CivArchive
    PornMaster-wan 2.2-14b-I2V-bukkake - v1.4_high_noise
    NSFW

    通義萬相AI生影片—使用指南-中文

    --

    PornMaster-wan 2.2_14b_I2V_69 position blowjob

    這個lora的用途是方便讓你將她的臉添加精液，除了精液以外的功能都是WAN 2.2中內建。

    此lora使用RTX PRO 6000 WK培訓，訓練用的影片皆為4K的素材，所有影片皆是針對臉部特寫的精液做訓練，具有高品質的精液與詳細的皮膚紋理。

    此lora經過多次的失敗和測試後，終於能夠完美的保有臉部的一致性。

    注意事項:

    如果使用低細節和皮膚過度蒼白，或是過度曝光的照片，生成的影片將看不到臉部的精液。當使用高清照片時，生成的精液品質將會非常好。

    由於只針對臉部訓練，越遠的距離精液的成功率將會降低。

    建議先生成臉部特寫影片後，再接著製作鏡頭拉遠自全身的影片。

    影片訓練長度為4秒，每秒16幀。

    低噪聲和高躁聲lora必須一起使用。範例影片使用官方標準工作流。

    需簡單描述背景，這樣影片轉換場景時會更加穩定。

    --

    The purpose of this lora is to allow you to add semen to her face. Except for the semen, all other functions are built-in in WAN 2.2.

    This LoRa was trained using an RTX Pro 6000 WK GPU. All training videos are 4K footage, focusing on close-ups of semen on faces, with high-quality semen and detailed skin textures.

    After numerous trials and errors, this LoRa was finally able to perfectly capture facial consistency.

    --

    Note:

    If you use photos with low detail, excessively pale skin, or overexposed skin, the resulting video will not show semen on the face.

    When using high definition photos, the quality of the semen produced will be very good.

    Since it is trained only on faces, the success rate will decrease with further distances.

    It is recommended to first create a close-up video of the face, and then move on to create a video with the camera zooming out to the whole body.

    The training video length is 4 seconds and 16 frames per second.

    Low-noise and high-noise LoRa must be used together.

    The example videos use the official standard workflow.

    The background needs to be described briefly so that the film will be more stable when changing scenes.

    --

    觸發詞(trigger words):

    pornmaster bukkake, a woman, after a completely black screen, the camera quickly switches and excessive sperm on her face, excessive sperm on her hair,

    --

    使用最新的加速器，可以讓生成的影片更穩定。

    Using the latest accelerator can make the generated video more stable.

    Wan_2_2_I2V_A14B_HIGH_lightx2v_4step_lora_v1030_rank_64_bf16

    wan2.2_i2v_A14b_low_noise_lora_rank64_lightx2v_4step_1022

    --

    我的其他檢查站和Lora。

    My other checkpoints and Lora.

    --

    如果你喜歡我的lora，可以用"buzz"鼓勵我。

    If you like my lora, you can encourage me with "buzz".

    --

    客製化lora:我的Telegram ID:@iamddtla0620

    我的微信ID:iamddtla

    我的line ID:iamddtla

    時間寶貴，請直接說明來意，例如:訂製lora.

    Customized lora: My Telegram ID: @iamddtla0620

    My WeChat ID: iamddtla

    My Line ID: iamddtla

    Time is precious, please state your intention directly, for example: customized LORA.

    請勿將此lora用於違法用途或是真實存在的人物。

    Please do not use this lora for illegal purposes or for real people.

    Description

    低噪聲和高躁聲lora必須一起使用。範例影片使用官方標準工作流。

    Low-noise and high-noise LoRa must be used together.

    The example videos use the official standard workflow.

    --

    觸發詞(trigger words):

    pornmaster bukkake, a woman, after a completely black screen, the camera quickly switches and excessive sperm on her face, excessive sperm on her hair,

    --

    需簡單描述背景，這樣影片轉換場景時會更加穩定。

    The background needs to be described briefly so that the film will be more stable when changing scenes.

    --

    使用最新的加速器，可以讓生成的影片更穩定。

    Using the latest accelerator can make the generated video more stable.

