After using MidJourney for the last two years, I find myself reflecting on an incredible journey of creativity, growth, and innovation. What started as a fun experiment has transformed into a new tool which we can leverage for our clients with impressive results.

Here’s a look back at how we’ve used these groundbreaking tools to date.

The Early Days: Memes and Exploration

As a tech enthusiast I’ve always been an early adopter.

From the early internet through to Minidisc and VR, there’s always been something inside of me that has bristled with excitement at the promise of technology, all shiny and new – a tantalizing glimpse of the future.

And so it was that in late 2022 I saw a new path laid bare, a glimpse of a new technology that would change the future of creative image making and it was with excitement that I began to explore a new tool that had been generating hype within my creative networks: Midjourney

But this shiny new toy came with a catch. Its interface was Discord, a chat app mostly used by gamers—a revolution still figuring out user experience.

Bewildered at this unfamiliar interface, our old friend Google provided the answers.

I typed out the prescribed command: /imagine

I froze for a moment. Now was the chance to enter any description known to mankind. I could reference any art period through history, any fictional fragment of an idea, the only limitation was my imagination.

I tuned to my kids and following their response I typed:

/imagine…A hamster playing guitar mixed with Darth Vader.

Full disclosure –  the last part was my own addition.

Moments later a grid of four thumbnails began to appear. At first blurry, abstract notions, resolving into clear images. I later came to learn that this process was called diffusion (Or more accurately reverse diffusion – gradually dreaming existence from an initial noise pattern).

For the next hour, I was hooked. I created high-brow art, some monstrous creatures, oil paintings of my long deceased Labradors, and even indulged in a bit of schadenfreude, picturing a down-and-out Boris Johnson begging on the streets of London.

Three fingers, a hoof and a nice dress.

As I became more familiar with MidJourney, I began to see beyond the memes. The AI’s capability to create high-quality, unique images opened my eyes to its broader applications.

I began to explore other tools such as Dalle 2 – The OpenAI solution.

I started exploring how these tools could be integrated into more serious projects, such as digital art, marketing materials, campaign imagery.

‘Out of the box’ however the results were mixed. Ever improving but cursed with defects. Extra limbs, unnatural poses, and bizarre quirks. It was easy to see progress with every version update but the tools were clearly not ready for commercial use. It was hard to art direct images and using the tools felt more like a lottery than something capable of generating consistent results.

I had some small success creating abstract graphic assets for a friends film pitch deck (A hallucinogenic, Ayuaschca style treatment featuring ‘sacred geometry’ and other shamanic symbols). The tools could be used where imperfection was a feature not a defect.

How apt that it’s first use was creating a hallucinogenic set of signs and symbols.

A one click wonder?  Refining the Workflow: Efficiency and Precision

Somewhere around version 5 or 6 of MidJourney, things began to change. Despite the persistent annoyance of misshapen hands with missing digits, the overall fidelity and intent of the images improved significantly.

With my growing confidence in MidJourney’s capabilities, I began to refine my workflow. While the term ‘prompt crafting’ might make me cringe, there is indeed a method to the madness.

I learned a glossary of new commands, from —Sref to —Cref and —iw, and everything in between. Prompting started to feel like collaborating with a fellow creative—a stubborn savant with unlimited potential. Success came when I committed to learning how to communicate on its terms.

This phase was all about efficiency and precision. I developed a systematic approach to prompt creation, learned to predict the AI’s responses more accurately, and streamlined the process of selecting and editing the final images.

While the algorithms improved, the tool was rarely used in isolation. When MidJourney encountered limitations, I would fine-tune specific areas in Krea AI combining these outputs in Adobe Photoshop using traditional retouching techniques. Art direction was finally within reach—a hybrid approach that maximized the strengths of both AI and human creativity.

