Skip to article
CGDANNYS
RU
Coca-Cola trucks on an illuminated bridge in a winter landscape
Coca-Cola: throughout my childhood, I waited for these trucks every New Year

AI / BRAND IDENTITY

When the wow effect becomes commonplace

AI makes spectacular visuals available to almost everyone. What does that mean for brands built on uniqueness? A look at real campaigns and the future of the production workflow.

0

There is a question I find more useful than the next AI-video demonstration: what happens to the value of an impressive image when almost any brand can obtain something similar? For production, it concerns the choice of tools. For a business, it also concerns recognition.

Generation lowers the barrier to complex imagery. A team can explore an unusual setting or compare several visual directions sooner. But complexity alone explains less about why the audience should remember this particular product. Its relationship to the brand becomes more important.

When spectacle becomes a production line

An elaborate transformation or photoreal world once visibly represented modelling, animation, light, rendering and compositing. Some comparable images can now be explored through generation, and audiences encounter them more often.

I therefore distinguish a striking frame from a distinctive one. A frame may attract attention while fitting several competing companies. Removing the logo is a useful thought experiment: which remaining decisions identify the work with the product?

A brand sells more than an image

A product has its own form, materials, character and history of use. Communication adds composition, type, colour and tone. Consistent choices across those properties can distinguish a brand even when competitors use similar tools.

Luxury also carries audience expectations. A company that describes craftsmanship and attention to detail creates a reason to expect comparable precision on screen. A striking but interchangeable image may sit uneasily with that promise.

That does not mean every AI-assisted image weakens a brand. The role of the technology matters. Generation can reveal product character, or it can create an attractive setting in which the particular product makes little difference.

Global brands are already testing the limits

Coca-Cola holiday campaign image from the company’s official announcement

Coca-Cola used AI in its 2025 holiday campaign, returning to the recognisable Christmas trucks. The interesting condition is the audience’s existing memory. A new version will be judged partly through its relationship with earlier advertising.

McDonald’s sign in winter, from The Guardian — Beata Zawrzel / NurPhoto / Shutterstock

McDonald’s withdrew its Dutch AI advert after criticism in December 2025. A single production cannot establish a universal judgement about an entire method.

Gucci AI campaign collage from Fast Company — images Gucci

In February 2026, Gucci published AI imagery before its Milan show and received some critical responses. The relevant tension is between experimentation and expectations. An audience can welcome a new technology and still compare the result with its understanding of the house.

Title page of When AI Doesn’t Sell Prada, Journal of Advertising Research

In When AI Doesn’t Sell Prada, researchers used three experiments to examine AI advertising. Disclosure affected perceived effort and luxury-brand authenticity, while more original imagery reduced the negative effect. Those are findings under particular study conditions, not proof that any brand using generation will lose its audience.

Access to an impressive image does not by itself establish its value for a particular product.

In Russia, AI advertising is already part of everyday practice

Russian advertising already uses generation at different scales. Published cases describe animated photographs, generated footage, digital characters and environments for live-action material. Calling all of them an “AI film” conceals important differences in production.

In my view, the pace of advertising contributes to adoption. Teams need new material and formats while adapting to access restrictions and changing services. Speed of delivery is still a separate quality: it does not automatically establish visual accuracy or campaign performance.

М.Видео-Эльдорадо / Sostav

M.Video-Eldorado: AI visuals for a major promotion

In January 2025, M.Video-Eldorado presented a Cyber Sale promotion whose footage, according to the company’s description, was almost entirely generated. It featured representations of M.Team members and applied the method to a large retailer’s promotion.

METRO и Nectarin / Sostav

METRO: an AI ambassador in national television advertising

The METRO and Nectarin case describes an AI version of Konstantin Ivlev for national advertising. The authors reported seven working days from idea to broadcast. Their process included image and voice generation, animation, lip sync, editing and motion design, already several distinct tasks.

Ushatava and Love Republic: technology in fashion’s visual language

Fashion campaigns show other uses. Ushatava animated photographs for Garderobe, while Love Republic used AI in material for its 2025 New Year collection. It is useful to examine which product properties remained fixed and what generation added to the presentation.

