Yoni Attlan

AI-driven creative 🚀 | @Magnific AI Partner | CPP Kling | CPP Runway |CPP CapCut | SCENARIO AI | CPP SEELAB AI

[Yoni Attlan @Yoni_attlan]

Third and last riff on @D_studioproject's recipe-grid trick. This one finally shows its teeth.

A cozy village recipe. Grandmother's instructions for "Feeding Day." Gather the harvest, knead the dough, carry the loaves up the old hill at dusk.

Sweet. Until you see who it's for.

Same method: GPT Image builds the 8-panel grid, Seedance 2.0 reads it as reference and animates every cut. The whole reveal lives in one pull-back.

🎬 GPT Image 2 × Seedance 2.0 on @dreamina_ai
#DreaminaCPP

(3/3 — the recipe was never for us)

[Yoni Attlan @Yoni_attlan]

On @reve I changed one word on a sign. Nothing else moved. Not one pixel.

The shot: a rusted metal plate from my series, red emergency light, bolts, cracked concrete. I rewrote the text on the plate, and the rust, the light, the bolts stayed frozen exactly where they were.

On any other tool, changing that text means re-rolling the whole image. You get your new word and you lose the shot you'd already locked. Region editing means you touch one thing and break nothing.

The job was never "make one beautiful image." It's changing one detail without wrecking the twenty around it. That's the line between a generation tool and a filmmaking tool.

One question for the directors: on a shot like this, what do you change first, the text, the light, the frame?

[Yoni Attlan @Yoni_attlan]

@D_studioproject built a recipe grid that comes to life in Seedance. Wholesome. Kimchi. Sweet.

I took the method and broke the promise.

Same setup: a sourdough tutorial, grandma-coded, every caption telling you to be gentle with the dough.

Then Seedance 2.0 reads the whole grid as a storyboard and animates it.

Watch the dough while it "rests." 🍞

The text stays sweet. The dough doesn't.

🎬 GPT Image 2 × Seedance 2.0 on @dreamina_ai
#DreaminaCPP

(2/3 — one more tomorrow)

[Yoni Attlan @Yoni_attlan]

@LudovicCreator Amazing !!! 😍

[Yoni Attlan @Yoni_attlan]

Been deep in @playVoyage lately. Let it bleed into a test.

Took the recipe-grid → Seedance trick going around and dropped it into an orc's kitchen.

The move: GPT Image builds the storyboard 8 panels, captions, every shot direction baked in. Then Seedance 2.0 reads the whole page as reference and animates every cut from it.

No start frame. The grid IS the storyboard.

The recipe reads sweet. Watch what's actually in the pot. 🍲

(1/3 — new world tomorrow)

[Yoni Attlan @Yoni_attlan]

So here's the test.

Four looks. The machine picked one. Before I reveal which, which would YOU have animated?

[Yoni Attlan @Yoni_attlan]

I gave an AI the final cut of my film. It chose like a coward. And I'm the one who taught it to. 🧵

Full teardown, built in @magnific Spaces 👇

[Yoni Attlan @Yoni_attlan]

Then it got worse.

It optimized exactly what I'd told it to weigh. It didn't fail. It obeyed.

The blandness we blame on the model was written into the brief by a human who looked away. That human was me.

[Yoni Attlan @Yoni_attlan]

@GenIArt_Fr That looks so cool! Great job, the render is super clean. It really makes me want to play the old GTA games.Classic design at its best. Well done.

[Yoni Attlan @Yoni_attlan]

@GenIArt_Fr Merci super recap !

[Yoni Attlan @Yoni_attlan]

Photorealistic, hyper-detailed, indistinguishable from real live-action footage.

An ordinary man in his mid-30s, natural realistic face with real skin texture, pores and natural imperfections, wearing a plain worn grey coat, stands perfectly still at the exact center of a vast ashen-grey glade at blue hour, arms at his sides, eyes open, an expression of calm wonder. He is the only motionless point in the frame. One continuous Steadicam orbit wrapping a full 360° around him at eye level, the man locked dead-center throughout, no cuts, one continuous take, time-ramped to roughly 20% speed.

0–3s: Cold colourless glade, thick real volumetric fog, faded wintry grass moving in the wind. Orbit begins from the front. The man stands motionless, breath visible in the cold air.
3–6s: As the camera passes his side, the ground beneath him greens outward in a spreading ring, fog parting, first wildflowers waking.
6–9s: Behind him, giant luminescent flowers unfurl and rise, glowing spores lifting into the air, ribbons of warm light raking across his real skin.
9–12s: Foxes and deer made of soft living light emerge from the treeline and circle him, drifting wisps orbiting his body, a colossal ancient tree behind him igniting with bioluminescence.
12–15s: The orbit completes back to the front. Full bloom, aurora ribbons across the sky, the whole world alive and moving around him while he remains completely still, the single fixed point.

Shot on ARRI Alexa 35 with anamorphic 40mm lens at f/2.0, realistic natural lighting, true-to-life materials and physically accurate light behaviour, naturalistic god rays through real atmospheric haze, realistic depth of field, authentic film grain, lifelike color. The fantastical elements rendered as photorealistic VFX, seamlessly integrated into live-action, zero CGI look, zero cartoon look. Natural organic motion throughout. 9:16 vertical composition.

The man stays completely motionless the entire shot; only the environment animates around him. Realistic human proportions and natural skin tones, no distortion. No text, no subtitles, no watermarks.

[Yoni Attlan @Yoni_attlan]

Last time @0kncn shared a Seedance 2.0 prompt: a masked shinobi cutting down seven enemies in one continuous orbiting shot. Clean, precise work.

I took the same skeleton and pointed it the other way.

One 360° orbit. Timestamped beats. One take. But instead of a man cutting through the world, a man standing perfectly still while a dead world blooms back to life around him.

His prompt films someone acting on the world. Mine films someone the world acts on.

Same skeleton, opposite gravity.

Built on Seedance 2.0 via @dreamina_ai . #DreaminaCPP

Prompt 👇

[Reve @reve]

What is an image?

[Yoni Attlan @Yoni_attlan]

@reve Waouw i Love it !!!!

[Yoni Attlan @Yoni_attlan]

i filmed three gods from a fishing boat this weekend.
no VFX. no studio. no team. just Seedance and a guy too dumb to stop recording.

took @umesh_ai 's colossus prompt and pushed it somewhere it had no business going: pure POV, a human screaming, and it HAD to look real.

the thing that nearly killed it? the word "phone."

@dreamina_ai · #DreaminaCPP

[Yoni Attlan @Yoni_attlan]

The secret to making AI video feel real isn't more detail. it's less.
kill the cinema. shoot it ugly — grey light, rain on the lens, shaky hands, framing like you panicked. the model stops drawing "movie" and starts drawing "footage."

[Yoni Attlan @Yoni_attlan]

Everyone's testing AI agents to see if they're good.

Wrong question.

The real one: what do they take from you when they actually are?

I gave Octo, @dreamina_ai 's creative agent, a 45-second track. No script. No images. Just sound.

It built a whole world. Characters, locations, a storyboard, four video clips, coherent shot to shot.

Stunning. And completely unusable.

Because the agent didn't remove the boring work. It removed the work that makes you an author, the direction, the choices, the human eye.

Great for snack content. Useless for anything with a soul. For now.

#DreaminaCPP #AIFilmmaking

[Yoni Attlan @Yoni_attlan]

@GenIArt_Fr @LudovicCreator @PamelaFtn @invideoOfficial Ahahaha excellent !! J’adore !