CivArchive
    Wan2.1 I2V Lightx2v Step/CFG NVFP4 - v1.0
    NSFW

    This model is the result of a partial NVFP4 quantization of Wan2.1-I2V-14B-480P-StepDistill-CfgDistill-Lightx2v by lightx2v, produced using convert_to_quant by silveroxides. Some layers have been kept on their original BF16 format, while others were quantized as MXFP8 or NVFP4, mostly.

    Wan2.1-I2V-14B-480P-StepDistill-CfgDistill-Lightx2v is an image-to-video generation model built on Wan2.1-I2V-14B-480P. It applies step distillation and classifier-free guidance distillation to reduce inference to 4 steps without CFG, cutting generation time substantially while preserving output quality.

    IMPORTANT

    Since NVFP4 is only supported on NVIDIA Blackwell architecture GPUs, running this model requires a Blackwell GPU with its corresponding support enabled in torch, along with a recent version of ComfyUI and comfy-kitchen built against CUDA 13. You'll also need at least torch 2.10 to make it run, so this is not for the faint of heart with all the corresponding, if you know what I mean.

    The model can be used in ComfyUI with the following parameters, based on the distilled model's own recommendations:

    • Shift: 5.0

    • Sampler/Scheduler: euler/simple

    • CFG: 1.0

    • Steps: 4

    Other combinations are possible, such as lcm, heun/linear_quadratic, etc.

    Description

    This version tries to protect the following layers, among others.

    • cross_attn.k

    • cross_attn.q

    • self_attn.k

    • self_attn.q

    FAQ

    Comments (3)

    SR_RakibApr 24, 2026
    CivitAI

    can you tell me please this model run 3060 12GB?

    InsecureErasure
    Author
    Apr 24, 2026

    It's an NVFP4 quant, so only supported on Blackwell architectures, i.e. 50xx series.

    ialocalefacileMay 16, 2026
    CivitAI

    Hello, the 14b i2v model is too heavy for me, I am happy to start getting very consistent results with your model with: 480x480, shift 5, 6 steps, cfg 1.4, denoise 0.80, rife x2 and upscale x2, on an RTX 5070 Ti 12GB laptop. I shared the screenshot of my workflow on this link. I'll share the video on this gallery.

    Checkpoint
    Wan Video 14B i2v 480p
    by InsecureErasure
    Download (Beta) View on CivitAI
    base model
    wan
    img2vid
    lowvram
    i2v
    wan 2.1
    nvfp4
    blackwell
    rtx50xx
    gpu-poor

    Details

    Downloads
    240
    Platform
    CivitAI
    Platform Status
    Available
    Created
    4/23/2026
    Updated
    6/12/2026
    Deleted
    -

    Files

    wan21I2VLightx2vStep_v10.safetensors

    Size:
    10.09 GB
    SHA256:
    6b81cabaa4771722e5f7111c0928d63bdb588b0e91c5117bdf56fd9fe4ccc6a9

    Mirrors

    CivitAI (1 mirrors)

    wan21I2VLightx2vStep_v10.safetensors

    Size:
    602.86 MB
    SHA256:
    644857118a5e8f5f8bb13e56c860eca7bdc1a32dd12ffdc97ed9cd9460254945

    Mirrors

    HuggingFace (5 mirrors)
    CivitAI (1 mirrors)

    wan21I2VLightx2vStep_v10.safetensors

    Size:
    1.18 GB
    SHA256:
    64a7ef761bfccbadbaa3da77366aac4185a6c58fa5de5f589b42a65bcc21f161

    Mirrors

    HuggingFace (197 mirrors)
    CivitAI (2 mirrors)
    CivitasBay (1 mirrors)
    ModelScope (2 mirrors)
    ModelScope CN (4 mirrors)

    wan_2.1_vae.safetensors

    Size:
    242.06 MB
    SHA256:
    2fc39d31359a4b0a64f55876d8ff7fa8d780956ae2cb13463b0223e15148976b

    Mirrors

    HuggingFace (302 mirrors)
    CivitAI (4 mirrors)
    TensorFiles (1 mirrors)
    CivitasBay (1 mirrors)
    ModelScope (2 mirrors)
    ModelScope CN (4 mirrors)