Sergio Cardona  Creative Technologist

Becurious

Generated with AI. Directed, finished and signed off by people.

Curious by design.
Human by default.

As a creative technologist, I work where the idea meets the tools. I collaborate on pitches, build the AI workflows and prompt systems behind film, imagery and packaging, and train teams to use them. I keep up with what new technology can do and the laws that shape it, then push the work as far as it can go.

Work19

  1. 01EnbridgeHi, BobcaygeonPitch film1:51
  2. 02New York LifeAnthemPitch film1:15
  3. 03SkechersSoccer, Kids, Work, RunningPrompt workflow · social2026
  4. 04DuracellMessi brand activationFilm · pack · character system0:30
  5. 05Black Box WinesOnce Upon a WineCTV spots · idea to delivery0:30
  6. 06Santa Cruz SkateboardsSecond LifeStory film0:59
  7. 07Buck MasonFall knit filmPitch film0:30
  8. 08King Ranch Saddle ShopWorn in, never worn outPitch film0:32
  9. 09East EastBeauty still lifeProduct imageryStill
  10. 10Bio.meGut health, lighter than airProduct loops5 scenes
  11. 11KiaHalloween & what's possiblePitch · creator spot · film concepts0:22
  12. 12GerberGet'Ems! SqueeziesFlat lays · Figma Weave0:06
  13. 13Dr Teal'sWhere 3D ran outVFX rescue · real footage + AI3 tests
  14. 14G-ShockBuilt for the coldSpec ad0:22
  15. 15Game of Thrones: DragonfireA new dragon arrivesStock footage + AI0:16
  16. 16Amazon LunaMake the pitch feel playablePitch · Gen AI key art0:10
  17. 17Node-based editingBuilt for designers, by designersOur platform · vibe-codedv0.6
  18. 18AlcaChinoA coffee cocktail brand, built from the ground upBrand partnership · Packaging0:34
  19. 19SunBuzzClassic cocktail energy, reimaginedNew venture · Packaging · Social0:10
    No. 01Enbridge · Pitch

    Hi,Bobcaygeon

    Pitch concept | Community storytelling

    Enbridge needed a way to introduce a new gas-delivery model to a Canadian community without reducing a meaningful local transition to a technology story. Hi Bobcaygeon was a pitch built around residents, workers, and the community itself. The idea framed Enbridge as a partner, showing how a more efficient service could be introduced with care, clarity, and respect.

    My role I helped shape the creative approach and develop the spot, turning a complex infrastructure conversation into a people-first story designed to earn understanding and support.

    DeliverableBrand film · sizzle 1:51
    CraftPhotoreal storyboardsAI video generationVoiceover & editCompositing & grade
    Snow-covered house with porch light on
    Kids playing hockey on a frozen pond at sunrise
    Teacher at a chalkboard
    Enbridge worker in frost on the phone
    Enbridge patch on a hi-vis jacket
    To fuel a community is to be trusted by it, and trust is not given.From the voiceover
    House at night under a warm protective dome
    Nice to heat you.
    No. 02New York Life · Pitch

    One life,not line items

    Pitch concept | Brand storytelling

    New York Life believes in people. The opportunity was to make that idea tangible before a frame was ever shot. Anthem turned a broad brand promise into an emotional, human story and gave the client a clear picture of how the campaign could look, move, and connect.

    My role I helped develop the creative approach and visualization, shaping the idea into a compelling presentation that could move from pitch to production.

    DeliverableAnthem film · 1:15
    CraftAI casting & performanceScene generationEdit & finishing in After Effects

    Turningpoints

    Six life moments, one visual language: soft window light, close lenses, real-feeling homes. Consistency across faces and rooms is what makes generated work read as a film instead of a mood board.

    Couple reacting to a pregnancy test
    Hands exchanging a ring at a wedding
    Family meeting with an advisor at a kitchen table
    Woman reading a letter in her kitchen
    Parents cheering in the bleachers
    Nobody plans their life in a straight line.Opening line
    No. 03Skechers

    Speed ofsocial

    AI workflow | Prompt engineering | Team training

    During the World Cup, social conversation moves faster than a traditional production pipeline. Skechers needed a way to respond at the speed of culture and become part of the conversation. I built a node-based workflow using Figma Weave, prompt systems, and structured creative controls so the team could develop social-ready ideas quickly while maintaining quality, consistency, and brand relevance.

    My role I led prompt engineering, created the workflow, trained the team, and helped troubleshoot the system as it evolved. The method gave the team a repeatable way to make work faster and opened new opportunities with the client.

    My roleBuilt the original workflowTrained the team: node-based editing & promptingTroubleshooting support
    LaunchesSoccer · Back to schoolWork · Running
    DeliverablesSocial static & motionPaid digital, every size

    01 · SoccerA seat atthe table

    A new player in a category built on Nike, Adidas and Puma, with the World Cup landing in its own backyard. Rather than outspend the giants, the team leaned into being the challenger with real-time reactive creative. Over 200 relevant assets in weeks stood out in a sea of pink boots.

    Harry Kane and player posts
    Bored of pink boots posts
    USA match-day posts
    Mexico match-day posts

    02 · Back to schoolRecessready

    Toddlers to teens, parents choosing durability and kids choosing what's trending. Instead of one campaign stretched thin, the workflow produced age-specific variations at speed: one back-to-school story, told fluently to every audience.

