Pierrick Chevallier | IA

VFX artist. House of David S2 (Amazon).
I test AI video in real pipelines and post what survives a job.
Adobe Firefly Ambassadorr 🚀

[Pierrick Chevallier | IA @CharaspowerAI]

These action scenes have nothing to do with each other. Could Firefly turn them into ONE cinematic showreel? 💥
That was the test: take my footage from different genres and bring it together into 60–90 seconds of action.

I wanted an opening that grabs you immediately, sharp cuts, a few moments to breathe, and a finale that hits hard.

I uploaded my clips into Adobe Firefly’s video editor and gave Quick Cut a prompt describing the direction I wanted. Then I moved into the timeline to refine the sequence and pacing.

That’s the part that interests me as a VFX artist: getting a first assembly to react to, then making the creative decisions that shape the final edit.

Which shot deserves to open the reel? Where should a moment last longer? Where does a cut make the action feel stronger?

Having the first cut and the timeline in the same workspace gives me a starting point I can keep working on.

Watch the showreel below. I’m also sharing the process with the full Firefly interface, from importing the footage to prompting Quick Cut and refining the edit, so you can follow the steps with your own clips. 👇

Try editing your own videos in Firefly:
https://t.co/IK4BWHdHNd

Created in paid collaboration with Adobe as an Adobe Firefly Ambassador.

#AdobeFireflyAmbassadors #Ad #FireflyVideoEditing

[Pierrick Chevallier | IA @CharaspowerAI]

Minimax H3 Image is almost here! 👀🔥
I've had the chance to play with the early access version, and here are a few of my first experiments.

Really enjoying exploring what this model can do, especially when it comes to bringing different visual styles to life. 🎬

Curious to see what everyone will create when it drops!
#H3Image
@Hailuo_AI

[Pierrick Chevallier | IA @CharaspowerAI]

Ultra-realistic first-person blockbuster action, one continuous 25-second POV shot. True first-person camera, visible hands when needed, fast natural head movement, speed ramps, dust, falling debris, bright daylight, premium architectural VFX, grounded physical camera behavior.

Modern downtown city avenue with sidewalks, traffic lights, glass buildings, construction barriers and narrow side streets.

0-5s: POV sprint through a normal city street. Hands occasionally enter frame while running. Everything feels stable and realistic.

5-10s: The pavement begins shifting ahead. Streetlights tilt, signs slide sideways and building façades subtly move out of alignment.

10-15s: A gap opens across the street. POV jumps over it, camera pitches downward in midair, hands visible reaching forward, landing hard on the other side.

15-20s: The architecture becomes impossible. Entire buildings rotate upward while the narrow path beneath the runner remains horizontal. Dust, windows and street objects shift around the moving corridor.

20-25s: The runner races toward one open doorway. The entire city folds vertically behind as the POV dives through the entrance. A massive white light fills the frame at the exact moment the world closes behind.

True POV only, coherent running motion, readable hands, realistic jump physics, stable pathway, premium impossible-architecture VFX, no teleportation, no gore, no text, no logos.

[Pierrick Chevallier | IA @CharaspowerAI]

POV creators, steal this one. 👀🔥
The challenge isn’t making a city fold.
It’s making the viewer genuinely feel like they’re RUNNING through it while reality collapses around them.
True first-person camera. Real jump physics. No teleportation. No cuts.
Basically: parkour inside a broken reality. 💀
Full prompt below 👇
#DreaminaCPP @dreamina_ai

[Pierrick Chevallier | IA @CharaspowerAI]

Pick any movie. Turn it into the LEGO collector set you wish actually existed. 🧱🎬
That’s the prompt.
Swap [MOVIE TITLE], choose the most iconic vehicle, creature or location…
and let GPT Image 2.5 do the rest.
This one is dangerously addictive if you’re a movie fan. 😂

Full reusable prompt below 👇
Create an epic LEGO-style cinematic collector poster inspired by [MOVIE TITLE]. Show the most iconic [vehicle / location / creature / object] in detailed brick-built form, with the main characters as LEGO-style mini figures. Add dramatic lighting, dynamic action, recognizable visual elements from the movie, and a strong atmosphere matching the film’s genre. Make it look like a premium movie poster, colorful, polished, highly detailed, and worthy of a limited-edition LEGO collector set.

What movie are you trying first?

[Pierrick Chevallier | IA @CharaspowerAI]

This is where Seedance 2.5 gets REALLY interesting. 👀
Not another pretty sci-fi shot.
A full 30-second survival-horror trailer with actual pacing, tension, dialogue and a controlled reveal.

The goal was to make it feel directed, not generated.
And Seedance handled the atmosphere way better than I expected.
Full prompt below if you want to test it 👇
#DreaminaCPP @dreamina_ai

[Pierrick Chevallier | IA @CharaspowerAI]

Being able to push AI-generated videos into 4K HDR with Topaz Astra clearly changes the game. 🔥
It’s not just about making the image sharper.

