Pinned Tweet
Would you stay in this hotel for ONE night? 👀
For my latest collaboration as an Adobe Firefly Ambassador, I wanted to try something fun for Halloween: create a luxury hotel ad… with a staff you’d probably run away from 😂
Meet THE THIRTEENTH BELL. 🔔
I used @AdobeFirefly Boards to bring the logo ideas, hotel designs and character sheets together before moving into video.
What I really liked was being able to compare everything side by side. The staff had to look terrifying while still looking like they belonged in a five-star hotel.
Having all my references in one place helped me catch inconsistencies, refine the designs and spend less time searching for the right assets.
Here’s the result, with a walkthrough of my process in Firefly.
Try Adobe Firefly: adobe.ly/SpookyFF
#AdobeFireflyAmbassadors #Ad #HowToAdobeFirefly
Logo explorations and character sheets in the next tweets. Full workflow below 👇
🤝 Paid partnership
When it comes to motion design and kinetic typography, Minimax H3 is still the GOAT. 👑🔥
The way it handles text animation, transitions and complex visual compositions is seriously impressive.
And the craziest part?
It's OPEN SOURCE. 🤯
I'm dropping the exact reference images I used to create this video below. 👇
GM creators ☀️
The week is flying by, but the render queue definitely isn’t. 😂
Still plenty of time for AI, VFX, motion design and a few cinematic experiments before the weekend.
Let’s cook. 🎬🔥
What are you creating today? 👇
Pierrick Chevallier | IA retweeted
I still don’t understand why more people aren’t using Minimax H3 for dialogue scenes. 👀
The voice delivery, pauses, facial reactions, timing between characters…
There’s a LOT more here than simple image animation.
H3 can actually make characters feel like they’re performing a scene.
Seriously underrated use case. 🎬🔥
Every window reveals a different ocean.
A train carriage stays intact while the world outside changes from daylight to the deep sea.
The VFX challenge: make the passing pillars control the transition—and the light prove it happened.
Full prompt below 👇
#DreaminaCPP @dreamina_ai
Ultra-realistic cinematic science-fiction wonder, one continuous 25-second interior tracking shot, landscape 16:9. Premium passenger-train set, natural acting, extraordinary exterior VFX, controlled evolving interactive light.
One adult woman in a dark green coat seated beside the window. A few quiet adult passengers deeper in the carriage. Stable faces and costume.
An elegant train carriage with broad windows separated by opaque structural pillars. Camera travels slowly down the aisle parallel to the windows.
The physical carriage never changes. Exterior environments change only while an opaque interior pillar completely covers the relevant window area. Three exterior stages: coastal daylight, shallow underwater reef, deep ocean. Interior illumination changes to match each new exterior. This is a fantastical train journey, not an accident.
0–6s: Track beside the seated woman as bright coastal scenery moves outside. A large wave rises beyond the window, briefly higher than the horizon. She turns toward it. Camera passes behind the first broad pillar.
6–12s: The pillar clears onto the same carriage now travelling beside a sunlit underwater reef. Fish pass outside the sealed windows. Moving blue caustics touch her face, coat and seat. Her surprise becomes quiet fascination.
12–17s: Continue tracking past another pillar. Reveal dark deep-ocean water outside. Caustics disappear, leaving cool dim illumination and warm carriage lamps. Tiny bioluminescent organisms drift past.
17–22s: A vast pale whale crosses slowly beside the train, close enough for one enormous eye to align with her window. Its movement is calm; the train and whale briefly travel alongside each other.
22–25s: Camera settles into her profile with the whale’s eye beyond the glass. She lifts her hand toward the window without touching it. The whale continues past, its immense skin filling the exterior. Hold.
Stable carriage geometry, pillar-motivated changes, matched interior lighting, coherent exterior movement, no water entering, no random scenery morphing, no text or logos. Train rhythm persists while exterior sound remains muted behind sealed glass.
Pierrick Chevallier | IA retweeted
Would you stay in this hotel for ONE night? 👀
For my latest collaboration as an Adobe Firefly Ambassador, I wanted to try something fun for Halloween: create a luxury hotel ad… with a staff you’d probably run away from 😂
Meet THE THIRTEENTH BELL. 🔔
I used @AdobeFirefly Boards to bring the logo ideas, hotel designs and character sheets together before moving into video.
