Exploring AI Art with Midjourney

Midjourney Personalization: The thing that is simple and obvious but I missed it During the latest Midjourney Office Hours, the CEO clarified something simple that completely reframes how I approach personalization: > Profiles learn from prompts. Moodboards learn from images. Here’s the core insight: When ranking images to train a Profile, you're not just training it on what looks good... you're training it on the prompt structure behind that image too. The system tracks how your aesthetic preferences align with the prompts that generated those outputs. In contrast, Moodboards ignore prompts entirely. They extract only visual information from the images you upload: color, texture, composition, lighting. From now on, I’m ranking differently. I won’t just glance and pick A or B. I’ll open the prompt behind each image, look at the language, structure, and composition... and then choose. Because ranking is more than just training the look. You’re training Midjourney on how you think, what kinds of prompts resonate, and which patterns match your intent. Think of it like this: I’m building a dataset for Midjourney to understand what I actually love. Prompt: > 80s airbrush poster of a model with braided hair, against a red background with black and purple highlights. The model is wearing beads and the image is a close-up face shot, in the style of Hajime Sorayama. --chaos 15 --ar 85:128 --profile j5d43pf --stylize 250 --v 7 #MidjourneyAI #AIArt #PromptEngineering #VisualLanguage #CreativeAI #MidjourneyV7 #HajimeSorayama #Personalization #Moodboards

A year ago, AI image generation felt full of promise—but the results were often clunky. Distorted hands, garbled text, and inconsistent detail were common. Fast forward to 2025—and the progress is hard to ignore. Still not perfect, but a huge step forward. 🛠 What’s Changed? - Photorealism: Tools now capture lighting, texture, and fine detail with near-photographic accuracy. - Readable text: Clean, embedded text in images—finally solved. Perfect for ads, memes, and branded content. - Creative input: Upload a sketch or rough idea, and refine it instantly. Think logos, mockups, or 3D concepts. - Speed: Newer models generate high-quality visuals in a fraction of the time. - Bonus: You can now create an image of yourself as an action here. Go ahead everybody's doing it. 🔥 Tools to Watch -GPT-4o (OpenAI) – Conversational image creation inside ChatGPT. - Midjourney v7 – Hyper-detailed, consistent visual style for concept and brand work. - Flux.1 – Photorealism tailored for product and campaign imagery. - Ideogram – Solves for crisp, usable text in image generation. 📌 Why It Matters You can now use these tools with little adjustment to create marketing content, product visuals, and other creative assets—quickly and at a high quality.

Part 8 of my Q & A Series Q. How Do You Keep AI Images Consistent Across an Animated Short Film Series? I’ve been developing a series of AI-assisted animated short films that rely on visual continuity as much as storytelling. Every scene needs to belong to the same world—same mood, same light, same story realism. AI IS MERCURIAL. PHOTOSHOP IS PREDICTIVE Consistency is easy to lose once you start mixing tools. One image emerges glowing and cinematic, the next looks like it wandered in from a Saturday-morning cartoon. My reference point for cohesion is Ciro Marchetti—his worlds are rich, luminous, and believable. Reaching that level of cohesion across multiple AI frames takes more than a strong prompt. It takes direction. What follows is the workflow I use to keep an AI-generated film series visually consistent: using MidJourney to generate, and Photoshop to refine, balance, and ground the results in one cinematic universe. I. MidJourney is a generator. Photoshop is a director Generative tools are great for “let’s see what happens.” You type in a prompt, cross your fingers, and sometimes you get gold. Diffusion models reinterpret your prompt every time—tiny changes shift the entire look. If you’re maintaining consistency across characters and scenes, that unpredictability works against you. II. Photoshop doesn’t guess. It lets you decide. When you composite in Photoshop: • You control scale, lighting, and placement. • You paint realistic shadows and reflections instead of hoping the algorithm gets it right. • You match tone and texture with a few smart adjustments instead of wrestling with “style strength.” III. The best workflow blends both worlds. Here’s what worked for this cinematic AI project: 1. Design characters and backgrounds in AI. Use #Midjourney, #OpenArt, #Seedream, etc. to generate assets at high resolution with clean edges. 2. Composite in #Photoshop. Drop every element into its own layer. Resize, relight, and blend. • Use Match Color to unify lighting. • Add a light Gaussian Blur to push distant figures into depth. • Paint shadows manually—AI sometimes misses those. 3. Run a cinematic regrade. Once your composition looks natural, export it and regrade (Seedream 4.0) for tone, light, and atmosphere. 4. Animate in #LumaDreamMachine. Luma handles camera motion with cinematic depth. 5. Finish in #CapCut or your preferred editor. Add timing, music, and sound. IV. Why this matters AI is mercurial. Photoshop is predictive. If you’re building a narrative world, you need both. V. Final Thoughts I’m not anti-AI. I’m anti-automatic. AI tools are brilliant, but they still need human direction. The best results come when you treat them as collaborators—AI for ideas, Photoshop for intention. So when your next scene looks too soft, too plastic, or too “AI,” don’t scrap it. Open Photoshop. Adjust the light. Fix the edges. That’s where your film starts to feel real. #lyndacathcart #motiondesigner #aianimation #curio