The premium segment is taking part too

For the premium Ostrov Pervykh residential development, Crop media described historical imagery and a series of static and moving assets. The concept drew on the character of St Petersburg’s Kolomna area. It provides a specific campaign idea to examine rather than a general claim about novelty.

Darina и Eich / Sostav

Darina: live action and AI in one commercial

In the Eich and Darina production, live action was combined with generated and revised environments. The published case also describes retouching, compositing and colour work. Several sources of material were connected through one production task.

These examples show different applications. They do not establish equal effectiveness or comparable budgets without additional data. They do allow a more precise discussion of what a tool supplied at a particular stage and what the team did next.

Where AI can really improve production

For my own work, I see particular value in exploration before main production and limited operations after it. They offer ways to compare directions or complete a defined task without making the whole project depend on one generation method.

Pre-production needs concepts, storyboards and animatics. They help select a direction and discuss composition and content. The material needs enough clarity for the next decision; final precision in every detail is not always necessary at this stage.

Main production can have different requirements. Product shape, branding, related views and approved revisions need control. Live action, 3D and animation provide ways to preserve selected properties while changing individual parts of a scene.

Post-production can use AI for local revisions, environments and supporting elements. The chart below gives my illustrative allocation of attention: 70% preparation, 10% production and 20% post. It is an author’s model, not market data or a budget breakdown.

PRE-PRODUCTION 70%, PRODUCTION 10%, POST-PRODUCTION 20%70%10%20%AIWORKFLOW

AI across the production process

  1. PRE-PRODUCTION

    Concepts, storyboards, previs

    70%
  2. PRODUCTION

    Direction, CG, live action

    10%
  3. POST-PRODUCTION

    Compositing, cleanup, VFX

    20%

Fully AI-generated ads won't disappear

Fully generated films will also have uses. A quick concept test, short campaign or short-lived piece of content may be served by this method. The requirements of that production should not be projected onto every product film.

In an authored project, generation itself may form part of the artistic idea. Variation then belongs to its language. For exact product demonstration, the same variation can become an obstacle. The task changes the criteria.

My prediction

I expect production methods to keep mixing. AI may support exploration in one project, supply final shots in another and remain a supporting operation in a third. Revision cost, precision and the required subject matter will influence that decision.

For a luxury product, recognisable details deserve particular attention. They need to be identified before choosing how to reproduce them. The discussion returns to art direction: what are we showing, and why should the image look this way?

Different markets can accommodate different approaches. In Russian advertising, I often notice an emphasis on fast response and conspicuous presentation. In international luxury work, precise screen representation of the product receives more attention. These are observations of particular work, not a measured rule for every company.

It is too early to declare which approach will perform better over time. That would require information about campaign goals, audiences and sales. An attractive case or a loud social-media response reveals only part of the picture.

The availability of similar effects already creates a practical question for brands: what makes them recognisable? If a visual device is easy to repeat, differences need to come from the idea, product representation and consistent communication.

Subscriptions to Higgsfield, ChatGPT and Claude can provide tools. They do not explain why a brand has its particular visual language. That answer still needs to be made visible in the work.

What this means for learning

A service can be explored in an evening. Working with its images requires more specific decisions: product form, light, shot order and text placement. Those conditions give a request substance and an output criteria for review.

I would start with a small product and a short sequence. Prepare a still frame, test one movement and build an edit. It becomes possible to see where generation helps and where a manual scene or graphic revision gives better control. Reviewing that result teaches more than comparing broad claims about models.

SourcesCoca-ColaMcDonald’sGucci / City A.M.When AI Doesn’t Sell PradaМ.Видео-Эльдорадо / SostavMETRO / Nectarin / SostavFashion / SostavCrop media: «Остров Первых»Eich / Darina / Sostav

Practical lessons

Related exercises can be studied in my lessons. The planner shows the topics, number of meetings and total before submission.

View the lesson plan
Back