Wan2.2-T2V-A14B-4steps-lora-rank64-Seko-V1_low_noise_model
Back

Wan2.2-T2V-A14B-4steps-lora-rank64-Seko-V1_low_noise_model
0 1 10

Other

Wan2.2-T2V-A14B-4steps-lora-rank64-Seko-V1_low_noise_model

Wan2.2-Lightning

We are excited to release the distilled version of Wan2.2 video generation model family, which offers the following advantages:

  • Fast: Video generation now requires only 4 steps without the need of CFG trick, leading to x20 speed-up

  • High-quality: The distilled model delivers visuals on par with the base model in most scenarios, sometimes even better.

  • Complex Motion Generation: Despite the reduction to just 4 steps, the model retains excellent motion dynamics in the generated scenes.


Training details

High_noise_model and low_noise_model are distilled simultaneously. On 81x720x1280, 81x1280x720, 81x480x832, 81x832x480 size options.

Tips on usage

In the distillation process, euler scheduler is used, with time_shift = 5. Specifically, the timesetps should be [1000.0000, 937.5001, 833.3333, 625.0000] In inference, please set the scheduler accordingly. Besides, please set cfg = 1.0 so that NFE = 4 but not 8.

This model is sourced from an external transfer (transfer address: https://huggingface.co/lightx2v/Wan2.2-Lightning ）,if the original author has objections to this transfer, you can click，
Appeal
We will, within 24 hours, edit, delete, or transfer the model to the original author according to the original author's request

suoaao

suoaao

Other

Model Information

Original author：
lightx2v
Model Type：
LoRA
Basic Model：
WAN2.2
Resource Name：
models/loras/Wan2.2-T2V-A14B-4steps-lora-rank64-Seko-V1_low_noise_model.safetensors
MD5：
9aac82334f2ec4b4a975422a1718b691

Wan2.2-Lightning

We are excited to release the distilled version of Wan2.2 video generation model family, which offers the following advantages:

  • Fast: Video generation now requires only 4 steps without the need of CFG trick, leading to x20 speed-up

  • High-quality: The distilled model delivers visuals on par with the base model in most scenarios, sometimes even better.

  • Complex Motion Generation: Despite the reduction to just 4 steps, the model retains excellent motion dynamics in the generated scenes.


Training details

High_noise_model and low_noise_model are distilled simultaneously. On 81x720x1280, 81x1280x720, 81x480x832, 81x832x480 size options.

Tips on usage

In the distillation process, euler scheduler is used, with time_shift = 5. Specifically, the timesetps should be [1000.0000, 937.5001, 833.3333, 625.0000] In inference, please set the scheduler accordingly. Besides, please set cfg = 1.0 so that NFE = 4 but not 8.

This model is sourced from an external transfer (transfer address: https://huggingface.co/lightx2v/Wan2.2-Lightning ）,if the original author has objections to this transfer, you can click，
Appeal
We will, within 24 hours, edit, delete, or transfer the model to the original author according to the original author's request