AI Creative Portfolio · 2026
Lisa
Catherine
Joechler
AI Creative Director & Video Creative
Scroll to explore ↓
AI Creative Portfolio · 2026
AI Creative Director & Video Creative
Scroll to explore ↓
A portfolio of AI-generated campaigns, built with the same creative precision applied to
16 years of destination wedding and brand filmmaking, now powered by generative AI workflows.
Lisa Catherine Joechler is a visual storyteller, filmmaker and creative director with 16 years of
experience directing and producing commercial and editorial content across Europe, Bali, Hawaii and beyond.
As co-founder of Lisa & Sven, a photography and film production company based in Germany, she has built a creative practice defined by precision,
emotional narrative and an uncompromising commitment to visual quality. Her photography has been adopted by luxury resorts and international brands.
She is now applying that foundation to AI-native creative production. Not as an experiment, but as a practice.
Lisa builds end-to-end AI video workflows that produce cinematic, brand-quality content efficiently and at scale.
Working with tools including Higgsfield, Kling, Runway and ElevenLabs, she combines traditional filmmaking principles, shot design, pacing, lighting,
emotional arc, with the logic and language of generative AI. Her approach is documented, systematic and built for professional creative environments.
She believes the filmmakers who will define the next decade are the ones fluent in both languages: the craft of traditional cinematography and the capabilities of AI generation.
Based in Germany · Remote-ready · hello@lisajoechler.de
Where the cliff meets the Indian Ocean.
The Edge Bali is one of Uluwatu's most dramatic luxury wedding venues: a clifftop property with an infinity pool suspended above the Indian Ocean, a ceremony terrace open to the horizon, and interiors that blur the line between architecture and landscape.
The photography was originally produced as a self-directed Concept Production: an editorial shoot whose resulting stills were subsequently adopted by the resort and used across The Edge Bali's official marketing channels. The next step was to transform that still photography into a cinematic hero film, designed for seamless loop playback as a website header.
AI photo animation must preserve the photographer's original intent precisely: the color grading, the compositional balance, the quality of light. The technical difficulty multiplies with architectural subjects. Geometric lines must stay absolutely stable while organic elements animate naturally. An infinity pool that glitches or a horizon line that drifts immediately breaks the illusion. Each shot demanded a different solution, and none could look AI-generated.
Each image was analyzed for its inherent motion potential: what movement would exist in this space at this exact moment? The ceremony terrace at sunset calls for a barely-there breeze through pampas grass and clouds drifting on the horizon. The infinity pool at midday calls for a gentle surface ripple with no camera movement. The reception table calls for candle flame and the subtlest focus shift. The interior suite calls for an ultra-slow pull through the room.
Motion prompts were built in two directions: explicitly describing what should move, and what must not. Architectural lines, furniture, tableware: locked. Water, sky, foliage, fabric: animated with intention.
A 42-second cinematic venue hero film built entirely from still photography, ready for seamless website loop deployment. The final cut adds a second layer of atmosphere through sound design: a calm, meditative music bed underneath location-specific sound FX, with ocean waves rolling through the exterior pool sequences and soft applause rising under the dinner reception scene. The audio transforms the visual loop into a full sensory experience.
No video crew. No return trip to Bali. A production approach that scales for luxury hospitality brands working across multiple properties.
Some nights leave a mark.
Self-initiated spec project: create a cinematic 25-second fragrance ad for fictional luxury brand NUIT using AI tools exclusively, from concept through final edit. The goal was to demonstrate end-to-end AI video production with a luxury brand aesthetic and zero compromise on visual quality.
AI video tools have no memory between generations. Each scene is produced independently, with no knowledge of previous outputs. Maintaining a consistent character, hands, and product across 5 separately generated scenes required a systematic solution.
Additional challenges: navigating content moderation on intimate imagery, and directing motion without common AI artifacts including facial morphing, identity drift, and unwanted camera shake.
Before any scene was generated, a 10-shot character sheet was produced to establish the NUIT protagonist, testing different angles, lighting conditions, and expressions using the same base reference. Each shot was evaluated for identity stability, lighting quality, and cinematic presence.
01 · Front
02 · ¾ View
03 · Full Body
04 · Profile
05 · Walking
06 · Contempl.
10 · ★ Anchor
Shot 10, a hair movement portrait with soft dramatic side lighting, was selected as the sole master anchor. From this point, this single image was the only character reference used across all subsequent scene generations.
Three dedicated reference images were established before any scene was generated: the master character anchor, a dedicated hand reference controlling skin tone, finger shape, and absence of jewelry, and a standalone product/flacon reference. All three were passed simultaneously into every generation where those elements appeared.
Character Anchor
Hand Reference
Product / Flacon
A two-sided motion prompt structure was built, explicitly describing what should happen (ultra-slow camera push, hair movement in wind) and what should not happen (no shake, no morphing, no identity drift). Higgsfield Kling 3.0 was used for all video, with preset override to bypass pre-baked templates that ignore reference inputs.
