Creative direction / Brand films / Generative AIParis / 2009-2026

ArnaudLabbé

I shape the idea. Then lead its making.

Brand films, campaigns and visual worlds. From creative strategy to final delivery, I bring together writing, art direction, production and generative AI whenever the brief calls for it.

+15years in creative production
3-10specialists per team
€10k - €100kproduction budgets
53 %successful public / private pitches · 09/25-09/26

Seven projects, seven approaches.

Each entry brings together concept, art direction, production and, when available, the relevant making-of material.

Provisional editorial visuals indicate projects whose shareable production media are not yet included.

+15 years creating moving images

Art director and motion designer since 2009, and cofounder of PLISSKEN Production, I work from concept to finish: storytelling, shot design, editing, animation and compositing.

Since 2024, I have progressively integrated generative tools into my professional workflows whenever they unlock a visual solution, accelerate a stage of production or offer a different way to make the work.

Art directionMotion design 2DEditing & rhythmCompositingAI pipelines
In production / 2026

#IciCestMaPlace

Build a consistent digital campaign across six professional sectors, avoiding literal job stereotypes while retaining a shared brand identity.

Scope6 sectors / 21 branches
Format6 × 10-second looping spots
Intended channelsDOOH · social media
ResponsibilityCreative concept & art direction

The challenge was never just six videos.

The campaign needed a shared visual grammar across six sectors, while leaving room for each headline to feel distinct and staying faithful to the “Ici, c’est ma place” identity.

Initial collage/print approaches, directions that strayed too far from the brand identity, and overly literal representations were dropped. A “3D POP” direction was chosen, to be refined for screen readability and a more mature tone.

01One headline
02One motion idea
03One sector
04One shared brand language

Explanatory design framework - not a final campaign visual.

One direction, multiple expressions.

Each headline becomes the starting point for a visual concept. Shot planning then translates the idea into moving images, reusing brand system elements to maintain continuity across six short films.

ESET.

A cinematic language to convey what cybersecurity makes possible.

Selected concept / In production
01 / Core ideaProgress, embodied by a continuous turquoise flow
02 / Visual grammarLive action + generative imagery
03 / EnvironmentsHealthcare · manufacturing · logistics / European signature

Make progress the protagonist.

The turquoise thread does not stand for an interface or antivirus software: it embodies progress already in motion. The film connects B2B worlds and European locations through a coherent narrative, bridging live-action footage and generated imagery. The concept was selected through a competitive private pitch.

The source archive contains no production visuals for this film; no fictitious footage is shown.

Case 04 / Image creation

Develop packshots and stylised imagery for the VOCABULAIRES project, from the brief to the finished visual assets.

Script · Moodboard · Generation · Photoshop · Film16:9 and 9:16 edits

Main edit 16:9

Vertical edit 9:16

01 / Frame
VOCABULAIRES moodboard exploring books, materials and packshot imagery

Translate the editorial intent into clear visual rules and a concrete direction applicable to the project.

02 / Generate
Development board and generated packshot for the Jardin volume

Explore the generations and select the strongest working foundations.

03 / Finish
Final packshot of the Jardin volume in a botanical set

Retouch, harmonise and prepare every visual for animation.

Packshot of the Vitrail volume in a yellow set crossed by coloured reflections
Development board and generated packshot for the Mobilier domestique volume
Jardin packshot variation in a set made of soil and vegetation

Case 01 / Client production

Create a series of six films focused on hospitality professions, using a consistent, repeatable visual system designed for social media.

Work in progress

ClientAKTO
Project6 hospitality profession films
RoleArt direction · Editing · Motion · AI
Year2026

Symbolic yet tangible: reality recomposed

The project combines performers filmed on green screen, recomposition assisted by AI and traditional postproduction. The challenge was to extend each performer’s identity, create symbolic, consistent and photorealistic imagery for every profession, and generate all illustrative shots while maintaining visual identity and spatial continuity. These shots replaced a heavy preproduction phase and a demanding shoot across several locations.

Upstream, the script and technical breakdown define what needs to be generated. AI production follows the logic of a traditional production pipeline. It is never added as an afterthought.

