Product Labs🔬: Decoding Midjourney

Midjourney (fondly called “MJ” in this article on several occasions), hands down one of the best AI image generators, has taken the world by storm since its launch in February 2022.

Trouble was one of the early adopters and has now elevated to a pro (check out his tutorial on how you too can generate photorealistic images with Midjourney)

Midjourney’s Growth Journey

Midjourney is your AI muse, turning your wildest text prompts into stunning images - not calling it a magical pencil, because all AI today is magic!

Midjourney actually made its debut even before ChatGPT did 🤯. It started as a closed invite-only method, with a pool of 500 invitees who were allowed to extend invitations to 500 more making it a total of merely 1000 users. Today, it is used by a staggering 16M+ users, making it the most active Discord channel by a huge margin!

In a span of less than 2 years, Midjourney has improved by leaps and bounds. We have gone from Midjourney v1 (Feb 2022) to the latest & stunning v6 (Dec 2023)

Let’s give you a feel for what this difference actually looks like…

In hindsight, Midjourney v1 seems quite disappointing. However, this laid the foundation for the concept of text-to-image, which seemed pure magic at that time!

Midjourney applied the MVP (minimum viable product) concept by Eric Ries.

Midjourney launched the simple v1, which was still largely a work in progress. The creators leveraged it to gain quick user feedback through regular micro-polls on Discord & social listening.

It followed it up with a series of quickly improved versions (v1 through v4 were released iteratively, in a span of 8 months), rather than waiting a year to directly launch Midjourney v4 or beyond.

With each iteration, Midjourney has only gotten better…and the most significant leap was observed between v3 & v4.

Let’s take a deeper look at this evolution of Midjourney. Taking this same prompt, we’ll showcase the progression from v1 to v6

large interior by Kengo Kuma, Harmonious blend of natural elements and modern design, an eco-friendly structure, pools and falling water --seed 10293

Early Stages: Embracing the Artistic Spark (v1-v3)

The initial versions of Midjourney showcased a clear artistic vision. Images leaned towards dreamlike landscapes and surreal portraits, prioritizing mood and atmosphere over precise details. This resonated with a specific user base who appreciated the painterly quality and unique aesthetic. However, some users desired a more realistic approach.

Listening and Adapting: The Rise of Realism (v4)

MJ v4 is when Midjourney really came to life. Colours became richer, details sharper, and overall fidelity improved. This shift towards realism addressed the desires of users who craved a more photorealistic experience. Along with better performance & quality, Midjourney also introduced introductory customizations like upscaling & custom aspect ratios.

Finding Balance: Refining the Tool (v5-v5.2)

Things were great with v4, but it still struggled, especially regarding the hands & fingers of AI-generated people. Users complained a lot about this on platforms like Reddit & Twitter and hoped this would get fixed soon.

Midjourney v5’s release was a big relief to these users, with a significant improvement in hands and fingers.

MJ v5 continued the focus on photorealism, with textures and lighting mimicking real-world photography - the boundary between AI “art” & photography started to blur.

While undeniably impressive, some users lamented the potential loss of the artistic spirit. This highlights the challenge in customer-driven development – balancing the needs of diverse user groups.

v5.2, launched in June 2022, not only improved the photorealism of generated images but mesmerized users by introducing a “zoom-out” feature. This is a prime example of how MJ balanced customer-centricity with innovation.

A New Era: Power in Your Hands (v6)

The release of MJ v6 in December 2023 was pivotal for Midjourney to continue its successful run because by then OpenAI had already released Dall-E 3 (Oct 2023), which was freely available in Microsoft Copilot and produced images of quality that could now easily compete with MJ. Plus, Dall-E 3 was able to represent text very well in images.

With v6, Midjourney users not only get to generate much more coherent & photorealistic images but can also finally “draw text”!

Not only this, MJ v6 also introduced the --cref & --cw parameters to address users’ concerns about being unable to generate characters with consistency across multiple images, which hindered the flow of visual storytelling.

All these features demonstrate a strong commitment to user feedback and a willingness to refine the tool based on evolving needs.

Is Discord creating discord in the users?

Discord, while popular among tech enthusiasts and gamers, can be a maze for new users. Navigating servers, understanding commands, and competing for generation slots could be a barrier for some.

Enter Midjourney Alpha - a standalone website designed specifically for generating images.

This translates to a streamlined experience. Users now have a clear interface, intuitive workflows, and dedicated tools for crafting their artistic visions.

Here's what this move means for you:

• No More Chat Chaos: Forget navigating your way around Discord commands, Midjourney Alpha puts image generation front and center.

• Focus on Creation: A dedicated interface means less clutter and more room for unleashing your creativity

• Simplified Workflow: Intuitive tools and clear instructions make crafting your dream image a smoother process

Currently, Midjourney Alpha is available only to users who have generated over 1000 images, but it will be rolled out to everyone soon! Trouble needless to say already has access, so check out some quick examples.

This migration isn't just about convenience – it's about opening the door to a wider audience. Midjourney Alpha removes the geeky barrier of Discord, making AI-powered art creation accessible to anyone with an artistic itch. Whether you're a seasoned creative professional or a curious newcomer, Midjourney Alpha is poised to be your launchpad into the world of AI-generated art.

Where is the intrigue?

Though Midjourney was the first bully in town, there have been many image generators subsequently.

Here’s a quick comparison of Midjourney against Stable Diffusion & Dall-E 3 (integrated in Microsoft Designer & Copilot)

• Need ultra-realistic images that exactly match your description? DALL-E 3 is your best bet. Think of photorealism or capturing a specific mood.

• Midjourney is your go-to for stunning, dreamlike art with a touch of the surreal. Great for concept art and artistic exploration.

• For super precise control over every detail, Stable Diffusion is the tool. Perfect for technical illustrations or fine-tuning specific parts of an image.

With Midjourney Alpha on the horizon, AI art creation is becoming more accessible than ever. Whether you're a seasoned artist or a curious beginner, Midjourney offers a powerful tool to unleash your creativity and explore the potential of AI-generated art.

Whatcha Got There?!🫣

Buckle up, tech fam! Every week, our dynamic duo “Spark” ✨ & “Trouble”😉 share some seriously cool learning resources we stumbled upon.

Your Wish, Our Command 🙌

You Asked 🙋‍♀️, We Answered ✔️

Question: With the rise of generative AI, how do you see the role of data scientists changing? Will it reshape data scientists' work, shifting them from task-oriented roles to more strategic ones?

Answer: Absolutely! Generative AI is equipping data scientists with new tools to tackle bigger problems. Tools like GitHub Copilot, which uses generative AI, can churn out code based on your prompts. This frees up data scientists from tedious tasks, letting them focus on the big picture.

Think of it like a sous chef. Before, data scientists were prepping all the ingredients (data cleaning, feature engineering). Now, generative AI can handle those repetitive tasks. This lets data scientists become the head chef, strategizing the recipe for success (model building, interpreting results).

Gartner predicts that 20% of top data science teams will morph into "Cognitive Science" consultancies by 2026, highlighting this shift. It's not about the data scientists being replaced, but rather being augmented - as long as they are ready to upskill proactively, with a growth mindset.