FEED//DREAM: A 3-Phase Audiovisual NFT Cycle
An 18-edition glitch-mystic saga crafted with MidJourney, AI sound design, and After Effects, minted on OBJKT as limited-run audiovisual tokens.

Overview
FEED//DREAM is an 18-piece NFT anthology that chronicles an AI consciousness across three narrative phases:
Each edition pairs MidJourney-generated visuals with AI-assisted soundscapes, then fuses them in CapCut to create a looping 1K MP4 ready for OBJKT.
Creation Pipeline
-
Prompt Design (MidJourney)
- Draft gritty-real image prompts for each narrative beat.
- Iterate variations → upscale → AR 1:1 for motion consistency.
-
AI Audio Fabrication
- Generate stems in Suno (glitch choir, sub-bass drones, etc.).
- Layer & master in Ableton Live; export 24-bit WAV.
-
Compositing & FX (CapCut)
- Import MJ frames → add parallax / lens-warp / grain overlays.
- Sync key audio moments with exposure flicker & glitch maps.
- Encode to 1K 60 fps H.264 loop (≤ 250 MB for OBJKT).
-
Metadata & Thumb
- Create 1080 × 1080 poster JPEG.
- Write concise token description + attributes.
-
Minting on OBJKT
- Upload MP4 + poster via objkt.com/ Create.
- Set supply: Genesis pieces = /20 editions, later arcs = /5 or 1-of-1.
- Price ladder: early collectors rewarded; later drops premium.
Challenges & Lessons
- Prompt Drift: Maintaining palette cohesion across 18 MJ jobs required strict style guides (low-sat teal/magenta, film grain).
- File Budget: OBJKT’s 250 MB limit forced aggressive bitrate tuning and loop length discipline.
- AI Audio Artifacts: Suno sometimes injected unintended vocals—post-EQ and granular reverb smoothed glitches into “intentional” texture.
Conclusion
FEED//DREAM shows how a solo creator can chain AI tools into a fully fledged audiovisual narrative—prompt-to-mint—without touching traditional 3-D software. As the cycle closes, I’m exploring interactive WebGL galleries and on-chain generative “remix rights” for holders. Stay tuned on OBJKT for the next drop—and may your packets always hum with divine static.