for achieving the right look:
    1. Prompt for the ceiling, sky, or whatever the viewer can see beyond the girl's head in order to help the AI figure out the perspective.
    2. Generate in square or landscape orientation, or add a negative prompt for things like legs or thighs, in order to discourage the AI from showing too much of the girl's legs, ruining the illusion.

    Tested on Hassaku XL Hentai v1.3.

    Description

    FAQ

    Comments (5)

    ____NULL____Jul 16, 2024· 4 reactions
    CivitAI

    lovely pov concept ❤️

    LonelydetJul 22, 2024
    CivitAI

    Don't know why, but I feel like most of the images I generate look more like they're lying in front of that person rather than the perspective of lying on their legs.Unable to achieve the effect shown in sample images 2 and 3. Any suggestions?(ChatGPT translate)

    alxpnt2
    Author
    Jul 22, 2024

    It can be a bit difficult to get the look right because the AI doesn't really understand what's happening in the pose, it just knows how to approximate the from-below perspective. Here are some suggestions that might help:
    1. Prompt for the ceiling, the sky, or whatever the viewer can see above the girl's head. This helps it understand how low the perspective should be.
    2. I suggest using either a square or landscape orientation. This discourages the AI from showing too much of the girl's legs, which ruins the illusion.
    I added these suggestions to the description.

    LonelydetJul 23, 2024· 1 reaction

    @alxpnt2 I conducted a rough test and found that when adding additional suggestion prompts and negative prompts, the probability of obtaining the correct perspective increased to 5/10. When the quality prompts from the sample images were added, the probability increased to 9/10 (only tested with 100 images, so it cannot be guaranteed to be completely accurate).

    alxpnt2
    Author
    Jul 23, 2024

    @Lonelydet Thanks for testing so thoroughly, I'm glad you were able to find a solution.

    LORA
    Pony

    Details

    Downloads
    2,022
    Platform
    CivitAI
    Platform Status
    Available
    Created
    7/1/2024
    Updated
    6/22/2026
    Deleted
    -
    Trigger Words:
    pov lap pillow
    ceiling, from below

    Files

    POV_Lap_Pillow_XL.safetensors

    Mirrors

    HuggingFace (1 mirrors)
    TensorArt (1 mirrors)
    TensorHub (1 mirrors)

    Available On (1 platform)

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