AI-generated visuals can be a big opportunity for a brand, but they can also cause a backlash. Here’s how we use AI at Sublimio.

—
This is the eleventh article in a series describing Sublimio’s approach to branding projects. In case you haven’t read the previous one about packaging, you can find it here.
—

To generate or not to generate, that is the question.

You know you can create brand visuals with AI, but should you? Is that going to make your brand look cheap or fake? Will it just join the sea of AI slop that is flooding social platforms?

The answer is – as usual – nuanced. On one hand, the use of AI is risky: while it feels magical, it’s also an imperfect technology, which can easily spin out of control if used at its base level. On the other hand, AI can quickly expand your brand universe and make your brand vision more defined and engaging. 

Here is how we approach it in theory and most importantly in practice.

Evaluating the role of AI in your project

So, why use AI? This is a good question to ask ourselves when starting a new project. “It’s cheaper” some will say. Yes, it is – even though not necessarily as cheap as some think – but if that’s why you are using it, you might be on the wrong path.

Using AI also communicates something to your audience. If used lazily, it will communicate the wrong thing: your brand just doesn’t care (just think of Gucci’s latest fumble). 

‎

There are, on the other hand, more intentional ways to use this technology. First of all, it has to be clear that you are not trying to represent reality in a purely objective way.

AI is not fit to tell a neutral truth the way photography can do (though even the idea of a neutral photography is debatable). It’s rather a tool for augmented storytelling.

‎

AI allows you to expand quickly from a root idea, to turn a vibe into a whole visual world. It works very well when you want to convey a grand vision rather than just create a single visual.

Your audience should be able to perceive your creative intention behind the tool. Simply looking good is now almost worthless. So how do you achieve that?

Art directing before you prompt

When you pick AI as your production method, it’s good to be aware about both the advantages and the drawbacks. It will probably be faster and you will be able to do more, but at the same time you can expect a certain loss of control.

‎Compared to other production methods, controls in AI tools are still quite blunt and unpredictable.

This is one of the main causes of the “slop” look, where visuals default to the model’s base style and end up looking like everyone else’s. Which is quite obviously poisonous for any brand.

‎Reducing randomness in AI is not – pardon the pun – a random result. Defining the visual direction takes two kinds of action (at least, that’s what we do).

‎

The first is the creation of a reference board to guide the visual style. It can have different appearances based on the tool used: what matters is the work of art direction that makes it possible.

Taste is what makes the difference at this stage and most importantly a good grasp of how the brand should feel.

‎

The second step is the creation of a technical process that makes this look reproducible in a reliable way. Again, this can happen in many different ways depending on the tools and the project, but it’s vital to ensure that the brand can access its visual style at any time.

Usually, prompting isn’t enough: to guarantee consistency, you need more structured workflows.

‎

maikasui: AI visuals with a soul

At Sublimio, we made use of AI to bring to life the story and world of maikasui, a Japanese brand of hand care products. ‎

maikasui has a well defined world of values, where authenticity, gentleness and the relationship with nature guide every action.

To bring this into visuals meant developing a style that would go beyond simple product representation: our aim was to make the images deeply emotional, to use them to recreate that very feeling of inner peace.

‎

On Instagram, we used AI to expand the visual world of maikasui in a coherent way. Not just product images, but a whole mood that also includes lifestyle visuals. 

This has allowed us to grow the narrative with more places, more micro-moments. All images belong to the same visual and emotional universe, and that’s by design.

The product had to fit within these micro-moments, so that was a double challenge: preserve the product look while also respecting our overall mood.

‎

On maikasui’s website, we did something more challenging. Since maikasui’s products are meant to be used throughout the day, we created living pictures (cinemagraphs) that seem to present us with a moment frozen in time, from the lives of Japanese women. 

We used AI, combining image models, video models, traditional editing and coding to create a minimal website that would feel alive, showing the product as embedded into the women’s lives. 

‎

Expressing the nuances of life through artificial intelligence is only possible if you have a very clear vision of the final result you want to achieve, not just visually, but emotionally. That’s a human job.

‎This way of working ensures that the world of maikasui won’t feel like an AI created it, but rather like humans used AI to channel it. And that’s what a brand should aim for.