    Dragon sneaker illustration and Recess Ready posts
    Recess Ready kids posts
    Cat illustration lifestyle posts
    First Day Fit and Coffee Vibes posts

    03 · WorkThree brands,one pipeline

    A Skechers Work collaboration with John Deere and Goodyear, launched to work, outdoor and fashion audiences at once. Agents trained on all three brands' guidelines checked assets as they were made, so the work kept moving without a failed brand review.

    John Deere and Goodyear work boots
    Max Cushioning and Slip-ins posts
    Work boot scenes
    Indoor hospitality footwear posts

    04 · RunningCredibility,not noise

    A signature shoe entering one of the most crowded, credibility-driven categories. The push earned trust with runners on their own terms, grounded in the shoe's real technical story, and stayed accurate from tech sheet to social ad to reel.

    Aero Burst glow product posts
    Aero Burst night running scenes
    Aero Elite Built for Speed
    Aero Elite product posts
    No. 04Duracell

    Powerwithin

    Character system | Packaging design | Node-based editing

    Duracell’s iconic Bunny had to stay unmistakably itself in every frame. This pitch showed how AI and node-based editing could turn a slow, expensive traditional pipeline into a flexible character system: any approved pose, season, costume or setting, from 3D-style development to social-ready assets, without losing the Bunny’s identity.

    My role I developed the prompt-engineering and node-based workflow, protecting the Bunny’s signature anatomy and character while making fast, scalable production possible.

    DeliverablesBrand activation film · 0:30Uniform & limited-edition packBunny Perfect Pose SystemPose & costume library
    CraftAI video · Topaz upscaleNano Banana 2 · Nano Banana ProNode-based canvas
    Black and copper number 10 jersey
    Limited-edition Duracell Power Boost AA 8 pack
    Battery with the number 10 tattoo design

    The PerfectPose System

    A mascot only works if it is the same bunny every time. Every generation is checked against a locked anatomy: ears and back fur, face, jersey, battery and back mount, hands, feet and shorts. The pose can change. The bunny can't.

    Node canvas overview of the bunny pose system
    1. Pose reference lab with pose sheets, a reference photo and a pose geometry extractor
      Pose reference lab

      Any reference, from a gesture sheet to a photo of a mascot on a bench, is read into structured pose geometry: camera, framing, head, torso, each arm and hand.

    2. Bunny anatomy matrix of approved locks
      Anatomy matrix

      Approved locks for every part of the character. The pose node controls the body; the locks have final say on identity.

    3. Costume lab generating twenty bunny variants
      Costume lab

      One approved bunny, twenty wardrobes: firefighter, site crew, cave explorer, holiday sweater, Halloween cape.

    4. Finishing chain: inner ear fix, background replace, hi-res and edge-safe cutout
      Finish

      Inner-ear correction, background replacement, hi-res pass and an edge-safe cutout, ready to drop onto any layout.

    Bunny running
    Bunny pointing
    Bunny with arms crossed
    Bunny as a firefighter
    Bunny in hard hat and hi-vis holding a battery
    Bunny as a cave explorer with a flashlight
    Bunny in a holiday sweater with gifts
    Bunny in a Halloween cape with a pumpkin bucket
    Approved ear and back-fur lock
    Hand and arm reference sheet
    No. 05Black Box Wines

    Once upona wine

    Hybrid production | VFX | Prompt engineering

    Black Box Wines became our first full hybrid AI and VFX production. The challenge was to create polished, dimensional animation without relying entirely on a conventional and costly 3D VFX pipeline. We developed a prompt-engineered system that carried visual consistency from one scene to the next, then combined AI-generated material with After Effects and traditional finishing.

    My role I developed the prompt and production workflow, helping create a seamless hybrid process that kept the work in-house and reduced the VFX budget by roughly 40 percent.

    DeliverablesCTV · 16:93L :30 & :15 · Tetra :15Fireworks :13
    CraftNano Banana Pro · SeedanceFigma Weave · JSON master promptGreen screen · After Effects
    Wine bottle and glasses rising from an open book
    Tetra packs bursting from a paper fan
    Pop-up flamingo and umbrella around the Tetra trio

    Green screen,one master prompt

    Books and packs were generated on green, then every plate was composited in After Effects, where the movement is controlled frame by frame. Between AI passes, one JSON master prompt kept every transition the same: one continuous turn, a subtle tilt back, a clean landing. One prompt to rule them all.

    Green plateBook open · for After Effects
    Figma Weave canvas: the initial cover on green with the prompt note
    Figma WeaveInitial cover
    master_prompt.json
    {  "task": "video-to-video motion refinement",  "primary_goal": "preserve the successful look, but correct the book motion",  "motion_priority": {    "description": "The book moves in one continuous motion from frame one to frame two.",    "movement_type": "single start-to-end transition",    "rotation": {      "instruction": "As the book tilts back slightly, it also rotates smoothly around its own vertical center axis.",      "style": "controlled turntable-like rotation with a subtle backward pitch",      "match_reference_rotation": true    },    "tilt": {      "instruction": "The book should tilt back only a little, like the reference video.",      "amount": "subtle",      "limit": "never tilts all the way flat or horizontal"    },    "opening": {      "instruction": "While rotating and tilting back, the book begins to open naturally.",      "style": "smooth, elegant, physically believable"    },    "landing": {      "instruction": "The book lands in the final space and final orientation shown in frame two.",      "hold_final_pose": "hold the final position for the last few frames"    }  },  "position_and_path": {    "instruction": "The book should travel directly from the first frame position to the final frame position.",    "no_extra_motion": true,    "no_down_then_up_motion": true,    "no_flattening_motion": true,    "no_return_motion": true  },  "camera": {    "movement": "locked-off camera",    "framing": "stable",    "zoom": "none",    "shake": "none"  },  "negative_prompt": [    "do not make the book go flat",    "do not make the book lay horizontal",    "do not make the book drop down and come back up",    "no bouncing",    "no bobbing",    "no floating upward",    "no exaggerated tilt",    "no flipping",    "no wobbling",    "no extra midpoint pose",    "no reset motion",    "no changed final position"  ]}
    Figma Weave graph: one initial cover feeding Tetra, 3L and Chardonnay frame chains
    One cover, every packTetra · 3L · Chardonnay frames
    Tetra · Stories & Reels · 9:16 · 0:15