What I really liked was being able to compare everything side by side. The staff had to look terrifying while still looking like they belonged in a five-star hotel.
Having all my references in one place helped me catch inconsistencies, refine the designs and spend less time searching for the right assets.
Here’s the result, with a walkthrough of my process in Firefly.
Try Adobe Firefly: adobe.ly/SpookyFF
#AdobeFireflyAmbassadors #Ad #HowToAdobeFirefly
Logo explorations and character sheets in the next tweets. Full workflow below 👇
🤝 Paid partnership
Okay, I think I just found my new favorite fighting game character. 💀🔥
Forget the tier lists.
I'm maining this absolute legend from day one. 😂🎮
🤖 Made with AI
Pierrick Chevallier | IA retweeted
This is the kind of AI footage that could fool you for the first few seconds. 👀📱
No cinematic camera. No polished Hollywood look.
Just a random smartphone recording… until gravity starts behaving like it has completely given up. 💀
The amateur realism is what makes this VFX concept so fun.
I'm sharing the exact 20-second prompt below. 👇
#DreaminaCPP @dreamina_ai
Starlight Precise 2.6 from @topazlabs is becoming an essential part of my AI video workflow. 🔥
Taking an 8-bit SDR generation → 4K HDR 10-bit completely changes what you can do with the footage afterward.
Better grading latitude. Better finishing. Much more viable for serious post-production.
Generation is only step one now.
The finishing pipeline matters just as much. 🎬
GM everyone ☀️
Halfway through the week and still plenty of pixels left to push. 🎬
AI, VFX, motion design, cinematic shots…
Let’s make something today that deserves the full-screen treatment. 🔥
What are you working on? 👇
Wait… ONE IMAGE to a fully rigged 3D character in about 60 seconds?! 🤯
Viggle just introduced JST-2.
Generate a character.
Animate it with text or video.
Turn another image into a 3D world.
Drop your character inside.
All powered by the SAME foundation model.
And we're not talking about a video you can only watch.
We're talking about 3D content you can actually interact with.
This is getting seriously interesting. 🔥
🤝 Paid partnership
Today we are releasing JST-2, the world's first full-stack physics-native world foundation model.
Many world models learn in pixel/video space. We went back to first principles and built the entire model stack in 3D.
Generate a 3D scene in seconds and expand it in real time. Create characters with motion built in, all through one general architecture.
This is the kind of AI footage that could fool you for the first few seconds. 👀📱
No cinematic camera. No polished Hollywood look.
Just a random smartphone recording… until gravity starts behaving like it has completely given up. 💀
The amateur realism is what makes this VFX concept so fun.
I'm sharing the exact 20-second prompt below. 👇
#DreaminaCPP @dreamina_ai
Ultra-realistic viral smartphone footage, 20 seconds, raw daytime street video. Strong hand shake, autofocus breathing, digital zoom artifacts, natural traffic audio, frightened English reactions, no cinematic polish.
Large city intersection with cars, bicycles, pedestrians, fountain and traffic lights.
0-4s: Normal intersection. The filmer records a strange floating plastic bag. FILMER: “Why isn’t that falling?”
4-8s: Loose papers, leaves and water droplets begin rising vertically. People notice and step back.
8-13s: Bicycles, café chairs and lightweight objects slowly lift from the pavement. Water from the fountain streams upward into the sky. FILMER: “Oh my God… gravity’s gone!”
13-17s: Parked cars begin lifting several feet into the air, gently rotating while people move safely away. Camera struggles to frame everything.
17-20s: Massive wide reveal: the entire intersection is suspended in midair around the buildings while the filmer backs away. FILMER: “RUN—what is happening?!” Hard cut.
Progressive gravity inversion, realistic inertia, believable amateur reactions, no injuries, no text, no logos.
Freeze the hit. Move the camera.
That’s the bullet-time brief.
Pick a moment worth inspecting: the glove compressing, the face reacting, the sweat leaving.
