There’s a moment in every workflow where you stop feeling like an artist and start feeling like a traffic controller in a collapsing airport.
Mine arrived somewhere around image 24.
One browser sighed and gave up. Another browser picked up the slack like nothing had happened. Somewhere in between, a machine decided that 9:16 was optional. I did not argue with it.
I just kept going.
This is what it feels like now: not creation exactly, but coordination under polite chaos.
And strangely… it’s working.
The spark: two browsers and a quiet kind of madness
I didn’t plan to build redundancy into my creative process.
That sounds too engineered. Too intentional.
What actually happened was simpler:
- Grok started timing out in the middle of batches
- A Chrome-based session would stall
- So I opened Brave
- And then I opened another account
- And suddenly I had two machines doing the same thing in parallel like nervous twins trying not to blink at the same time
There’s something funny about it.
I’m sitting here in Palouse, Washington, watching two browsers behave like rival projectionists in a very small cinema, each trying to keep the film rolling while the other catches its breath.
And I am… apparently the manager of both.
Lab note:
At no point did I feel “advanced.” Mostly I felt like I was trying to keep a vending machine from eating my quarter twice.
The process: building a two-headed production creature
Here’s what the workflow actually looks like when you strip away the romance:
1. Midjourney stills
Everything begins as a frozen moment.
A character. A hallway. A storm that looks like it remembers something.
These are the bones.
2. Grok image-to-video generation
Each still gets:
- a prompt
- a mood direction
- a gentle suggestion of motion (never too much, never too literal)
The AI does what it wants anyway. I’ve stopped trying to micromanage gravity.
3. FluxGrok batch automation
This is where things start to feel like machinery.
- queues of prompts
- automatic submission
- background downloads
- occasional stubborn silence when something desyncs
Lab note:
Automation doesn’t remove attention. It redistributes it. Mostly into waiting.
4. The 24-image wall
One account hits a limit.
Not a crash. Not an error.
Just a polite, bureaucratic “that’s enough for now.”
So I switch:
- Chromium stops
- Brave begins
- The second account takes over like nothing happened
It feels less like failure and more like swapping batteries in a flashlight mid-storm.
5. The editing trick: time dilation as illusion
Here’s the quiet cheat code:
A 6-second clip becomes a 12-second memory.
I slow it down.
Not because I need more time…
but because the AI motion often feels more believable when it hesitates.
Fast motion looks like generation.
Slow motion looks like thought.
The tools that make the machine breathe
- Midjourney — still image generation (the frozen dream layer)
- Grok Imagine — image-to-video motion engine
- FluxGrok — batch automation and queue handler
- Chromium — primary production browser
- Brave — failover production browser
- Kdenlive — timeline editing and temporal distortion (slowdown, rhythm shaping)
- Screen-based download pipeline — whatever the browser decides to give me
Lab note:
The most important tool is not listed here. It is patience pretending to be workflow design.
Discovery: I am not making videos. I am managing constraints.
This is the part that surprised me.
I thought I was making creative decisions.
But most of my real decisions are now:
- Which browser is still alive?
- Which account has not hit its invisible ceiling?
- Which clip deserves to be slowed into something more emotional than it was originally intended to be?
The AI handles generation.
I handle limits.
And somewhere in that tension, the work becomes interesting.
Because constraint is no longer the enemy.
It’s the structure.
The mistake that wasn’t a mistake
One of the clips came back wrong.
Wrong aspect ratio.
It was supposed to be vertical. It rendered horizontal.
My first instinct was irritation.
Then I watched it.
And it actually worked better.
Not because it was correct.
Because it felt like a memory that had been slightly misfiled.
Lab note:
Happy accidents are not accidents. They are just decisions made without ego.
What I’ve learned from running 40 AI video clips a day
Here’s the honest list:
- Stability matters more than intelligence
- Automation doesn’t remove effort, it relocates it
- Two browsers is not redundancy—it’s survival strategy
- Most “errors” are just the system revealing its limits
- Slow motion is not aesthetic—it’s emotional compensation for synthetic movement
And maybe the biggest one:
- The work doesn’t scale in creativity. It scales in coordination.
The emotional layer nobody warns you about
There is a strange feeling that comes from watching machines almost succeed at making something beautiful.
Not failure.
Not success.
Something in between.
Like listening to a piano recital where one note is always slightly late, but never enough to ruin the song.
I don’t think I’m directing these videos anymore.
I think I’m negotiating with them.
Closing reflection: the alchemist and the queue
I used to think creativity was about inspiration.
Now it feels more like:
- queue management
- timing
- recovery
- and deciding when to let something imperfect be finished anyway
There is something oddly peaceful about it.
Not because it’s easy.
But because it is finally honest.
The machine doesn’t pretend to understand me.
And I’ve stopped pretending I control it completely.
We meet somewhere in the middle.
Between a prompt and a result.
Between two browsers.
Between intention and whatever survived execution.
Steve Teare
video alchemist
TerminallyBored.Monster
Palouse, Washington USA