    Cabana,cannonball ready

    The Tetra goes poolside in a vertical cut made for the feed. It strikes a pose, takes the plunge in a burst of splash, surfaces and floats off in a ring, with the line landing on the last frame: threeish glasses of premium, resealable.

    1. 01Pose
    2. 02Splash
    3. 03Cannonball
    4. 04Surface
    5. 05Float
    Blender clay preview: the Tetra pack sitting in a pool float
    ClayBlocked in 3D
    Figma Weave canvas: clay render, pool plate and real packaging wired into Nano Banana Pro and a Kling video
    Figma WeaveClay + real pack → frame
    Story Frames group: storyboard and prompts driving each Cabana frame
    Story framesBoard → five frames
    No. 06Santa Cruz Skateboards

    Secondlife

    Experimental film | Personal storytelling

    Second Life explores what it feels like to learn AI from the inside: awkward at first, full of trial and error, and satisfying when the right result finally lands. The piece connects that feeling to skateboarding, a world I know well. Both require repetition, instinct, failure, adjustment, and the willingness to keep trying until something clicks.

    My role I created the concept as an exploratory piece, using skate culture and AI experimentation to tell a personal story about process.

    DeliverableBrand film · 0:59
    CraftAI video generationOriginal score · MiniMax MusicAfter Effects edit
    Snapped Santa Cruz deck on concrete
    Skater riding a bowl at golden hour
    Looking up out of a drained pool
    Skater pushing down a street with Santa Cruz title
    No. 07Buck Mason · Pitch

    From the rackto Route One

    Pitch concept | Fashion storytelling

    Buck Mason’s visual language is built on lived-in texture, timeless clothing, and recurring relationships with its models. The opportunity was to explore an AI-supported fashion concept without losing that authenticity or the people behind it. The pitch paired Buck Mason’s vintage sensibility with a responsible production model that respected talent, protected likeness, and kept the work from feeling synthetic.

    My role I helped develop the pitch and visual approach, showing how AI could extend a fashion story while keeping the brand’s human character at its center.

    DeliverablesFashion film · 0:30Lookbook stills
    WorkflowsClothing to modelModel angles & locationsClothing transfer
    ToolsGemini 3 Pro in Runway · VeoFirefly upscale · After Effects
    Studio lookbook, front
    Studio lookbook, high angle
    Studio lookbook, three-quarter

    Same knit,every angle

    Node-based workflows lock the garment first, then move the model through angles and locations. The sweater's cable pattern survives the trip from studio sweep to sun-flared windshield.

    Point-of-view shot driving a classic convertible on the coast

    Two graphs,one wardrobe

    The workflows behind the pitch, in Figma Weave. The first takes one model in one outfit through seven shots: low, high and medium angles, three-quarter, a fabric close-up, a Paris subway and a vintage Mustang. The second takes a flat lay of real product and four face references and dresses each model in the same shirt, chinos, sneakers and tote.

    Figma Weave graph: one model reference feeding seven prompt and Nano Banana Pro chains for angles, a fabric close-up, a subway and a convertible
    Model angles & locationsOne reference · seven shots
    Figma Weave graph: a flat lay of shirt, chinos, sneakers and tote plus four face references feeding full-body looks
    Clothing transferFlat lay + four faces · same outfit
    No. 08King Ranch Saddle Shop · Pitch

    Worn in,never worn out

    Product storytelling | Prompt engineering

    King Ranch Saddle Shop needed product imagery with a point of view: tactile, crafted, and distinctive enough to match the quality of the goods themselves. A signature wide-angle photographic look had become difficult to reproduce at scale. I developed a prompt-engineered approach that recreated the feeling of that lens across a large product catalog.

    My role I created the visual workflow and prompt system that helped secure the continuation of a 666-product shoot. The work continues to evolve across new campaigns and product stories.

    DeliverableBrand film · 0:32
    CraftAI scene & product generationMacro material studiesEdit, grade & score
    Macro of saddle stitching on canvas
    Macro of worn leather
    Leather travel bag on a workshop bench
    Woman in a western shirt at a ranch fence
    Man pulling on a quilted vest in a tack room
    Rancher walking a field with two hunting dogs
    No. 09 · East East

    Still,life

    A study in texture, material, light, and the small details that make a beauty image feel alive. The work explores how AI-assisted image making can support still-life direction while retaining the precision, restraint, and sensuality that beauty brands require.

    My role I explored the visual language, developed prompts and references, and helped shape a repeatable approach to highly controlled beauty imagery.

