Some ideas just won’t leave me alone. A character I imagine, a movement I can see in my mind — it nags at me until I make it exist. I light up Midjourney, Grok, WAN, SeeDream… and suddenly the fun collides with reality: every prompt costs money, every tool has quirks, every clip is a negotiation.
I love that tension. It’s part of the challenge. Watching pixels respond to my vision, hearing audio snap into rhythm, seeing micro-motion suddenly feel alive — that’s the payoff. And yes, the budget keeps me honest. Every dollar spent is a choice, a tiny commitment to the story I want to tell.
Lab note: Creativity isn’t just inspiration. It’s obsession + friction + delight + cost. That’s the dance I live in.
The Spark of the Frame
Before any workflow or credit card enters the picture, it’s the spark — a pose, a glance, a shoulder tilt, or a fleeting emotion. These micro-moments are everything.
I start in Midjourney for still images. Relax mode lets me pour through variations, exploring light, shadow, and pose until something feels alive. These stills aren’t final products — they’re breadcrumbs, hints, prompts for motion, timing, and audio.
Lab note: The still image is never the product. It’s a lure, pulling me toward animation and subtle movement.
Grok: The Workhorse
Once a visual idea exists, Grok handles most of the heavy lifting. Image-to-video generation is forgiving and surprisingly fast. The auto-SFX sometimes surprise me — a breath, a flicker, a texture I didn’t ask for but suddenly need.
SuperGrok Lite is inexpensive: $10/month, predictable, and perfect for experimentation. Most clips live here. I save WAN and SeeDream for moments where subtle expression counts.
Lab note: Cheap AI tools are liberating. They let me fail fast, iterate, and explore ideas without guilt.
WAN and SeeDream: Precision Tools for Key Shots
There are video moments Grok can’t reach. Facial expressions, cinematic angles, subtle gestures — that’s WAN and SeeDance territory.
These are pay-as-you-go luxuries: WAN clips with lip sync cost $0.50 for five seconds, $1 for ten seconds, and SeeDance credits come in $10 or $30 chunks (but 5-second clips are $0.28 each). I reserve them for moments where emotion matters.
Lab note: Expensive AI sharpens intention. It forces me to plan, pre-visualize, and respect timing.
Audio: The Pulse of the Clip
Sound gives life to movement. I isolate vocals from music with LALAL.AI — your current balance is measured in minutes (Not GPU time. Music duration.) That lasts for months of experimentation. Each voideover or lyric stem is sliced into 5- to 10-second clips in Audacity, then fed into WAN for lip-sync and motion.
ElevenLabs covers voiceovers at $5/month — mostly for narration or character lines. Together, audio and video become heartbeat and breath.
Lab note: Micro-motions in sound are just as important as in pixels. A pause, a soft consonant, a subtle exhale — these are what make clips feel alive. ElevenLabs v3 will annotate the audio for best emotion if you click the Enhance button and then the “keep” popup. This method usually outperforms ChatGPT bracketed annotation for embedded enhancement.
EXAMPLE:
[inhales deeply] Sometimes my mind leaps — ideas, flashes of thought.[exhales sharply] I chase them, feel the spark.
[long pause] Then I pause.
[long pause] I soften the edges, frame them lightly…
[whispering] almost as if I’m speaking to no one —
[whispering] or to someone I hope will understand.
[inhales deeply] It’s not about proving anything.
[exhales sharply] It’s about letting curiosity exist safely.
Stacking the AI Assistants
Here’s the lineup producing my shorts:
- Midjourney: inspiration, stills, ideation
- Grok SuperGrok Lite: image-to-video with auto-SFX
- WAN / SeeDream: high-fidelity precision image-to-video shots
- LALAL.AI: vocal isolation, stems
- Audacity: trimming, slicing 5–10-second audio segments
- ElevenLabs: voiceover generation
Lab note: Each tool is like an instrument. The magic is in how they play together — rhythm, tension, and timing.
Budget Reality
The AI stack isn’t free:
- SuperGrok Lite: $10/month — cheap, reliable, low stress
- WAN AI — $0.50 to $1 per clip, $10 to $30 per credit chunk
- SeeDream: pay-per-use, $0.27 per 5-sec per clip
- LALAL.AI: $40 chunks (credits), lasting months if managed carefully
- ElevenLabs: $5 per month
I’ve been on Midjourney since the very beginning at $30/month (all-you-can-eat when using relax mode). Adding all the tools together, yes, it’s a small fortune (for a poor man). But every expense is a deliberate choice for quality, experimentation, and speed.
Lab note: Cost is attention, patience, and iteration — not just dollars.
Making the Intangible Tangible
The thrill is turning imagination into film. Tiny gestures, subtle glances, shoulder tilts — all captured and made real through this stack. And the best part? It’s within reach without a Hollywood budget. Every second, every micro-motion, every credit is a choice that brings vision closer to reality.
Lab note: The tools I have now make ideas tangible. The tools I hope to add will make them believable, repeatable, and cinematic.
Discovery and Joy
The biggest takeaway? Friction and cost amplify delight. Every happy accident, every serendipitous auto-SFX, every micro-motion that finally works feels earned. Watching clips breathe — even in a five-second WAN segment — helps me feel alive.
This stack lets me explore, experiment, and iterate without fear. Planning, slicing, and deliberate spending become part of the creative rhythm. The AI assistants don’t replace me — they amplifies my vision.
TL;DR: Build a workflow that balances cheap experimentation with expensive precision. Respect micro-motions. Spend deliberately. Celebrate happy accidents. Let AI be your orchestra, not your master.
Steve Teare
video alchemist
TerminallyBored.Monster
Palouse, Washington