EP 030 · 66 MIN PROMPT: 'CYBERNETIC ANATOMY OF A SNAIL'

Midjourney Moodboards The Deeper You Go, the Crazier It Gets

February 9, 202566 min Rory Flynn & Drew Brucker

The short version

Episode 30 is a live moodboard-stacking session: Rory and Drew layer SREF codes, personalization profiles and small themed moodboards (goo, kaleidoscope, Japanese movie poster, film negative) on top of each other, testing a shared "cybernetic anatomy of a snail" prompt and an album-cover collage prompt side by side. They also recap Midjourney office hours, including a claim that Rory's own Google Veo output pushed David to move up Midjourney's video timeline, and preview an upcoming premium moodboard pack.

Key takeaways

  1. 01

    Build tiny, single-trait moodboards (5-15 images) for one job, like a "film negative" texture or a specific color grade, then stack two or three together.

  2. 02

    Test a moodboard's built-in bias by prompting an empty character (a period or slash) or a random emoji; Rory's fairy-emoji prompt kept returning mushrooms.

  3. 03

    You can weight your global personalization profile separately from a moodboard's weight, shifting the result toward the profile or the moodboard depending on the ratio.

  4. 04

    Rediscovered setting: Style Version 1 (SV1) still produces distinct, usable results layered on top of current moodboards and personalization stacks, per Rory.

  5. 05

    Rory says Midjourney's David reportedly moved his video timeline up after seeing Rory's own Google Veo output, from a 2026 target toward sometime in 2025.

  6. 06

    AI video tools still can't do sports (basketball, soccer, gymnastics) or complex physical motion like a fly-fishing cast, because no two bodies or motions repeat identically.

  7. 07

    Rory's AI-video read: Sora regenerates just a bad few seconds inside a clip, Runway targets production teams, and MiniMax is oddly creative.

Receipts · said on the record

“I was multi-prompting in the prompt: I had about six mood boards, four profile codes, two SREF images, and then I'd use style version on top of that to try and figure out if there would be a significant difference.”
Rory Flynn Stress-testing how many moodboards and codes can be stacked in one prompt Play from 8:02
“Why box yourself in? I can't even tell you how many different avenues I've gone down, both things that are very predictable that I'd like, and also unpredictable that I'd like.”
Drew Brucker Against locking into one narrow aesthetic lane Play from 9:38
“In my experience, sports, Veo 2 is really bad with basketball, still can't get it, basketball and soccer … it's impossible. You can't replicate it because there's no two plays that are exactly the same.”
Rory Flynn Why AI video tools still fail at sports footage Play from 52:58
“That's why hands are so hard. Hands do really weird things. And when you get photos of hands, you're not getting the same photo of a hand, you're getting a hand in the 360, up here, down low, from the side.”
Drew Brucker Why hands are uniquely difficult for AI generators Play from 53:22
“I like to tell people to go, if you have complaints about the hands, try to go draw one. Go try to draw a hand or some eyes. It's hard.”
Rory Flynn Pushing back on casual criticism of AI-generated hands Play from 53:58

Questions this episode answers

What is moodboard stacking in Midjourney?

It's combining several smaller, single-purpose moodboards, for example a texture board, a color-grade board, and a subject-style board, on top of a personalization profile in one prompt. Rory and Drew build dedicated 5-to-15-image boards for one job and then layer two or three together for a blended look.

How do you find a Midjourney moodboard's built-in bias?

Drew suggests prompting an empty character like a period or slash, or a random emoji, with the moodboard active, to see what it defaults to. Rory did this with a fairy emoji and kept getting mushrooms.

Did Google's Veo change Midjourney's video plans?

Rory and Drew relayed that Midjourney's David originally pegged 2026 as the year of video, but after seeing Rory's own Veo output he reportedly said he'd changed his mind and now expects a breakthrough sometime in 2025.

Can AI video tools handle sports or complex physical motion?

No, according to Rory: basketball, soccer, gymnastics and even a realistic fly-fishing cast still defeat tools like Veo 2, because no two bodies or motions repeat identically and the underlying physics are too complex to fake convincingly.

How do Sora, Runway, Kling and MiniMax compare for AI video?

Rory calls Sora's ability to regenerate just a bad few seconds inside a longer clip a standout feature, sees Runway as built for production teams and editors, calls Kling an emerging all-in-one with try-on and scene-mixing tools, and calls MiniMax "the little engine that could" for oddly creative results.

Show notes

MJ:FH Ep. 30—Midjourney Moodboards The Deeper You Go, the Crazier It Gets (with Drew Brucker & Rory Flynn)

Every word Read the full transcript

If you liked this one