You’re getting footage much closer to something you can actually bring into a professional grading, VFX and finishing pipeline.
That gap between “AI clip” and “production-ready asset” is getting smaller fast.

[Pierrick Chevallier | IA @CharaspowerAI]

GM everyone ☀️
Fresh coffee. Fresh timeline. Way too many ideas already. 😂
AI, VFX, motion design, cinematic experiments…
Let’s see which rabbit hole wins today. 🔥
What are you creating? 👇

[Pierrick Chevallier | IA @CharaspowerAI]

Ultra-realistic sci-fi horror trailer, 30 seconds, original industrial survival aesthetic. ARRI Alexa 35, anamorphic lenses, cockpit close-ups, shaky crash energy, helmet lights, slow push-ins, wide planetary reveals, subtle handheld tension, practical ship lighting, dense fog, metallic interiors, oppressive atmosphere, premium Hollywood realism, non-graphic.

Main crew: four adult space explorers, clearly distinct. Commander: calm adult woman. Pilot: adult man, tense but skilled. Science officer: adult woman, observant and uneasy. Engineer: adult man, rugged and practical. Stable suits, helmets and faces throughout.

Damaged spacecraft, hostile unknown planet, black rocky terrain, toxic mist, strange organic-looking formations, narrow canyons, abandoned alien structure, dim emergency-lit corridors.

0-5s: Trailer hook Violent cockpit turbulence. Warning lights flash. Close-up on gloved hands gripping controls.

Computer voice: “Impact imminent.”

The ship crashes through thick orange clouds.

5-10s: Aftermath Silence. Emergency lights flicker inside the damaged ship. The commander removes her helmet.

Voice-over: “They were never meant to land here.”

10-16s: First exploration The crew steps onto the unknown planet. Huge wide shot: endless dark landscape, drifting mist, towering rock spires. Helmet lights scan strange shapes in the distance.

Pilot whispers: “We’re not alone.”

16-22s: Threat escalation Inside a vast alien structure, flashlight beams reveal wet walls, strange egg-like forms, moving shadows just out of sight, and a faint organic sound echoing through the corridor.

Science officer: “Something is watching us.”

22-27s: Panic beat One crew member turns quickly as a motion tracker spikes. Red emergency light flashes. A door slams shut behind them. Breathing intensifies.

Voice-over: “On this world… curiosity has a price.”

27-30s: Final cliffhanger Close-up on the commander’s terrified face as she raises her light. Reflection in her visor: a towering biomechanical silhouette unfolding in the darkness behind the team.

Hard cut to black.

Original sci-fi horror design, oppressive industrial realism, stable crew continuity, readable geography, unseen threat for most of the trailer, strong atmosphere, premium sound-driven tension, no gore, no visible attack, no text, no logos.

[Pierrick Chevallier | IA @CharaspowerAI]

This is where Seedance 2.5 gets REALLY interesting. 👀
Not another pretty sci-fi shot.
A full 30-second survival-horror trailer with actual pacing, tension, dialogue and a controlled reveal.

The goal was to make it feel directed, not generated.
And Seedance handled the atmosphere way better than I expected.
Full prompt below if you want to test it 👇
#DreaminaCPP @dreamina_ai

[Pierrick Chevallier | IA @CharaspowerAI]

Pick any movie. Turn it into the LEGO collector set you wish actually existed. 🧱🎬
That’s the prompt.
Swap [MOVIE TITLE], choose the most iconic vehicle, creature or location…
and let GPT Image 2.5 do the rest.
This one is dangerously addictive if you’re a movie fan. 😂

Full reusable prompt below 👇
Create an epic LEGO-style cinematic collector poster inspired by [MOVIE TITLE]. Show the most iconic [vehicle / location / creature / object] in detailed brick-built form, with the main characters as LEGO-style mini figures. Add dramatic lighting, dynamic action, recognizable visual elements from the movie, and a strong atmosphere matching the film’s genre. Make it look like a premium movie poster, colorful, polished, highly detailed, and worthy of a limited-edition LEGO collector set.

What movie are you trying first?

[Pierrick Chevallier | IA @CharaspowerAI]

Sometimes the best VFX shot is just ONE impossible moment executed perfectly. ❄️👀
No fight. No destruction montage. No chaos for the sake of chaos.
Just a monk standing still while an entire avalanche has to bend around him.
Simple concept. Massive scale.

I’m sharing the full prompt below — steal it 👇

[Pierrick Chevallier | IA @CharaspowerAI]

Minimax H3 Image is almost here! 👀🔥
I've had the chance to play with the early access version, and here are a few of my first experiments.

