CivArchive
    Blowbang I2V - v1.0 HN
    NSFW

    A double handjob blowbang.

    Include the keyword bl0wb4ng in your prompt and wan will handle the rest.

    v1.1

    Tweaked a few knobs and pushed the training a bit further.

    🟢 Less hallucinations
    🟢 Better hand skills
    🔴 Slightly reduced facial accuracy

    Usage

    • Prompting: keep it simple; avoid NSFW vocabulary when possible.

    • Shift: keep it low. 7 is OK with most of schedulers and evenly split steps.

    • Cut: the cut happens at the 9th frame.

    • Lower Strengths (high-noise):

      • ~0.5: makes a progressive transition instead of a cut.

      • ~0.7: less crowded scene.

    Misc

    All the showcase videos have their workflow embedded.
    Tested only:

    • with lightx2v LoRAs enabled (1030 or 1022 on high-noise and 1022 on low-noise).

    • in ComfyUI.

    • on the following base wan models: Q4, Q5, Q6 and nvfp4.

    Training

    • v1.0: 67.5 epochs, 72 5-seconds clips (4860 steps).

    • v1.1: 80.0 epochs, 72 5-seconds clips (5760 steps).

    Description

    High-noise version.

    FAQ

    Comments (11)

    boobkake22Jun 14, 2026· 8 reactions
    CivitAI

    Buzz worthy! Been hoping a better one of these would show surface. Nice work.

    Edit: The results can be good - best in class, even. But it does have a lot of unpredictability. Edit works really well: diverse body types, ages, races. No clear face bias beyond normal i2v identity drift. Anatomy is decent, but can be helped by other LoRA's.

    However, it can also manifest:
    - Extra hands/penises/branching/snapping
    - Haze/smoke/particle efects/body horror

    It does seem sensitive to sampler configuration. The results seem to vary quite a bit based on strategy. This isn't unexpected, but more than ideal.

    qdr1en
    Author
    Jun 14, 2026

    Thank you. This one can still be a bit unstable sometimes though.

    boobkake22Jun 14, 2026

    @qdr1en Added some feedback. Still testing. (I've used your workflow, mine, and a future variant that uses a similar technique to yours. FP16 model.)

    qdr1en
    Author
    Jun 14, 2026

    @boobkake22 Thanks, that's helpful feedback.
    Counter-intuitively, from my tests, the shorter the prompt, the better the results.

    boobkake22Jun 14, 2026

    @qdr1en I've been trying to use things pretty close to your examples.

    ss9999Jun 27, 2026

    Which ones have you compared this one with?

    janssenbuttonJun 14, 2026
    CivitAI

    just curious. what is this terminate-node for? :o (works fine without)

    qdr1en
    Author
    Jun 14, 2026

    It allows to stop the ongoing generation without having to wait for the step to complete.

    janssenbuttonJun 14, 2026· 1 reaction

    @qdr1en interesting. thanks

    dailydoseofaiartJun 14, 2026· 1 reaction

    @qdr1en what comfyui version are you on? i thought that's integrated by now

    qdr1en
    Author
    Jun 16, 2026

    @dailydoseofaiart Looks like it's not needed anymore indeed

    LORA
    Wan Video 2.2 I2V-A14B

    Details

    Downloads
    560
    Platform
    CivitAI
    Platform Status
    Available
    Created
    6/14/2026
    Updated
    7/12/2026
    Deleted
    -
    Trigger Words:
    bl0wb4ng

    Files

    bl0wb4ng_HN_67.safetensors

    Mirrors

    HuggingFace (1 mirrors)
    CivitAI (1 mirrors)