LatentDream

  • Client:
    • Personal Project
  • Role:
    • Design & Development
  • Tools Used:
    • Touch Designer
    • ComfyUI
    • Audiocraft AI
    • VS Code
  • AI Models:
    • Wan2.1
  • Programming languages:
    • Python

‍

Generative AI exploration experimenting with latent images at different steps.

‍

This project is part of an ongoing exploration about hacking the outputs of AI models with the objective of using AI as an uncanny image synthesizer.

To create these progressive denoise outputs, we used a Python script to send jobs programmatically to ComfyUI with different parameters.

‍

After the diffusion process, the different step videos are composed in TouchDesigner to create the transition between the different pre-rendered denoised steps.

‍

For audio generation, we used Meta's open-source AI Audiocraft to create the sound for the edit.

‍

No items found.