    Skincare and haircare bottles in a studio still life
    Node canvas with a Composition Studio and a Retouch Studio
    01 · Composition studio

    A product bible of six labeled packs, a camera and mood reference, and a studio sweep go into Nano Banana 2. The prompt allows no invented products and no borrowed details: shape, cap, pump and finish come only from the bible.

    02 · Retouch studio

    Nano Banana Pro strips every line of copy, then the true labels are rebuilt from flat packshots of Oribe, Paula's Choice, Allies of Skin, Medik8 and Augustinus Bader, word for word, with nothing paraphrased.

    What happened next

    From one shotto a partnership

    Matching the shoot turned into an ongoing relationship. We became the digital photographer for East East’s client Dermstore, and we’re now producing their holiday campaign. The partnership continues.

    Now · Dermstore holiday campaign · images coming soon
    No. 10Bio.me

    Gut health,lighter than air

    Packaging to social system | Prompt engineering

    Bio.me needed a way to turn packaging design into a living social-content system quickly, consistently, and with enough flexibility to keep pace with the channel. The approach transformed existing brand assets into an adaptable set of social-ready images and ideas without starting from zero every time.

    My role I directed the creative approach, developed the prompt system, and built the workflow that allowed the team to generate and refine content at the speed of social.

    DeliverablesSocial loops · 4:5Packaging imagery
    RangeDaily Prebiotic FiberKids Daily · Psyllium Husk
    01Citrus
    02Syrup
    03Pour
    04Balance
    No. 11Kia · Pitch

    What ifit's possible?

    Social content | Asset transformation

    Kia had strong existing assets. The challenge was to make them feel newly alive for social: faster, more surprising, and more culturally relevant. The project explored how AI could transform brand materials into fresh creative content while keeping the core identity intact.

    My role I helped develop the creative direction, prompt engineering, and workflow for reimagining existing assets into engaging social material.

    DeliverablesCreator spot · 9:16Storyboards · film conceptsBackground replacement
    VehiclesEV6 · Sorento
    Halloween

    Dressedto haunt

    Two AI creators, Luna and Tio, take on a Kia Halloween car challenge: cardboard bat wings, fangs on the grille, spiders across the hood and a trunk full of candy for trick-or-treaters. Native vertical, shot like real UGC, made entirely with AI.

    You're not ready for what we built.Luna, AI creator
    Storyboard: garage opens, the Kia reverses out in fog
    Story 01The Portal
    Storyboard: trunk-or-treat station with a skeleton driver
    Story 02The Candy Crew
    Storyboard: ghost sheets swaying inside the cabin
    Story 03Haunted Ride
    1. Objectives

      Understand the goal and the platform it lives on.

    2. Pre-vis

      Storyboard and sketch every beat before generating a frame.

    3. Prompt

      Write the shot twice: as prose and as structured JSON.

    4. Compose

      Finish AI elements into the final film or key art with traditional design tools.

    Idea 01

    Desert heat,city night

    An EV6 crosses the Mojave in golden light. The dust off its tires rises, turns to smoke and opens onto a neon Tokyo alley, wet pavement and manhole steam carrying the transition.

    Pencil sketch: EV6 in the desert and in a neon alley
    The prompt

    Prose,then JSON

    Every idea is written as a traditional prompt and as a JSON prompt: subject, environment, transition, camera and output, each set on its own line, so a shot can be changed one setting at a time.

    JSON prompt for the desert-to-city shot
    Idea 02

    The car'sGlamBot moment

    Borrowed from the red carpet: a robotic arm sweeps a spotlight over grille, headlights, cabin and taillights. Not a walkaround. A performance.

    Pencil sketch: robotic arm spotlighting the car
    Idea 03

    Dreamcommute

    Plug in, smile, 100%. A calm charge-to-road story in two looks, everyday suburbia and a near-future city, so the client can pick the world.

    Pencil sketch: charging, plug-in, 100% battery

    Any street,no shoot

    One packshot becomes a season of locations. The car stays the real car: paint, lights and badge untouched, while the world changes around it.

    EV6 packshot on white
    00Original
    EV6 on a sunny city street
    01Summer
    EV6 on an autumn street
    02Fall
    EV6 on a grey, wet street
    03Overcast
    EV6 on a neon-lit street at night
    04Neon
    Original EV interior photo
    Original photographyCabin
    AI-modified EV interior with dark trim
    Modified with AINew trim
    9:41
    getemsOriginal audioFollowing•••

    256 likes

    getems A little fuel, a lot of flavor ✨ Meet NEW Squeezies from Get’ems!

    View all 15 comments

    No. 12Gerber · Get'Ems!

    Twist,then squeeze

    AI-assisted stop motion | Art direction

    Gerber needed stop-motion food photography at a moment when the fruit required for the shoot was out of season. Traditional post-production could move objects, but it could not achieve the photorealism and tactile quality the concept demanded. We used AI as part of a controlled hybrid process, working with stencils and visual references to guide object placement and preserve the handmade feel of stop motion.

    My role I developed the prompt-engineering and visual-control process, using references to guide the system toward a specific physical result.

    DeliverablesSocial · 9:16Flat-lay key art
    CraftFigma WeaveNano Banana 2
    FlavorsCherry LimeadePeach Apple TwistApple Sweet Potato
    A tool used the way it wasn't meant to be

    Stencils,not guesswork

    Figma Weave was never built to art-direct a flat lay, so we bent it: a black-and-white stencil marks exactly where every slice lands, so the art director sets the layout and the model only fills it. The colour card is cropped to the flavor in play and a master prompt locks camera, light and surface. Swap the fruit and the colour, and the next flavor matches the last.

