hard_cut_wan_lora / README.md
neph1's picture
Update README.md
3394d11 verified
metadata
tags:
  - lora
  - diffusers
  - template:diffusion-lora
  - image-to-video
  - i2v
widget:
  - output:
      url: >-
        https://cdn-uploads.huggingface.co/production/uploads/653cd3049107029eb004f968/Fn-H_-3IQyBC2n9W4Q2xv.mp4
    text: '-'
  - output:
      url: >-
        https://cdn-uploads.huggingface.co/production/uploads/653cd3049107029eb004f968/U5R29yT6heQL-B4HSBLx_.mp4
    text: '-'
  - output:
      url: >-
        https://cdn-uploads.huggingface.co/production/uploads/653cd3049107029eb004f968/CflxbcXJM8eZAxG7mszc8.mp4
    text: '-'
  - output:
      url: >-
        https://cdn-uploads.huggingface.co/production/uploads/653cd3049107029eb004f968/7k1aCBT8J8WE3BRQrNrsk.mp4
    text: '-'
  - output:
      url: >-
        https://cdn-uploads.huggingface.co/production/uploads/653cd3049107029eb004f968/u3aBFjHY_ERTXh2Ud5RfU.mp4
    text: >-
      two medieval warriors in armor, dueling in a swamp. the man on the left
      attacks, swinging his sword. the camera makes a hard cut, to a close up
      shot as the swords clash, sending sparks flying. the camera makes a hard
      cut, to a mid shot view showing the strained face of the left warrior as
      they continue fighting
  - output:
      url: >-
        https://cdn-uploads.huggingface.co/production/uploads/653cd3049107029eb004f968/-HYqMPiOHpBlk4E0yHifd.mp4
    text: >-
      the woman is tenderly repairing the robot. she moves her hands to the body
      of the robot. the camera makes a hard cut, to a close up shot of a panel
      in the robots chest. wires stick out, and the hands of the woman are seen,
      working on electronics.the scene has a 50s technicolor look
base_model:
  - Wan-AI/Wan2.2-I2V-A14B
instance_prompt: null

Hard Cut Lora Wan2.2 I2V 14B

Mirror of: https://civitai.com/models/2088559/hard-cut

v1: hard_cut_200_wan_i2v_high.safetensors v2: hard_cut_2_100_wan_i2v_high.safetensors

As much as I love Wan, if it's one thing it's not good at, it's cuts.

This lora is an experiment in greater shot control, by being trained on hard cuts. Most of the clips are also cut on action, fwiw, to allow for a continuous flow.

It's captioned with "wide-angle", "mid-shot" and "close-up". But I think Wan already handles them well.

Prompt format:

[a brief description of your initial shot.]

the camera makes a hard cut to [resulting shot]

[what happens after]

Where's the low noise model? There isn't one. It seems to work well with only the high noise model. I'll still train and test a low noise model, to see if it will improve consistency during cuts.

Play with strength if the scene changes too much. Too low strength will make it a dissolve transition (at least testing seems to imply that).

Download model

Download them in the Files & versions tab.