An orbit around someone standing still won’t give you the same payoff.
Full prompt below 👇
Ultra-realistic premium boxing commercial, 20 seconds, landscape. Dark training gym, hard overhead light, sculpted sweat highlights, restrained handheld setup transitioning into a precise bullet-time arc.
One adult female professional boxer in deep-red gloves and black training wear. One adult coach holding a thick black focus mitt. Controlled professional training, no injuries.
Normal speed before and after the featured punch. At glove-to-mitt contact, the physical action slows to near-frozen time while the camera arcs sixty degrees around the contact point. Glove, mitt, sweat and body positions progress only minimally during the arc. Resume from exactly that action state.
0–5s: Close side tracking. Boxer delivers two quick controlled punches into the coach’s mitts, resets her feet and slips beneath a simulated counter. Sharp breathing and readable weight transfer.
5–8s: Medium three-quarter angle. She plants the rear foot, rotates through the hips and drives a powerful right cross into the presented mitt. Slow time at the instant glove padding visibly compresses.
8–13s: Bullet-time arc around the contact. The compressed glove dominates foreground; her focused face and shoulder rotation remain readable. Sweat droplets hang in the same spatial arrangement as the camera moves past them. Coach’s arm is visibly braced.
13–17s: Resume normal speed. The mitt recoils, droplets continue their original trajectories, her fist retracts and her weight settles back into stance. No repeated punch or positional reset.
17–20s: Hold close on her face behind raised gloves. Coach quietly says: “Again.” She gives one small nod. Cut as she begins the next step.
Precise frozen-action continuity, limited motivated camera arc, natural athletic anatomy, visible glove compression, coherent droplet trajectories, no duplicated limbs, no text or logos. Sound drops into a low restrained tone during the arc, then restores the impact and breath on release.
Your favorite video game character is one selfie away. 📸🎮
Upload your portrait.
Pick an iconic character.
Let GPT Image 2.5 handle the rest.
For this test, I went adventuring with Lara Croft… and the result is way too convincing. 😂🔥
I’m sharing the prompt so you can swap Lara for ANY character you want.
Full prompt below 👇
Use the uploaded portrait as the main character reference and keep his face, hairstyle, and identity clearly recognizable. Create a spectacular cinematic selfie featuring this man and Lara Croft together, both close to the camera in a dynamic wide-angle selfie composition. The man should look excited and confident, while Lara Croft stands beside him with her iconic adventurous attitude, ready for action. Place them in a dramatic action-adventure setting such as a lush jungle, ancient ruins, or a mysterious temple, with cinematic lighting, dust in the air, subtle motion, and an epic exploration mood. Make the selfie feel spontaneous but visually powerful, with realistic facial details, natural expressions, immersive depth, and a high-end blockbuster look.
Pierrick Chevallier | IA retweeted
Okay, that's it. 😂💀
Next time there's a protest in France, I'm taking the Tesla Roadster.
Apparently, this thing has a feature for EVERY situation. 🇫🇷⚡️
GM everyone ☀️
Tuesday, October 20.
The creative stack is loaded:
AI video. VFX. Motion design. A few questionable FPV ideas. 😂⚡️
Now let’s see which one takes over the day.
What are you creating today? 👇
Five-star service. Absolutely terrifying staff 😂
Next, I created five employees: the hotel director, bellhop, housekeeper, chef and bar manager.
This took a few iterations. I pushed the horror too far at first and ended up with torn, dirty uniforms. Definitely not the luxury hotel I had in mind.
The balance: haunted faces, immaculate outfits.
I kept their character sheets together in Firefly Boards to compare the faces, costumes and colors before using them as video references.
Who would make you leave before checking in? 👇
I wanted to make a luxury hotel ad. I think I made sure nobody would book a room 😂🔔
This was a fun way to explore how branding, character design and video can come together around one idea—with Firefly Boards keeping all my references in one place.
Would you check in… or turn around at the door? 👀
If you enjoyed this breakdown, like and repost the first tweet ❤️🔁
Follow @CharaspowerAI for more AI filmmaking experiments, prompts and behind-the-scenes workflows.








