CivArchive
    Claire Forlani | Wan2.1 T2V-14B - v1.0

    Claire Forlani is a British actress known for her roles in films like Meet Joe Black, Basquiat, and Mallrats. She was born in London in 17 December 1971.


    Fuel my GPU's caffeine addiction! (Ko-fi)
    https://ko-fi.com/cyberaimania




    Trained on Wan 2.1 T2V-14B model locally on RTX 4090 with 50 publicly accessible images using diffusion-pipe.

    Trigger: cforlani

    Example prompts:

    --

    (masterpiece, best quality), beautiful portrait of cforlani, looking directly at camera, soft smile, simple background, detailed eyes, cinematic lighting.
    --

    (masterpiece, best quality, photorealistic), cforlani bathed in warm golden hour sunlight, gentle smile, looking slightly off-camera, soft focus background, bokeh, ((captivating eyes)).
    --

    (masterpiece, fashion photography), medium shot of cforlani wearing an elegant black evening dress, sophisticated pose, studio lighting, sharp focus, vogue style.

    This is my first LORA, so I would REALLY appreciate any comments with suggestions, feedback, advice, or even negative opinions.

    Enjoy!

    Description

    Trained T2V but also works great on I2V

    FAQ

    Comments (4)

    whitespider9999Apr 13, 2025· 1 reaction
    CivitAI

    How does making wan loras differ from making traditional still image loras?

    CyberAImania
    Author
    Apr 13, 2025· 3 reactions

    Training Wan2.1 video LoRAs involves fine-tuning a large transformer model (not a U-Net), often requiring video clip inputs (prepared at specific FPS/frame counts) or images for T2V. It demands more aggressive VRAM optimization (FP8, checkpointing, 8-bit optimizers, maybe swapping) due to model size and sequence processing, generally leading to longer training times compared to standard image LoRAs. The core LoRA concept is similar, but the base model and data handling differ significantly.

    lowcaloriesyrupApr 14, 2025· 1 reaction
    CivitAI

    Thanks for your work and for including training data.

    Any idea why your generations look almost... rendered? Is this intentional?

    CyberAImania
    Author
    Apr 14, 2025· 1 reaction

    Thank you.

    This is my first LORA ever, and I haven’t used any other filters or LORAs to make the videos look more natural. I’m still learning, but I believe that with the help of my LORA, it's possible to generate more natural-looking results.

    LORA
    Wan Video

    Details

    Downloads
    173
    Platform
    CivitAI
    Platform Status
    Deleted
    Created
    4/13/2025
    Updated
    5/8/2026
    Deleted
    5/23/2025
    Trigger Words:
    cforlani

    Files

    dataset_claire.zip

    Mirrors

    CivitAI (1 mirrors)
    TensorFiles (1 mirrors)

    Claire_14B_r32.safetensors

    Mirrors

    CivitAI (1 mirrors)
    TensorFiles (1 mirrors)