Fast Hours

🎙️The AI podcast for creative degenerates / Tools, tactics, and the weird stuff in between / Hosted by Drew Brucker @dcbruck & Rory Flynn @ror_fly

[Drew Brucker ん @dcbruck]

0 VC funding. $500M(+) in revenue. Less than 200 employees. And now they're reinventing medical imaging.

@midjourney, you know, the AI image company.

Every founder hears the same thing: you can't build anything that matters without raising a mountain of venture capital $$$. But Midjourney never got the memo.

That structure buys the one thing money can't: nobody to ask for permission.

So last night, they finally pulled the cover off one of their secret hardware projects and the reveal was epic.

The world's first full-body ultrasound CT scanner. The goal: affordable full-body imaging for everyone on Earth.

How does it work?

You stand on a small platform and get lowered into a shallow pool of water. In under a minute, it builds a detailed 3D map of your body, capturing more than 25 organs and anatomical structures.

100x faster and 10x cheaper than an MRI. Oh, and zero radiation.

And if that wasn't enough, they're putting these in spas, not hospitals, with the first one set to open by the end of 2027.

I'm old enough to remember when people thought Midjourney was just an AI image model. But if you've followed along, they've said over and over again that they're a AI research and data company.

And now it's here.

cc: @FastHours Hours

[Drew Brucker ん @dcbruck]

Two years ago, every brand said this.

"We're going to wait for the AI space to mature a bit."

Waiting felt safe because regulations/safety were in the gray area and the tools were changing constantly, so why build on sand.

It felt like the responsible move.

Since mid-late 2025, I haven't heard that line once.

Most brands realized AI doesn't need to start as a giant public-facing campaign.

It can start small and low-risk (concept exploration, moodboards, asset variations, internal workflows, brand guardrails etc.)

That's the part people underestimated.

For the brands that got started early, the reps, taste calls, constraints, and thousands of small rejections taught them invaluable future lessons that the others are now trying to play catch-up on.

And those teams that started messy two years ago have exactly one advantage. They're two years into the part that can't be skipped.

[pratish @pratishhhhh]

my fav ai creator @jboogx_creative on my fav ai podcast

@FastHours

letsgoooo🔥🔥

[Fast Hours @FastHours]

@sudiprokaya Simi

[Drew Brucker ん @dcbruck]

True story: I've watched a team win a pitch with a single AI image.

But then by Month 2, they lost the account.

Boom, just like that.

One image is easy. Holding the look across a thousand frames, on deadline, while a client picks apart every one, is what breaks teams.

[Rory Flynn @Ror_Fly]

Weavy → SIMPLE batch-processing workflow.

>>Using image style guides<<

(Full workflow in thread)

PROCESS:
01. Input product + prompt idea
02. Input "style reference."
03. Auto-gen multiple prompts
04. Using visual style reference
05. Gen multiple images...in one click.

Simple...easy...malleable.

[Fast Hours @FastHours]

@dcbruck

[Reve @reve]

Today, we’re launching Reve 2.0, the best 4K image model in the world.

We invented a new way to generate and edit any image using precise layouts. For the first time, it’s possible to create images you can touch.

[Drew Brucker ん @dcbruck]

Edit models aside, make no mistake, Midjourney is still king.

[Drew Brucker ん @dcbruck]

@Ror_Fly and I covered this on the upcoming @FastHours.

Social experiments like this one act similarly to how metaphors can be used in comms. to quickly digest a concept/scenario.

[Drew Brucker ん @dcbruck]

Nana hit one combo and the Wi-Fi started calling her ma’am.

Text-to-video in Seedance. No references.

Prompt ⬇️

[Rory Flynn @Ror_Fly]

GPT2 + Seedance 2.0 → Yoga Flow.

Controlling pose transition.

(Prompt + process in thread)

PROCESS:
01. GPT2 Img: Create pose sequence in
02. GPT2 Img: Create base character
03. SD2 Omni-Ref: Animating scene
04. Codex for stitching + frame interpolation

While not perfect step-to-step...

Still impressive.

[Rory Flynn @Ror_Fly]

GPT2 + Seedance 2.0 → Capoeira Sequence.

Testing for more complex choreography.

(Full workflow + prompts in thread)

PROCESS:
+ GPT2: Choreography diagram
+ GPT2: Camera direction diagram
+ Midjourney: Setting base image
+ Seedance 2.0 Omni-Ref: Video gen

Didn't follow 100%.

Probably needs more than a 15s gen.

[Drew Brucker ん @dcbruck]

Ayyyyyyye

[Drew Brucker ん @dcbruck]

GPT-Image-2 is good. But is it nano good?

@Ror_Fly and I break it down.

New Fast Hours:
→ GPT-Image-2 vs Nano Banana Pro, head to head
→ Rory turns fiddleheads into a custom typeface
→ I turn my handwriting into a working font
→ Graphics that look scary good
→ Kling adds native 4K while Veo 3.1 falls behind
→ Claude Design
→ Why context > prompts

cc: @FastHours

[Drew Brucker ん @dcbruck]

Midjourney v8.1...

@Ror_Fly and I took the new model apart on Episode 66.

Mood boards, describe, stylize behavior, what's broken, what's fixed, and why --sv6 still fires but --sv4, --sv5, and --sv7 are dead in 8.1.

And of course, a Mac folder hack I lost a full day to.

cc: @midjourney

[Drew Brucker ん @dcbruck]

Walking the streets with Midjourney v8.1

[Drew Brucker ん @dcbruck]

Walking the streets with Midjourney v8.1

[Drew Brucker ん @dcbruck]

Walking the streets with Midjourney v8.1

[Drew Brucker ん @dcbruck]

Walking the streets with Midjourney v8.1