Olivio Sarikas frames AI art as more than technical pattern reproduction. Drawing on Kandinsky and animated films, he argues that human creators add introspection, emotion, subtext and cultural perspective that models cannot supply on their own. AI can reproduce established visual structures, while the artist still decides what a work should make viewers feel.
The practical demonstration explores Dreamcore imagery in Midjourney. Sarikas compares model versions, personalization and prompt specificity, and finds that shorter, less literal prompts can produce more surprising compositions. Midjourney is particularly useful for ideation, mood boards and distinctive still images, although its video mode offers limited control over motion, duration and camera behavior.
Sarikas experiments with loops, low-motion and high-motion animation, then pairs generated scenes with abstract sound. He shows how the same image can feel comforting, uncanny or threatening depending on the audio, and argues that sound gives creators a direct way to shape the emotional subtext of AI visuals. Music generators are less useful when they insist on turning atmosphere into a conventional song.
For production, he recommends using Midjourney to establish the artistic concept and moving the selected image into a more controllable video model when necessary. He also discusses local versus hosted generation, hardware limits, licensing advantages of original AI-assisted music, and why future creative work is likely to combine human judgment with specialized AI tools rather than rely on one model for every step. Casual livestream conversation is omitted.
Watch on YouTube