    See Gerber on Instagram ↗
    Node canvas: fruit reference, layout mask, flavor colour card and master prompt feeding Nano Banana 2
    Flat lay of cherries and lime wedges
    Cherry LimeadeFlat lay
    Flat lay of apple slices and sweet potato
    Apple Sweet PotatoFlat lay
    The launch

    Three flavors,one squeeze

    Squeezies launched in April 2026 in three flavors: Cherry Limeade, Peach Apple Twist and Apple Sweet Potato Splash. Made with real fruits and veggies and no added sugar, each pouch is a smoothie straight from the shelf or an ice pop from the freezer.

    9:41
    getemsOriginal audioFollowing•••

    256 likes

    getems Cherry Limeade 🍒🍋

    View all 15 comments

    01 · Cherry Limeade
    9:41
    getemsOriginal audioFollowing•••

    256 likes

    getems Peach Apple Twist 🍑🍎

    View all 15 comments

    02 · Peach Apple Twist
    9:41
    getemsOriginal audioFollowing•••

    256 likes

    getems Apple Sweet Potato Splash 🍎🍠

    View all 15 comments

    03 · Apple Sweet Potato Splash
    No. 13Dr Teal's

    Where 3Dran out

    Hybrid VFX | Photorealism

    When a traditional VFX partner could not achieve the photorealism required for a key creature moment, the team needed another route. By combining AI with established production tools, we found a more convincing way to integrate the lizard into the spot and complete the visual idea.

    My role I helped develop the hybrid workflow and prompt-engineering approach that pushed the work past the limitations of the original VFX solution.

    DeliverablesEnd frame · 16:9Lizard VFX shots
    CraftLive-action platesNano Banana ProLuma Modify · Kling 3
    Broadcast · 16:9 · 0:30Watch on YouTube ↗
    Mobile · 9:16 · 0:15View the reel ↗
    Motion testBlue set
    FinalEnd frame on product
    Two node graphs: Lizard leaf Animation 1 and 2
    01 · Swap the actor

    Nano Banana Pro drops the anole into a live-action plate and changes nothing else in the frame. Luma Modify carries the swap through the whole move.

    02 · Correct, then animate

    The tail is lengthened to match the reference, then Kling 3 animates between a first and last frame, with a negative prompt holding colour, background and set in place.

    No. 14G-Shock · Spec ad

    Built forthe cold

    Creative exploration | Prompt engineering

    An exploratory study of what AI tools can do when visual precision, texture, and product energy all have to work together.

    My role I used prompt engineering and visual experimentation to test the creative range of the tools and build a foundation for future product storytelling.

    DeliverablesSpec spot · 9:16 · 0:22
    CraftAI filmEdit, sound and finishing

    Five beats,one night

    The spot reads left to right as a single cold night: watch, mirror, wheel, road, ridge.

    G-Shock in snow, eyes glowing behind it
    01Watch
    Driver seen in the rear-view mirror
    02Mirror
    Gloved hand on the wheel
    03Wheel
    Car in fog on a night road
    04Road
    Wolf howling under a full moon
    05Ridge
    No. 15Warner Bros. Games

    A new dragonarrives

    AI-assisted retouching | Campaign production

    Game audiences are understandably cautious about AI. This project took a focused, practical approach: use the technology to solve small but meaningful production problems efficiently. AI was used to refine existing stock photography and footage, such as removing distracting details from a dragon’s eye, work that would otherwise require hours of traditional retouching.

    My role I helped apply AI as a precise production tool, supporting a successful campaign without replacing the existing creative process.

    ClientWarner Bros. Games
    DeliverablesSocial tease · 9:16 · 0:16
    ApproachStock footage + AI modification
    View the post ↗

    Stock in,dragon out

    The comparison. On the left, the licensed stock clip we started from. On the right, the same eye after AI: smoke, scales, the colour and the reveal rebuilt around it for a vertical feed.

    BeforeStock footage · iStock
    AfterAI · 9:16

    Modify,don't generate

    Every beat began as real footage. AI reshaped it into the character the game needed, which keeps the client's wishes intact and lets the team answer social at the speed it moves.

    Swirling smoke
    01Smoke
    Dragon scales emerging from smoke
    02Scales
    A gold dragon eye
    03Eye
    A new dragon arrives, 09.24.26
    04Date
    No. 16Amazon Luna · Pitch

    Make itplayable

    Pitch | Gen AI | Social content

    The idea was to show the Amazon team how they could take a game they already have and use the power of AI to bring it to life. Starting from the game’s own look, we built a campaign world of key art, characters, UI moments and motion, ready to become social media content.

    My role I helped shape the idea and the visual system, showing how AI can turn an existing game into a steady stream of social-ready content.

    ClientAmazon Luna · Amazon Games
    DeliverablesPitch deck · key artCharacter system · social · UI · motion
    NoteOriginal Gen AI concept art for discussion. Not official Amazon artwork.
    Poster: Even superheroes have one of those days. Luna is there when you need a reset.
    Key artThe reset button
    A glowing purple controller over the line Play your way
    Key artPlay your way
    Do not show them a prompt. Show them the poster.From the pitch
    1. References

      Two images set the world.

    2. Brief

      Luna is the reset; the crew brings the comedy.

    3. Prompts

      One look, written down so it repeats.