My single biggest pain point in AI video production is this the gap between aesthetic excellence and actual control. Midjourney’s cinematic look isn’t luck. It’s the result of a very opinionated model design that treats beauty as a core objective. That's why MJ images feel dramatic and polished and why other models struggle to match the vibe even with the same references. The whole system is optimized for a specific aesthetic lane: cinematic lighting, rich tonality, painterly textures, and strong visual cohesion. Prompt adherent models like GPT5, on the other hand, are built to be neutral and obedient. They triy to give you exactly what you ask for, even if what you asked for isn’t particularly beautiful. Midjourney tries to make the most beautiful version of whatever it thinks you’re asking for even if it has to change your instructions. A big part of this comes from the style baked directly into the model. MJ has opinions about lighting, palettes, composition, and camera language. Those opinions show up in every output. DALL·E intentionally removes that bias so it doesn’t impose a style on users. Great for accuracy. Less great for cinematic impact. Then there’s the post-processing. MJ’s upscalers don’t just sharpen. They generate new detail that fits its aesthetic. It’s effectively a creative pass layered on top of diffusion. Most competitors avoid this because it reduces controllability. For MJ, it’s a feature. And of course, the community plays a big role. Millions of daily aesthetic preferences push the model toward high-gloss, high-drama imagery. That feedback loop shapes the model’s visual taste over time in a way other platforms simply can’t replicate, because their user bases are broader and their objectives more neutral. This is why feeding DALL·E a Midjourney image won’t produce Midjourney-style results. Once you understand that, you start using these tools differently. Midjourney for taste. The others for control. And the real magic comes from knowing how to combine them in a workflow instead of expecting one model to do everything.

As industrial designers, we constantly strive to find better, faster ways to ideate and iterate. One of the most exciting developments in design workflows recently is leveraging AI tools like MidJourney’s Edit & Retexture functionality to transform basic CAD forms into high-quality visual concepts in minutes. It was a while since I used Midjourney. But thanks to seeing one of the LinkedIn posts by Hector Rodriguez , I was itching to try it. I recently experimented with this approach using a foundational CAD model. I had made this as one of the form explorations through CAD for a coffee machine.I prompted MidJourney to retexture and visualize it in various material and finish combinations. The results? A series of diverse, photorealistic outputs that allows me to explore design possibilities I may not have considered otherwise. This workflow highlights some key strengths: 1. Speeding Up Concept Ideation: AI tools can generate multiple aesthetic directions from a single CAD base almost instantaneously. This means you can explore and test design ideas quickly, without committing hours to detailed rendering or material adjustments in software like Blender or Keyshot. 2. Streamlining CMF Exploration: Traditionally, exploring different colors, materials, and finishes (CMF) can be a long-drawn-out process, requiring meticulous work in rendering software or Photoshop. With AI, you can bypass this step and instantly visualize multiple CMF options. This not only saves significant time but also allows for rapid iteration and refinement. 3. Accelerating Design Evolution: With rapid outputs, you can visualize the potential of your design’s form and materiality in real-world contexts. This allows for informed decision-making early in the process, saving time during later-stage refinements. 4. Enhancing Creative Exploration: By integrating AI tools, we can step beyond our usual design instincts and uncover unexpected design solutions. This not only enriches the process but also pushes boundaries in creativity and innovation. For industrial designers, this hybrid approach—merging CAD fundamentals with AI-enhanced retexturing—opens up new opportunities to iterate faster and more effectively. Once the most promising directions are identified, we can dive into refining the details, ensuring manufacturability, or rendering them perfectly in Blender, Keyshot, or similar tools. This newfound workflow feels like a game-changer to me, especially for balancing creativity with tight deadlines. What do you think about this tool? #industrialdesign #ConceptIdeation #CMF #CMFExploration #productdesign #MidJourney #ai