Later, Adobe released its own image generator, Adobe Firefly, which we have also used with success. Its well-designed UI and several tools to control the ‘structure’ and ‘mood’ of the image create further opportunities to enhance our creative output. Being able to define the silhouette of an image and prompt based on that structure is a gamechanger.

Embracing these tools has been a transformative journey, blending the best of AI and human ingenuity to push the boundaries of art direction.

Commercial practice: How we’ve used AI as a creative brand studio.

As we’ve grown in our confidence, we’ve shared our knowledge internally and the last 12 months has seen a rise in our use of AI to assist – and not always on the final creative deliverable itself.

Specifically we’ve:

• Used AI as a tool for early storyboarding for our Essendon Fields brand film. Using pencil sketch styles to indicate interviewees without steering the visuals in one direction or another.
• Using AI for creative storyboarding – representing locations through consistent styles and helping the client to understand the final creative intent without drawing on stock imagery.
• Using AI as a creative tool to help visualize campaign ‘mood’ images. These images were then used in a range of collateral from brochureware to web outcomes.
• Creating a range of digital avatars for Coburg Industries for business types that might want to lease warehouse spaces.
  Digital mockups for our own case study pages.
• Developing a series of high resolution campaign images for Mirvac for 7 Spencer Street. A retail property development. Elevating the outcome beyond budget expectations.

Navigating ethical concerns.

We prioritize the ethical application of AI and strongly advocate for the human touch in all our endeavors. Collaborating regularly with photographers, illustrators, and a diverse range of creatives, we eagerly anticipate the continuation of these partnerships.

Within our studio, certain practices are firmly off-limits, such as prompting based on an individual illustrators style.

We commend the utilization of tools like Adobe Firefly, which harness the knowledge from stock images.

Upholding the rights of artists to receive fair compensation for their contributions, we assert that artists should have the choice to participate in AI training data, ensuring they are duly remunerated for the utilization of their imagery, even in training contexts.

Looking ahead, we envisage the potential of Web 3 and blockchain technologies to establish a tokenized internet, where digital assets can be tracked in their usage and appropriately monetized for the benefit of the source creator.

A brave new world for Video.

As image generation technology has advanced and user experience has improved, we’ve witnessed remarkable strides in the quality of generative video tools. Tackling the challenge of maintaining seamless transitions between frames in videos has been particularly complex due to their temporal nature.

While OpenAI’s incredible Sora technology lies on the horizon we have already harnessed existing tools like RunwayML and Pika to infuse vitality into our visual projects. From crafting animated cinemagraphs for websites to developing compelling case studies for our own online presence, these tools have proven invaluable.

In a significant announcement made by Midjourney CEO and Founder David Holz in late 2023, it was revealed that Midjourney is actively engaged in advancing 3D and video generation technologies. With a commitment to delivering a solution that surpasses current standards by tenfold, Midjourney is poised to redefine the landscape of video creation.

And the change of pace is exponential. The last fortnight has seen the release of Kling AI, currently only available to those within China.

Days later ‘Dream Machine’ was released – The Luma AI competitor offering similar visual fidelity.

Furthermore, we’re closely monitoring advancements in music generation tools like Suno.ai, as well as exploring the capabilities of 3D generative applications.

To infinity and beyond!

Reflecting on my first years with MidJourney, I am awestruck at the exponential growth in these tools. The rapid advancements in AI technology suggest that we are only scratching the surface of what is possible.

Time will tell whether AI will be a net positive for the industry. Are we driven by temptation, opening Pandoras Box? An irresistible yet cursed vessel? Or is this Ouroboros the emblematic serpent of ancient Greece eating it’s own tail? In fact an expression of unity of all things which never disappear but perpetually change form in an eternal cycle of destruction and re-creation.

As a creative studio we’re investing our time into being at the forefront of these tools. To use them carefully with tempered expectation, alongside our human endeavors.

As we look ahead, we are confident that the best is yet to come, and we are excited to continue pushing the boundaries of creativity with AI alongside traditional tools.