    4. Worlds

      Couch, missions, cast and game worlds.

    5. Deck

      The pitch arrives as finished work.

    One look,many assets

    The same style carried across every format the campaign would need: hero key art, social tiles, a character system and mission UI. A pitch that already reads as a campaign kit.

    Heroes slumped on a couch with controllers, a glowing Luna device on the table
    01Couch
    A player reaching into a holographic mission board as heroes burst out
    02Mission
    Friends cheering as a hero leaps out of the TV
    03Living room
    Character system: the cast in a row with icons and stickers
    04Character system
    Grid of social tiles in the same style
    05Social tiles
    Game worlds opening around players on a couch
    06Game worlds
    A dispatcher picks a hero card from a glowing mission console
    07Mission UI
    The opportunity

    Own the celebrationof gaming culture

    Game marketing has become entirely functional, and that creates an opportunity. With Dispatch, a comic-book superhero workplace comedy free with Prime, Amazon already had every asset it needed to start the conversation.

    Comic-book heroes crowd together with pizza and a controller
    Key artThe crew
    A hero meditates above a glowing controller in a dark living room
    Key artDispatch · Free with Prime
    Social

    Readyfor the feed

    The same world, turned into posts for Luna’s channels: quick, funny moments that talk like the people who play.

    Social post: Who's up for a speed run?
    01Speed run
    Social post: Ever have one of those days?
    02One of those days
    Social post: Squad-based carbs!
    03Squad-based carbs
    Social post: Doesn't everyone wear gloves when they're gaming?
    04Gloves
    Social post: When he says he doesn't use his phone for big purchases
    05Big purchases
    The heroes and friends gathered on a couch around a glowing Luna device
    Key artSquad night
    A hand on a glowing controller in purple light
    Key artSo what could this mean for Luna?
    MotionThe controller, warping in
    An eagle, a terrier and a red squirrel in an autumn forest, composed in Loom
    No. 17Our platform · vibe-coded

    The future of node‑based editing.

    Built for designers, by designers.

    Product concept | Vibe coding | MCP connectivity

    This is my vision for node-based editing, if I were to build one. Most node-based tools are powerful, but they are not built around the way designers actually think and work. This project asks what a node-based environment would look like if it began with the creative process. The idea brings together familiar tools such as Photoshop, After Effects, Blender, and new AI systems in a layered, connected workspace.

    My role I am designing and prototyping the concept through prompt engineering, vibe coding, MCP connectivity, and an understanding of how large language models can connect tools into useful creative systems. The goal is to give designers more control.

    My roleDesigned & vibe-coded itMac app
    Live linksPhotoshop · After EffectsIllustrator · Blender
    Client-safeClient folders · brand kitsSpend tracked per project
    Loom canvas: an image and a prompt wired into Nano Banana Pro and Kling 3.0, feeding live Photoshop and After Effects blocks
    9,000+Lines of code, vibe-coded by me, a designer.
    26Block types, from prompt and palette to camera and contact sheet.
    18Image and video models behind one set of wires.
    4Live links: Photoshop, After Effects, Illustrator and Blender.
    01 · Nodes

    Wire it,don't re-prompt it

    Prompt, reference, palette, camera and model blocks connect into a graph. Change one input and everything downstream updates. The graph is saved, so the next designer can open it and run it again.

    PromptPaletteCameraEnhance promptCompareContact sheet
    02 · The real tools

    Photoshop andAfter Effects, live

    A result lands in an open Photoshop document as live layers, or on an After Effects timeline, ready for a designer's hand. Finish it there and it comes back to the canvas. Here a forest, an eagle and a palette are composited in Photoshop, then the whole scene turns to snow.

    PhotoshopAfter EffectsIllustratorLayers
    Loom → PhotoshopSession · 3× speed
    03 · 3D, live

    3D insidethe canvas

    Work live in Blender and shape the result as you go. A small add-on, Loom Link, streams the Blender viewport into a block on the left of the canvas; move the camera, change the model or relight the scene and the creation on the right follows. Here a paper architectural model in Blender becomes a photo of a real model on a studio table, without leaving the canvas.

    Blender, liveLoom LinkCameraModel viewer
    Blender, livePaper architectural model
    Blender → Loom → photoSession · 2× speed
    04 · Spend

    Know what it costs,as it runs

    Every paid generation is priced the moment it runs and logged with its project. Each block shows about what one run costs before you press go, Run all totals the batch, and the chip in the top bar keeps a running total for the project, model by model. Free tools count as zero, and the admin sets prices for the whole studio.

    Cost per blockRun-all totalSpend per projectStudio price list
    Loom spend tracker: the project total broken down by model, with cost estimates on each block and on Run all
    Spend trackerDemo project · list-price estimates
    The code behind it

    Vibe-coded,end to end

    Loom is built in-house. I vibe-coded it as a designer, one conversation at a time: the canvas, the model library, every live link and the Blender add-on. When a job calls for a new block, I write it that week.

