The first part is about the Steerable Motion Workflow by Peter O'Malley (POM). I've made slight modifications to it, including adding a FaceDetailer to meet my specific needs but this is by far the best image interpolation workflow I've worked with.
🙋♂️I used 6 images from my Nike video for the first example.
You can control the motion by adjusting the influence graph and using one of three Motion LoRAs.
🤿Another technique I've been experimenting with in my creative explorations is generating depth maps within ComfyUI for various effects and transitions. One method I'm showcasing here uses the Fake Depth LoRA to convert your prompt into a depth map with AnimateDiff.
👽my text prompt:
spinning - overlapping boxes, 4k
Depth map, black background, high contrast
Then you can to use it as the source for your ControlNet in other workflows, which is much faster than relying solely on 3D applications like Houdini or Blender to generate masks.
Links:
https://github.com/banodoco/Steerable...
https://civitai.com/models/392921/dep...
/ ardiology
--------------------------------------
👩💻 sign-up to our free newsletter:
newslounge.co/subscribe
--------------------------------------