    Wan_2_2_I2V_A14B_HIGH_lightx2v_4step_lora_v1030_rank_64_bf16

    wan2.2_i2v_A14b_high_noise_lora_rank64_lightx2v_4step_1022

    wan2.2_i2v_A14b_low_noise_lora_rank64_lightx2v_4step_1022

    FAQ

    Comments (28)

    Rapidfire_sukNov 8, 2025· 1 reaction
    CivitAI

    Most underrated wan model. Hopefully now that it's available for free we will see more generations.

    iamddtla
    Author
    Nov 8, 2025

    I personally find this to be an interesting lora; you can add anything you want after the prompt "after a completely black screen."

    StawyNov 8, 2025· 9 reactions
    CivitAI

    It works much better without the black transition. Simply remove "after a completely black screen" from the prompt.

    iamddtla
    Author
    Nov 8, 2025

    Do you have a sample of the generated video?

    iamddtla
    Author
    Nov 8, 2025· 1 reaction

    Oh~ After testing, I only need to fill in the prompt "the camera quickly switches and excessive sperm on her face" to add other themes afterward, and the black screen can be omitted.

    fenasikerimNov 8, 2025

    @iamddtla Doesn't work for me. Which base model and step lora do you use? I get no scene cut with "the camera quickly switches and excessive sperm on her face", the character just makes random moves

    iamddtla
    Author
    Nov 8, 2025

    @fenasikerim I used wan2.2_i2v_14b_fp8_scaled to generate the video.

    fenasikerimNov 8, 2025

    @iamddtla Interesting

    blaNov 8, 2025· 1 reaction
    CivitAI

    Keep up the good work!

    MikeflowerNov 8, 2025· 1 reaction
    CivitAI

    Great work as always👍🏻

    CharlieBrown0115Nov 9, 2025
    CivitAI

    why am i getting every single time a video where the camera zooms in to the characters face??? i already tried prompting fixed point of view, static shot etc, and it doesn't work

    iamddtla
    Author
    Nov 9, 2025· 1 reaction

    You should post the prompts you wrote so that others can help you check them.

    CharlieBrown0115Nov 9, 2025

    @iamddtla no camera movement, static shot, fixed point of view

    pornmaster bukkake, a woman, the camera quickly switches ,and the woman is in the same position , excessive sperm on her face, excessive sperm on her hair, excessive sperm on her tongue, excessive sperm on her breasts, She looks at the viewer

    CharlieBrown0115Nov 9, 2025

    no camera movement, static shot, fixed point of view

    pornmaster bukkake, a woman, the camera quickly switches ,and the woman is in the same position , excessive sperm on her face, excessive sperm on her hair, excessive sperm on her tongue, excessive sperm on her breasts, She looks at the viewer

    iamddtla
    Author
    Nov 9, 2025

    @CharlieBrown0115 

    This is your problem: [no camera movement, static shot, fixed point of view, and the woman is in the same position]. You wrote these words, but of course she won't change anything. You can refer to the prompts in all my examples.

    CharlieBrown0115Nov 9, 2025· 1 reaction

    @iamddtla @iamddtla I wrote those prompts that way because I copied exactly the prompts from your examples and I was only getting videos with that annoying zoom too close to the person’s face.

    SlowmoeNov 9, 2025

    @CharlieBrown0115 The lora is most likely trained specifically on close-ups. You could try to lower the strength but then you might not get the "result" you want.

    iamddtla
    Author
    Nov 9, 2025

    @CharlieBrown0115 1. This LoRa file is primarily for close-up facial shots; you can use it in conjunction with my other LoRa files.

    2. Please read the instructions carefully.

    3. Didn't you see the other examples I created demonstrating how to move the camera?

    NEOSAMANov 12, 2025
    CivitAI

    This transition is like something out of a horror movie.

    axymeus991Nov 12, 2025
    CivitAI

    Was looking forward to it but... very random results. Often times the transition happens twice, or the cum is poured over like milk, or it splashes suddenly after the transition, plus other wonky stuff happening. Once in a while it works great but it takes a lot of effort.

    iamddtla
    Author
    Nov 13, 2025

    Are you sure you've read the instructions carefully? Have you referred to the example I created in detail?

    axymeus991Nov 13, 2025

    @iamddtla Your prompt actually triggers an odd effect where after the black screen, the lightning flickers to black once more. Removing the "black screen" mentions fixes that but not the rest. Like, if you prompt for something other than a face close-up you will get a transition to a close-up then a second transition to your desired shot.