    • server.py · app.jsCanvas, blocks, models
    • pslink · aelink · ailinkPhotoshop, After Effects, Illustrator
    • strand_link.pyBlender add-on, Loom Link
    • strand_mcp.pyConnector for Claude
    • clients · foldersClient folders, brand kits, provenance
    strand_mcp.py — Loom
    def call(name, args):    # Claude asks to see the live Blender viewport    if name == "blender_view":        raw, _ = http("/api/blender/snapshot", timeout=15)        return [image_block(raw)]    # ...or to look at what a node just made    if name == "look_at_node":        graph = remote("get_graph", {})        node = next((n for n in graph["nodes"] if n["id"] == args.get("id")), None)        if not node:            raise RuntimeError("No node %s" % args.get("id"))        out = node.get("output")        if not out:            return [{"type": "text", "text": "%s has no output yet" % node["id"]}]        url = STRAND + out if out.startswith("/") else out        if node["type"] in ("veo", "generate_video"):            return [{"type": "text", "text": "Video: " + url}]        with urllib.request.urlopen(url, timeout=60) as r:            return [image_block(r.read()), {"type": "text", "text": url}]    # everything else goes straight to the open canvas    return [{"type": "text", "text": json.dumps(remote(name, args))}]
    MCP

    Claude,at the controls

    Loom ships with its own MCP connector. Add it to Claude Code or Claude Desktop and Claude can read the open canvas, add and wire blocks, run them, look at what came back and check the live Blender view. One line installs it, it uses only Python's standard library, and API keys stay in the admin console, never in the browser.

    $claude mcp add strand --scope user -- python3 ~/.strand/strand_mcp.py
    get_graphRead the open canvas: every block, wire and result.
    add_nodeDrop any of the 26 blocks onto the canvas.
    set_paramsChange a prompt, a model, a camera or a palette.
    connectWire one block into another.
    delete_nodeClean up what the graph no longer needs.
    run_graphRun the graph, or just the part that changed.
    look_at_nodeSee a result the way a designer would: the image itself.
    blender_viewSee the live Blender viewport, right now.
    Camera block with orbit control, shot size and lens
    01Direction · camera
    Live Photoshop and Illustrator blocks
    02Live Photoshop · Illustrator
    Loom model library
    03Library
    No. 18AlcaChino · Brand partnership

    AlcaChino

    Brand partnership | Packaging | Product storytelling

    AlcaChino began as an original partnership: a chance to build a coffee cocktail brand with a distinct place at shelf, in hand, and in culture.

    My role As a partner and original founder, I helped build AlcaChino from the ground up. I created the branding and packaging, shaped the product world, and managed the look and feel throughout as the brand grew.

    My rolePartnerOriginal founder
    FocusBrand foundationPackagingProduct world
    PhotographyPrism House
    The AlcaChino product family with a cocktail shaker and an espresso martini
    No. 18 · AlcaChino

    Start withthe product

    The brand needed to feel considered before a customer ever tasted it. The work began with a clear product story, then carried that idea into packaging and a visual world that made the cans feel tangible, distinctive, and shareable.

    Vodka Spiked Mocha Latte can in coffee beans
    Two hands passing AlcaChino cans
    Bourbon Spiked Vanilla Latte can in coffee beans
    AlcaChino cans scattered across a tan backdrop
    Vodka Spiked Original Latte can in coffee beans
    Rows of AlcaChino cans in every flavor
    Three AlcaChino flavors scattered on a tan backdrop
    Campaign image: Happy hour every hour
    CampaignHappy hour
    Campaign image: Cocktails
    CampaignCocktails
    Campaign image: It's the boozy latte of your dreams
    CampaignBoozy latte
    What it made possible

    A launchpadfor what came next

    The brand foundation became a launchpad for the next venture: SunBuzz.

    Photography · Prism House
    No. 19SunBuzz · New venture

    SunBuzz

    New venture | Packaging | Brand world

    SunBuzz turns the things people already love about a cocktail (flavor, ritual, and a reason to gather) into a colorful, modern THC experience.

    My role As a partner and original founder, I created the SunBuzz branding and packaging, and I manage the look and feel across every touchpoint, from shelf to social. The goal is a new generation of cocktails, built for beverage platforms that keep changing.

    My rolePartnerOriginal founder
    FocusPackagingBrand worldSocial
    Starting pointLessons from AlcaChino
    SunBuzz Cosmopolitan bottle on pink with a disco ball
    No. 19 · SunBuzz

    Familiar,but new

    As the non-alcoholic market changes, the opportunity is not to leave cocktail culture behind. It is to bring its recognizable tastes and occasions into a new format, one that feels fun, social, and confidently current.

    Social

    A brand worldfor the feed

    Instagram templates that put each flavor into a moment: a bright flat lay, a desk, a tote, a cooler on the way out.

    SunBuzz Margarita bottle with limes and ice
    SunBuzz Jalapeño Pineapple bottle with pineapple and jalapeños
    SunBuzz bottle on a sunny desk
    SunBuzz bottle in a tote bag on a car seat
    Four SunBuzz flavors in a cooler of ice
    SunBuzz Cosmopolitan on a tiled bar
    Squeezing lime into a SunBuzz Margarita
    SunBuzz Jalapeño Pineapple with fresh pineapple on yellow
    SunBuzz Margarita with a coupe and lime on green
    SunBuzz Jalapeño Pineapple six-pack with a cocktail
    SunBuzz bottle beside a yoga mat in the sun
    The next chapter

    Immediatelyinviting

    A brand world designed to make a new kind of drink feel immediately inviting.

    Social templates made with AI
    Resume

    Sergio
    Cardona

    Creative Technologist · AI Design Lead

    Profile

    Creative technologist with roots in CPG packaging and a self-taught path into 3D. I design the AI workflows, prompt systems and tools behind commercial films and product imagery, train the teams who use them, and create work of my own, with photoreal accuracy and brand control at the center.

