I started with nothing but a raw list of 14 songs and a simple brief: build a vertical neon-noir series around a lonely small-town girl in a Blade Runner city. No script. No characters. Just the emotional temperature of each track. What came out the other end was a complete 7-episode arc that actually felt like it had a heartbeat.
Here’s the real, unfiltered process — and the tricks that made it possible in under two days.
The Lab Deep-Dive: How It Actually Went Down
1. Songs first, story second
I handed Grok the full list of 14 needle drops. We didn’t write a story and then try to cram songs in. Instead, we studied the core mood of each track and let the strongest seven carve the emotional spine of the series. The remaining seven became strong alternates.
Lab note: Starting with music as the foundation instead of forcing songs into a pre-written plot was the single best decision we made. Mood and character truth beat musical checklists every time.
2. Locked characters + one powerful style reference
We generated clean frontal references for “Her,” the dangerous blonde “Him 1,” and the gentle “Him 2.” Then we locked everything together with a single style reference from Midlibrary: Neon Dystopian Inkplay (--sref 3683525326 --sv 4). That one reference gave the entire series its signature rain-soaked, inky, teal-orange Blade Runner soul.
3. The talking-head trick
In every episode we added one intimate talking-head shot synced to the first-person narration. It’s a cheap psychological hack — the primacy effect makes the audience connect faster and deeper. Simple but ridiculously effective.
4. The cheap lipsync workaround
Normally people pay $0.50–$1.00 per clip on services like wan2video.com. We refused. Instead:
- Generate a short speaking clip in Grok from the script (even though the voice was wrong)
- Drag that clip into ElevenLabs Voice Changer using our custom narrator voice
- Perfectly synced lips + matching audio for pennies
We only needed one talking-head clip per episode. The rest of the shots stayed as beautiful static-to-video motion.
CHEAPER LIP SYNC
5. Prompt volume + automation
We built 112 structured prompts (8 main + 8 B-roll per episode). Using the Midbot Chrome Extension, we generated 4 variations per prompt → roughly 448 vertical images. What would have taken days became manageable in focused bursts.
Lab note: Speed preserved creative energy. We finished the entire production in roughly 24 to 48 hours instead of weeks.
Tools & Creative Stack
- Story & Song Mapping: Grok
- Images & Style: Midjourney V7 + Neon Dystopian Inkplay sref + Midbot Chrome Extension
- Talking Head Video: Grok (initial generation) + ElevenLabs Voice Changer
- Narration: ElevenLabs custom husky small-town female voice
- Editing: Kdenlive
- Final Format: Vertical 9:16 with 1-minute needle drop exactly at 3:00 per episode
Cost: Under $5 per episode. Insanely cheap for the final quality.
Discovery / Takeaway
What this project actually taught me:
- Let the music lead. The strongest emotional arc emerged when we followed the needle drops instead of imposing a traditional story structure.
- One killer style reference can unify everything. That single
--srefdid more visual heavy lifting than any other setting. - Talking heads are psychological cheat codes. One per episode dramatically increased connection.
- Automate the volume, protect the soul. Midbot handled repetition so we could focus on feeling.
- Cheap workarounds beat expensive tools. The Grok → ElevenLabs voice changer pipeline delivered better results for a fraction of the price.
- Speed is part of the art. Finishing fast kept the project exciting and prevented perfectionism from killing momentum.
TL;DR:
Start with the music. Cut ruthlessly to the best seven. Lock faces and style early. Add one talking head per episode. Automate the grind. Use clever cheap hacks for lipsync. The AI handles the rain — you handle the heartbreak. The result can feel shockingly human.
This was “just another experiment,” but it proved that disciplined process plus strong musical seeds can produce something that actually moves people.
I’m already eyeing the next playlist.
Got a set of songs you think deserves the neon treatment?
Steve Teare
video alchemist
TerminallyBored.Monster
Palouse, Washington USA