Many think making incredible AI art requires artistic genius. But Rory Flynn shares a secret - the art is actually in the prompt! He explains prompting AI is like storytelling. Vague prompts give vague results. But concise, detailed prompts set the vision to generate amazing images. Provide the context AI needs - mood, style, colors, composition - just like directing any designer or photographer. This clarity and direction focuses AI creativity. Listen now to learn prompting best practices to get stunning, consistent visuals from AI every time. Topics we discussed:  ✅ How Midjourney can create virtually any visual asset imaginable, from icons to infographics to full ad campaigns ✅ Clever ways brands are using AI-generated images to connect with customers on an emotional level ✅ Pro tips for writing detailed prompts to get the perfect lighting, angle, and composition ✅ The untapped world of AI upscaling, turning pixelated images into ultra high-res masterpieces ✅ How to ensure original, legal images that capture your brand essence Even if you're not a designer, Rory Flynn makes mastering Midjourney feel effortless. You'll walk away overflowing with ideas to step up your visual game using this amazing new technology. Get actionable insights for leveraging AI’s tireless capabilities while keeping talented people inspired. Become an artist through your prompts and unlock AI's creative potential!

Ever tried to be a superhero? 🦸♂️ I've been experimenting with AI image generation for a while now, and one thing is clear: the best way to learn is to get your hands dirty. My latest challenge? Crafting a Superman-inspired portrait of myself using Midjourney. Let’s just say the journey had its fair share of twists. The first few attempts? Far from perfect. Midjourney seemed determined to pair my business suit with a cape—close, but not quite there. The breakthrough moment? Learning about the character weight parameter (--cw). A tip from Ferdaus made all the difference. By setting the --cw to 0, I finally got the AI to focus on what I really wanted—the face. And just like that, I was flying high in full superhero mode. 🌟 This experience reinforced that the real learning happens when you dive in and experiment. Want to try it yourself? Check out this guide on character references in Midjourney: https://lnkd.in/g7KhDiKB Have you experimented with AI image generation? #AIImageGeneration #HandsOnAI #ContinuousLearning #AIinMarketing #AIExperiments

About 2 years ago, I started experimenting with AI and began including it as a part of my creative workflow, and it's truly shocking to see how much progress has been made with this technology in such a very short time Here comparing an image from one of my first generated series (Left) using #MidJourneyv4 and a more recent image (Right) using #MidJourneyv6 The aesthetic differences between MidJourney v4 and v6 stem from advancements in image quality, style realism, and control over fine details. Below are 5 key aesthetic differences i've noticed so far between the two versions: 1. Image Quality & Resolution: While v4 delivers visually impressive images, they tend to be more artistic and painterly, with softer textures and some limitations in sharpness. V6 on the other hand significantly improved image clarity, rendering hyper-realistic textures and intricate details with higher resolution and a more lifelike quality, even in complex scenes. 2. Realism vs. Artistic Style: v4 is known for its dreamy, stylized, and surreal aesthetics, it excels at creating abstract or exaggerated forms, often adding a painterly or illustrative feel to the output. V6 on the other hand leans more toward realism, capturing accurate lighting, materials, and proportions, which makes the results feel more polished and true-to-life. It still handles artistic styles but with much greater control over realism when desired. 3. Handling of Human Anatomy and Faces: Faces and human anatomy often suffered from distortions and inconsistencies, particularly in highly detailed prompts or complex compositions in V4 while V6 Significantly produces realistic human anatomy and faces, with far fewer distortions. Facial expressions, proportions, and skin textures are more natural and finely rendered. 4. Lighting & Shadows: Lighting effects in V4 can be dramatic but often lack the nuanced gradient transitions found in real-world photography, giving a more stylized or even flat appearance. V6 on the other hand produces realistic lighting and shadow play with smoother transitions, depth, and accurate reflection of light sources, enhancing the photorealism of scenes. 5. Complexity and Control: MidJourney v4 is great for generating dynamic and expressive compositions, but users often had less control over fine-tuning intricate details, which could result in more abstract or less precise outcomes. However, V6 Allows for more complex and precise control over elements like composition, color palette, textures, and fine details, making it easier to direct the aesthetic outcome. Bottom line, if you want to create art, use V4 but if you want to mimic detailed photography, use V6. _ #ai #midjourney #aiusecases