    Experience

    Mod Op

    Apr 2024 – Present

    Creative Technologist · AI Design Lead, AI Content Studio

    • Lead AI-enabled production for film, product imagery, packaging and social, for brands including Skechers, Duracell, Kia, New York Life and Black Box Wines.
    • Designed and vibe-coded Loom, a node-based editing platform that links generative models with Photoshop, After Effects and Blender.
    • Created the prompt-engineering workflow behind Skechers’ AI campaign scenes, trained the team in node-based editing and prompting, and supported them through troubleshooting.
    • Matched East East’s original photo shoot through prompt engineering, opening an ongoing partnership as digital photographer for Dermstore, including its holiday campaign.
    • Build multi-SKU product-image systems that keep labels, logos, proportions and materials accurate at scale.

    LAM Design

    May 2017 – Apr 2024

    Associate Art Director

    CBX

    Oct 2010 – May 2017

    Senior Designer

    Cornerstone Strategic Branding

    2006 – Oct 2010

    Junior Designer

    Sunbuzz

    Sep 2025 – Present

    Partner & Original Founder, Brand & Creative

    • Created and manage the brand look and feel for a new generation of ready-to-pour THC cocktails, across packaging, flavors, formats and social.

    Alc-A-Chino

    Completed venture

    Partner & Original Founder, Brand & Creative

    Education

    Fashion Institute of Technology (FIT)

    2006

    Bachelor’s degree, Packaging Design

    Expertise
    AI production

    Prompt engineering
    Node-based workflows
    Generative image & video
    AI product photography
    Digital twins

    3D & finishing

    Blender
    Photoshop compositing & retouching
    After Effects
    Photoreal rendering
    Packaging mockups

    Brand & leadership

    Packaging & brand design
    Art direction
    Team training
    Workflow design
    Creative tooling & vibe-coding

    Selected work

    The workflowis the asset

    Every project starts with the client: what they need, and how comfortable they are with AI. From there I shape a hybrid approach that respects and protects their intellectual property while working with LLMs, combining the latest tools, like node-based editing, with traditional craft. That combination is my advantage, and each project leaves behind a method the next team can run.

    Rights-cleared references. Client-owned packaging rebuilt by hand. Brand, legal and claims review before anything ships. Prompts, node graphs and QA checklists stored so the work can be repeated, not just admired.

    Stage 1Draft

    One person, one tool, one idea worth testing.

    Stage 2Pilot

    Live client work with a named reviewer.

    Stage 3Repeatable

    Someone else runs it and gets reviewable results.

    Stage 4Template

    In the Hub with prompts, references and QA.

    Stage 5Agent

    Stable enough to automate the busywork.

    Clearto ship

    The law caught up with generative AI in 2025 and 2026. Every frame on this site was made against three rulebooks, New York, California and the EU, plus US copyright, and a hybrid human-and-AI process built to meet them.

    1. WhereThe ruleIn forceOur answer
    2. New YorkS.8420-A
      Synthetic performer disclosure

      An ad that shows an AI-generated person, one who looks human but isn't a real, recognizable individual, must say so clearly and conspicuously. $1,000 for a first violation, $5,000 after.

      Jun 2026
      How we handle it

      Synthetic performers are flagged at concept stage, and the disclosure line is designed into the frame for New York-facing ads, not bolted on.

    3. New YorkS.8391 · Civil Rights Law §50-f
      Deceased performers' digital replicas

      Written consent from heirs or executors before an ad uses a digital replica of a deceased person's voice or likeness.

      Dec 2025
      How we handle it

      No replica of a real person, living or deceased, without a signed release on file.

    4. CaliforniaAB 2602
      Digital replica contracts

      A clause letting a performer's AI replica stand in for them is unenforceable without a reasonably specific description of the use, when the performer had no legal or union representation.

      Jan 2025
      How we handle it

      Talent releases spell out the exact replica use, channels and term.

    5. CaliforniaAB 1836
      Deceased performers

      Estate consent before producing or distributing a digital replica of a deceased performer. Liability of $10,000 or actual damages, whichever is greater.

      Jan 2025
      How we handle it

      Same release rule as New York, applied everywhere we work.

    6. CaliforniaSB 942 · AB 853
      AI Transparency Act

      Large generative AI providers must embed latent provenance in AI image, video and audio, offer a detection tool, and require licensees to keep it. Large platforms must surface it from 2027.

      Aug 2026
      How we handle it

      Licensed, enterprise-tier tools only, with provenance data kept intact through finishing and export.

    7. European UnionAI Act · Article 50
      Deepfake and AI-content transparency

      Whoever publishes a deepfake must disclose it as AI-made, with a lighter touch for evidently artistic or fictional work. Providers must machine-mark AI output. Not postponed by the Digital Omnibus. Fines up to €15M or 3% of global turnover.

      Aug 2026
      How we handle it

      A deepfake check on every asset, disclosure labels for EU markets, and the Commission's Code of Practice on marking and labelling as our reference.

    8. United StatesCopyright Office · Part 2 report
      Who owns AI output

      Prompts alone aren't authorship. Human selection, arrangement and modification of AI material can be protected, but the purely generated parts can't.

      Jan 2025
      How we handle it

      Human-made sketches, comps, retouching and edits on every piece, documented so the client's ownership holds up.

    An internal summary of the rules as of October 2026, not legal advice. Laws in this area move quickly; every campaign is cleared with Mod Op legal and the client's counsel before release.