Really enjoying exploring what this model can do, especially when it comes to bringing different visual styles to life. 🎬

Curious to see what everyone will create when it drops!
#H3Image
@Hailuo_AI

[Pierrick Chevallier | IA @CharaspowerAI]

GM everyone ☀️
Another day to turn caffeine into pixels, VFX and questionable amounts of rendering. 😂🎬
The creative lab is officially open.

[Pierrick Chevallier | IA @CharaspowerAI]

First test with Minimax H3 Image in early access!

This time, I'm exploring that iconic 90s movie aesthetic: cinematic lighting, film texture and the unmistakable feel of a classic blockbuster.

There's something about that era I just can't get enough of. 🔥

#H3Image @Hailuo_AI

[Pierrick Chevallier | IA @CharaspowerAI]

Ultra-realistic epic spiritual fantasy, one seamless 25-second shot. Slow push-in, low-angle hero framing, snow particles, dramatic mountain daylight, premium energy and ice VFX, blockbuster realism.

A blind monk meditates in a snowy mountain village as a colossal avalanche rushes toward the settlement.

0-7s: Calm mountain village. The monk kneels in meditation at the center while distant rumbling grows.

7-12s: He opens his blind eyes. Snow begins spiraling around him. Villagers look up as a giant wall of snow descends from the ridge.

12-18s: He rises and spreads both arms. A translucent blue energy dome blooms outward, crystallizing the air as camera circles him.

18-25s: The avalanche slams into the dome and parts around the entire village in a towering frozen wave. Snow hangs like sculpted ice on both sides while the monk stands perfectly still at the center.

Readable heroic moment, premium avalanche simulation, spiritual energy VFX, no gore, no text, no logos.

[Pierrick Chevallier | IA @CharaspowerAI]

Sometimes the best VFX shot is just ONE impossible moment executed perfectly. ❄️👀
No fight. No destruction montage. No chaos for the sake of chaos.
Just a monk standing still while an entire avalanche has to bend around him.
Simple concept. Massive scale.

I’m sharing the full prompt below — steal it 👇

[Pierrick Chevallier | IA @CharaspowerAI]

I’d wishlist this game immediately. 👀🔥

Tiny armored knight.
Cursed castle.
Spike traps. Lava. Collapsing towers.

Then the boss fills the ENTIRE screen…
and you beat it by reflecting its own fireball back into its crown.
That’s the kind of 2.5D action I want more of.
made in @LumaLabsAI

Full prompt below 👇
A tiny armored knight enters a cursed side-scrolling castle where floors collapse, spike traps rise from the ground and giant armored beasts block narrow corridors, the knight double-jumps over a lava pit, parries a flaming axe, bounces off a wall, grabs a hanging chain and swings through a collapsing tower before reaching a giant skull-faced boss filling the entire screen, premium 2.5D platformer aesthetic, layered parallax backgrounds, exaggerated silhouette readability, dynamic side camera, sharp hit-stop impacts, orange lava glow against deep violet shadows, ending with the knight reflecting the boss’s fireball back into its crown and the whole castle façade collapsing into the background.

[Pierrick Chevallier | IA @CharaspowerAI]

Running from zombies is easy. Looking COOL while doing it? That's an art form. 🧟‍♂️😂
When the apocalypse hits, I'm not just trying to survive.
I'm making sure every escape looks like a Hollywood action sequence. 🎬🔥
#DreaminaCPP @dreamina_ai

[Pierrick Chevallier | IA @CharaspowerAI]

Your selfie. But you accidentally spawned inside Dragon Ball Z. 💀🔥
That’s it. That’s the prompt.
Upload your photo, keep your identity, switch yourself into full DBZ anime mode… and casually take a selfie with Bulma at Capsule Corp. 😂
Best part? You can remix this with basically any anime universe + character + location.

Full prompt below — steal it 👇
Use the provided photo as reference and keep the same person clearly recognizable, preserving his facial identity, hairstyle, and overall look. Transform him into the Dragon Ball Z anime style and place him in a selfie scene with Bulma inside or in front of Capsule Corp. Keep a believable selfie composition with a slightly wide-angle camera feel, natural arm position, and both characters looking toward the camera. Show Bulma with her iconic anime look, stylish outfit, and confident cheerful expression. Recreate the Capsule Corp environment with recognizable futuristic architecture, clean sci-fi details, bright colors, and a polished DBZ atmosphere. Use strong cel shading, clean line art, vibrant anime colors, and a faithful Dragon Ball Z visual style. Make the final image look fun, dynamic, and fully integrated into the DBZ universe.

[Pierrick Chevallier | IA @CharaspowerAI]

This prompt works because the camera is intentionally TERRIBLE. 💀📱
Bad zoom. Lost focus. Shaky framing. People screaming off-camera.

Then you drop a blockbuster-sized creature into the middle of it.
That contrast between amateur footage + impossible scale sells the illusion ridiculously well.
Try it with your own creature.
Made in @runwayml

Full prompt below 👇