The internet's buzzing with Ghibli-inspired AI art! 🍄✨ From Elon Musk to fantastical landscapes, the distinct, enchanting style has captivated us all. It's a testament to the power of AI to replicate and reimagine artistic visions. But this is just the beginning! Let's explore how you can unleash your own creative potential beyond this trend: Beyond Ghibli: Dive into the AI Image Revolution! 🎨✨ Forget limitations – imagine crafting visuals that were once only dreams. Want to see your wildest concepts come to life? Here's how: Dive into Detail with Advanced Prompting: Don't just ask for a picture; tell a story! Use descriptive language, specify moods, lighting, and styles. Think "a cyberpunk cityscape at dusk, neon reflections on wet asphalt, dramatic cinematic lighting." Explore Hidden Gems: Midjourney: Known for its artistic and surreal outputs, Midjourney offers a unique creative playground. Stable Diffusion: Open-source and highly customizable, Stable Diffusion lets you fine-tune every aspect of your creations. Leonardo AI: A platform that gives you a large amount of free generations, and is very user friendly. Adobe Firefly: If you are within the Adobe ecosystem, this is a very good tool to use, and is very well integrated. Refine with Post-Processing: Take your AI-generated images to the next level with tools like Photoshop or GIMP. Adjust colors, add details, and create truly unique masterpieces. Combine and Remix: Don't be afraid to mix and match! Use multiple AI tools to generate different elements, then combine them into a single, stunning image. Use ControlNet: ControlNet is a neural network structure to control diffusion models by adding extra conditions. For example, you can control the pose of a person, or the edges of an object. This will give you much more control over the generated image. Why Wait? The future of creativity is here, and it's in your hands. Whether you're a designer, marketer, or just someone with a spark of imagination, AI image generation is your ticket to a new world of possibilities. What will you create? Share your AI art in the comments! Let's inspire each other. #AIArt #DigitalArt #CreativeAI #ImageGeneration #Midjourney #StableDiffusion #LeonardoAI #AdobeFirefly #ControlNet #ArtisticInnovation #CreativityUnleashed #GhibliAI

Want AI to produce *real*-looking photos? There’s a weird prompt hack floating around that actually works. No “shot on DSLR” or “cinematic lighting” required. Just this 👇 The hack? Use a camera-style *file name* in your prompt. Something like: IMG_190282009.HEIC That’s it. Drop that in front of your description. Why does this work? Let’s break it down 👇 AI image models like Midjourney are trained on tons of real-world data. That includes actual camera photos — and those photos often have generic filenames like: IMG_20230915.HEIC Using this tells the model to reach into the “real” part of its training set. Let’s test this theory: Prompt A (control): "Mountain landscape from the perspective of a hiker in digital camera style" Result: pretty, but painterly. Not camera-realistic. Now, Prompt B (hacked): "IMG_190282009.HEIC mountain landscape from the perspective of a hiker" Result: sharper, more lifelike, feels like a snapshot. Wild, right? Here’s another test: Prompt A (control): "3 friends on a street corner in NYC, digital camera style Looks more like concept art" Prompt B (hacked): "IMG_93282007.HEIC 3 friends on a street corner in NYC" Boom. Realer composition, digital photo vibes. 7Bonus tip: This trick works best in Midjourney v7 for working with humans or portraits, v7 (with personalization) gives better results (I used v6.1, which is why the people in the images above look very 'AI') But even in v6.1, this hack is a game-changer. Final thought: Prompting is part creativity, part reverse-engineering the model’s brain. This hack is proof. Sometimes, the smallest tweak (like a fake filename) unlocks the best results. Try it out for yourself with Midjourney, and share what you come up with!