Scenes were generated and reviewed one at a time before proceeding, catching issues early and preserving budget. Final edit in CapCut Pro and DaVinci Resolve: 16:9 format, 0.5s fade-to-black transitions, 25-second runtime.
A 25-second cinematic luxury fragrance spot created entirely with AI, with consistent character identity maintained across 5 independently generated scenes. A technical challenge most current AI video pipelines fail to solve reliably.
Delivered a replicable, documented Triple-Reference workflow for character-consistent AI video production.
Translating 16 years of visual language into AI.
A series of AI-generated campaign stills translating my photographic aesthetic, editorial, cinematic, movement, architecture, into AI imagery. Each still was directed with the same intentionality as a real shoot: specific light quality, exact fabric behavior, precise compositional weight. The brief was to produce luxury wedding campaign stills that could stand alongside real photography.
01 · Night Reception · B&W Editorial
02 · Ibiza Cliffs · Sunset · Color
03 · Reception · Puglia · Candlelight
04 · Intimate Portrait · B&W
Rather than prompting generically, each still was directed using my own reference photography as mood and color anchors. Model selection (soul_2) was deliberate, chosen specifically for fashion and editorial output quality. Prompt language was drawn from real cinematography: f-stop vocabulary, lighting direction, film stock references.
Four distinct visual categories: movement and architecture (Ibiza cliffs at golden hour), night editorial with reportage flash, reception tablescape with candlelight, and intimate couple portrait. Each requires a different prompting register. Each reflects a core photographic strength built over 16 years of real-world destination work.
Some places don't ask anything of you.
Self-initiated spec ad for fictional luxury villa brand VELUM, targeting the private stay market on the Aegean coast. The angle is nervous system recovery and transformation, not product features. The entire 25 seconds is designed to make the viewer feel something before they understand what they are watching.
The ad opens on "Day 3.", a single text card with no context, held one second longer than feels comfortable. High-retention opening built around emotional recognition rather than aspiration: I didn't know I needed it that badly.
Three A/B hook variants were scripted and structured for split-testing to identify which emotional nerve drives the highest 3-second retention on short-form platforms.
Prompt architecture followed the same two-sided logic developed for The Edge Bali: explicit description of what should happen, paired with a negative constraint layer for what must not move, morph, or shake. All clips edited at 0.85x playback speed before assembly. Transitions: cross-dissolve only. Sound design uses ambient ocean, minimal piano, and silence as a structural element. No beat. Space between sounds.
Three hook variants were scripted for A/B deployment targeting distinct emotional entry points: exhaustion ("I didn't realise how much I was carrying"), performance pressure ("I stopped performing my life for 72 hours"), and digital detox ("The first morning I didn't check my phone"). Each runs as a separate post. The hook with the highest 3-second retention becomes the primary creative.
I stopped holding my breath.
Self-initiated spec ad for fictional luxury wellness brand SOLÈNE, a plant-based adaptogen supplement. Emotional angle: not healing, not fixing. Recognition. The viewer should think "this is me" before they understand what is being advertised. Concept Production.
Where VELUM sells aspiration (I want this), SOLÈNE sells recognition: this is me. Recognition-first content bypasses the defensive filter that activates the moment a viewer senses an ad. SOLÈNE never names a problem. The nervous system names it for them.
The central production decision for SOLÈNE was to prioritise emotional consistency over face-matching across scenes. Current AI video models drift: the same reference image generates a slightly different face in each output. Rather than fighting the technology, the shot structure was built to work with it. The first three scenes show hands, a profile, and a close-up of a ceramic mug. No face. The face appears fully only at the emotional peak. When the viewer is tracking a feeling, small visual inconsistencies become invisible.
Reference Still · Scene 03 · Holding
Reference Still · 02 · Scene 06
Shot structure was designed to minimise face exposure across high-risk scenes: hands near laptop, hands holding ceramic mug, curtains in wind. These scenes carry the emotional weight without requiring identity consistency. The face is shown only in Scene 05 at the emotional turn, the moment it matters most and where a slight drift reads as softness, not error.
Edit rhythm: Hard cuts in Scenes 01 to 03 create tension. Cross-dissolves into Scenes 04 and 06 signal the shift. Color grade: warm, slightly desaturated, raised shadows. Copenhagen wellness editorial.
A 27-second wellness spot with a documented, replicable emotional-consistency workflow. The image-first pipeline (generate still, validate, then animate) reduced wasted video credits and produced a cleaner result than prompt-to-video generation would have allowed at this stage of AI tooling.
The strategic distinction between aspiration-led and recognition-led ad creative, built across VELUM and SOLÈNE as a paired project, demonstrates a systematic approach to emotional creative direction for short-form platforms.