    I'm using the same setup as you in regard to lightning loras but tried different approaches (incl. full 720p, no lightning). The cum still looks too liquid most of the times, but the weird pouring animation is really what breaks the fidelity.

    EDIT: Even in your own examples the issue is visible, half the times the cum is dripping immediately after the transition in a very liquid state, like it was just poured. Would expect something more static, like a "before-and-after" kind of results.

    axymeus991Nov 13, 2025

    Actually ran a few extra tests and might have found something. There are many versions of the lightning loras, but it seems the "MoE" version distilled by Kijai is the one that works best in conjunction with your work. The newer versions like 1022 and 1030 are giving me worse results on average.

    CharlieBrown0115Nov 14, 2025

    i know !!! same happens to me with this lora! :(

    CharlieBrown0115Nov 14, 2025

    @axymeus991 you mind posting the links of the lighting loras you are using

    axymeus991Nov 20, 2025· 2 reactions

    @CharlieBrown0115 sorry for some reason I don't get notified of comments. Now that I have more experience with this lora, I'm finally getting some great results relatively consistently, so here's what works for me. I'm rendering at 720p using the Q8 GGUF, but lower resolutions seem to work fine so you can adjust for less VRAM.

    I have Kijai's distill of the MoE lightning lora at 1.0 strength (https://huggingface.co/Kijai/WanVideo_comfy/blob/main/LoRAs/Wan22_Lightx2v/Wan_2_2_I2V_A14B_HIGH_lightx2v_MoE_distill_lora_rank_64_bf16.safetensors) as well as the old 2.1 rank 128 at 1.5 strength (https://huggingface.co/Kijai/WanVideo_comfy/blob/main/Lightx2v/lightx2v_I2V_14B_480p_cfg_step_distill_rank128_bf16.safetensors) although you can sometimes bypass that one (overall it improves the motion without much visible loss).

    For the low noise model you can use any lightning loras since they're all the same actually. I personally use the lightx2v1, I'm not even sure where it is on the repo lol, the first version of the 2.2 lightning pretty much. I also add the same rank128 one as above at 0.25 strength.

    5 steps, 2 high + 3 low. Euler+Beta. The following prompt structure has given me the best results so far: "The scene suddenly and immediately cuts to a black screen. The same girl is now covered in cum after an overhelming bukkake. Her face, hair, and body are covered in cum. Before-and-after. Same place, same person, same clothes."

    zakakkiNov 17, 2025· 3 reactions
    CivitAI

    The best cum lora of all.

    SawlikeNov 17, 2025· 10 reactions
    CivitAI

    This is more a jumpscare lora

    LORA
    Wan Video 2.2 I2V-A14B
    by iamddtla
    Download (Beta) View on CivitAI
    concept
    facial
    bukkake
    cum
    cum on face
    cum on hair

    Details

    Downloads
    7,191
    Platform
    CivitAI
    Platform Status
    Available
    Created
    10/24/2025
    Updated
    6/5/2026
    Deleted
    -
    Trigger Words:
    pornmaster bukkake, a woman, after a completely black screen, the camera quickly switches and excessive sperm on her face, excessive sperm on her hair,

    Files

    Pornmaster_wan 2.2_14b_I2V_bukkake_v1.4_high_noise.safetensors

    Size:
    292.60 MB
    SHA256:
    b8a3e9d2dd9676e873b0b1619dc4ed3f7e73ff8e5174c17af758ad382463a789

    Mirrors

    HuggingFace (71 mirrors)
    CivitAI (1 mirrors)
    TensorFiles (4 mirrors)

    Available On (2 platforms)

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

    TensorArt
    buk - low
    TensorHub
    buk - low