01
AKTO visual development board featuring graphic references, a performer and colourful sets

Design

Define the series’ visual language and grammar before production begins.

02
Excerpts from the AKTO script and annotated technical breakdown

Plan

Decide how each shot will be produced: live action, generation, animation or compositing.

03
Comparison between the performer filmed on green screen and recomposed in the final set

Shoot

Capture a real performance designed around the editorial requirements and also intended for AI recomposition.

04
Character sheet showing the performer from multiple angles

Stabilise

Build a sufficiently robust identity base to remain consistent across shots and test its reliability through video generations.

05
Location research showing several angles of the same restaurant

Build the space

Maintain geography, lighting and scale from one generated shot to the next by establishing a solid and logical spatial continuity.

06
Selection of generated keyframes for service gestures and situations

Keyframes

Generate every keyframe using the visual guides for identity and spatial consistency.

Case 02 / Hybrid 2D animation

Make a still illustration feel hand animated without smoothing out its linework or losing its energy.

Delivered formats

Full trailer. The same edit is adapted for social formats.

ProjectGLITCH / Ki-oon
ApproachAI + manual animation
TimingAnimation on threes
Year2026

Craft movement, not just an effect.

NB2 separates the characters, backgrounds and useful elements. Previously hidden areas are then automatically reconstructed through generative completion, turning each component into an editable layer. From this material, compositing resumes as a traditional workflow.

Seedance 2 / Kling 3.0 animates the previously isolated characters. Every pose from the selected output is retouched frame by frame to build animation on threes, followed by secondary motion, compositing and finishing in After Effects.

“The quality is very high and it makes you want to see more. You pushed it very far, and it shows on screen, to the point of elevating the original material.”

Fabien HyzardHead of Marketing and Sales · Ki-oon
01 / Source illustration
Source illustration for the GLITCH shot
02 / NB2 separation

The image is extended to create a comfortable, logical working canvas matching the delivery ratio. Each element is then processed, separated, extended and completed according to a traditional compositing workflow.

Separate layers for compositing06 assets
01NB2 layer 01 from the GLITCH shot
02NB2 layer 02 from the GLITCH shot
03NB2 layer 03 from the GLITCH shot
04NB2 layer 04 from the GLITCH shot
05NB2 layer 05 from the GLITCH shot
06NB2 layer 06 from the GLITCH shot
03 / Hybrid shot composition

Here, two different illustrations are used to create a single animated shot. The character’s pose in the opening image becomes the source for video generation.

Character prepared for the hybrid shot composition
A. Opening shot / In
Background prepared for the hybrid shot composition
B. Closing shot / Out
04 / Animation & compositing

The character’s halftone shadows are removed beforehand so the video generation model can interpret the image more clearly. They are then restored frame by frame to refine the result and make the AI assistance invisible.

Character composition and layer stack in After Effects

Frame by frame retouching for animation on threes.

Pose 1 from the motion breakdown01Pose 2 from the motion breakdown02Pose 3 from the motion breakdown03Pose 4 from the motion breakdown04Pose 5 from the motion breakdown05Pose 6 from the motion breakdown06Pose 7 from the motion breakdown07Pose 8 from the motion breakdown08

Case 03 / Visual system

Extend an existing illustrated visual identity through a controlled node based AI workflow while preserving its graphic codes.

Research · AI workflow · Motion design16:9 film · 9:16 version
Reference illustration board for AFNOR

01. Identify the constants: palette, shapes, density and composition.

Overview of the AFNOR node based AI workflow

02. Translate the visual identity into a controllable, repeatable generation system.

Close up of the prompt text within the AFNOR node based workflow

Focus. The images selected as a moodboard are connected to a prompt node, enabling and facilitating the understanding of multiple references for repeatable results.

Final film

Main edit in horizontal format.

03. Prepare and animate the new elements within the film.

“AI does not replace the visual identity: it becomes a tool for extending it.”

NETATMO-LEGRAND

NAC / NDL product films. Art direction, copywriting, brand consistency and leadership of a 3D specialists team. Deliverables in 16:9, 9:16 and 4:5.

Additional reference, intentionally placed after the AI-led work.