Guest · Ep 070
Tyler Bernabe aka jboogxcreative on Fast Hours.
On Fast Hours
Generative AI creator Tyler Bernabe (jboogxcreative) joins Fast Hours to break down the viral formula behind his food videos, a deliberate mashup of food and shock value, and his full creative stack: Midjourney 8.1 for imagery, Nano Banana and GPT for edits, Seedance for video, and a custom Claude-built YAML prompter that writes 10,000-character prompts by voice.
The 30-second version
- 01
His viral formula, in his own words: everyone loves food and wants sex, and mashing those two things together sends the velocity on a video through the roof.
- 02
He deliberately buries his Patreon call-to-action ('comment LEARN') at the bottom of captions so only viewers who read that far, not casual scrollers, get funneled in.
- 03
Translating a Seedance prompt into Chinese before running it through Artcraft's Chinese API reportedly makes prompt adherence about 90 percent better, he says.
- 04
Default platforms cap Seedance prompts around 3,500 characters; his custom Claude prompter writes up to 10,000, which he says is what makes his complex anime scenes possible.
The episode
The Weird AI Video Formula Getting Millions of Views
ft. Tyler Bernabe (jboogxcreative)
Generative AI creator Tyler Bernabe (jboogxcreative) joins Fast Hours to break down the viral formula behind his food videos, a deliberate mashup of food and shock value, and his full creative stack: Midjourney 8.1 for imagery, Nano Banana and GPT for edits, Seedance for video, and a custom Claude-built YAML prompter that writes 10,000-character prompts by voice. He also explains his Instagram-to-Patreon funnel and why he refuses to let AI fully automate his own creative decisions.
Open the episodeTyler's stack, as they told it
- Midjourney ▲ Rates it
His default imagery generator since 8.1, used for aesthetic vibe, brainstorming and creative inspiration.
V8.1 · V8.2 - Nano Banana ▲ Rates it
His go-to editor for tricky details like gold grills, which he says GPT can't render well.
- GPT Image / ChatGPT Images ◆ Mixed
His workhorse for professional brand edits, though he calls it 'atrocious' at rendering gold grills.
- Kling ▲ Rates it
Kept 'in his back pocket' for reliable start-frame/end-frame generations within the first few tries.
- Seedance ▲ Rates it
His main video model ('has me in a chokehold'), driven by a custom 10,000-character YAML prompter.
- Claude ▲ Rates it
His 'partner': he voice-dictates ideas that a custom project turns into structured YAML video prompts.
- Figma Weave (Weavy) ◆ Mixed
Has a pro account and alternates with Freepik Spaces, though an earlier 'Patchwork' feature confused him.
- Freepik ▲ Rates it
Works in Freepik's Spaces canvas (which he also calls Magnific) alongside Weavy for image generation.
Tyler, on the record
- 01 His explanation for why the food-video formula went viral
“Everyone loves food and has some sort of emotional connection to food, whether that's good or bad. And every everybody ev everybody wants sex. It's like it's those two things and you mash them together and the v velocity on that content is crazy.”
- 02 His answer to 'is AI art' critics
“Is the thing that was in my head that is now on the screen, and that makes me happy. If it resonates with anyone else, that's just a bonus.”
- 03 His filter for deciding which new AI tools to adopt
“I view everything through the lens of does this help my longevity or does this distract from my longevity? And those are two very important distinctions when I decide like between what tools I'm gonna use and what makes it into my stack.”
- 04 His hack for better Seedance prompt adherence
“The only nuance is and I've compared this one to one over and over and over, and it's been true every time. Translating your prompt to Chinese makes the prompt adherence like ninety percent better.”
- 05 Why he lets AI surprise him instead of fighting it
“I want AI to pull me in ways that I would have never thought of for myself.”
