AI

AI R&D, experiments, and sketches.

Midjourney

Comfy UI

A node-based, modular AI tool for generating and refining images with precision and flexibility.

Using Keyframes

comfyUI IMage & text prompt to video

Using still image from my 3d scene.

ComfyUI, MiniMax Hailuo Video 01, Image reference & text to video

ComfyUI, Image reference with text prompt, using Kling Pro 1.5

ComfyUI Vid2Vid Transfer

Transfers the motion from a source video onto a target image or object.

INPUT: Blender

OUTPUT: ComfyUI

CLEANUP: After Effects

COMPOSITE:Midjourney, Photoshop, After Effects

INPUT from traditional oil painting, driven by video reference, and distorted in After Effects.

original oil painting

photoshopped to match more closely to first video frame

video INPUT


runway ml

A powerful AI tool for generating and editing videos, blending creativity with machine learning.

runway Vid2Vid Transfer

Transfers the motion from a source video onto a target image or object.

Used practical footage of coffee beans in Runway to generate peanuts and candy.

Generated original music with Suno.

Quick character setup in Adobe Animate and composited in After effects.

3D render as input

Transfers motion of video to a targe style provided by image, text prompt, or preset.

Blender render from scene for INPUT

Runway ML Gen 1 Video to Video, using blender render for base, with Midjourney style reference.

Midjourney

Runway ML Gen 1 Video to Video, using blender render for base, with text prompt for style reference.

Runway ML Gen 1 Video to Video, using blender render for base, with Runway preset for style reference.

text prompt:
"A tiny plane glides through a tranquil sky, bathed in warm light and soft shadows.

Runway ML Gen 1 Video to Video, using blender render for base, with text prompt for style reference.

text prompt:

a tiny pink and blue retro plane navigates a bright, faded pink sky

runway image2Vid Transfer

Converts still images into dynamic motion, guided by an optional text prompt for added control.

“Stop motion set” video generated with Runway Gen 3 Alpha using base Midjourney image.

Runway Gen 3 video from digital painting.

original digital painting


disco diffusion

I began experimenting with Disco Diffusion in 2022. It’s an AI tool that generates dreamlike, painterly visuals from text prompts using diffusion models. Following are some tests from that time.


chatgpt + p5js

Following are visuals that were created using coding prompts in ChatGPT for executing in P5JS.

AI experiment, 2022. Used chatGPT to write code for P5JS. Music by @ruiqimusic

AI experiment, 2022, coded p5js visuals using chatGPT. Coded MIDI music using chatGPT. Some post in After Effects.

chatGPT > P5JS

music by dokfractal