Stanley Chen

Worth Accelerating. AI did not change how I think. It changed how fast I can test whether I am right. That is not a small distinction. Most creative directors are using AI to generate. I use it to verify. The thinking remains mine. The infrastructure accelerates it. If your value is in the output, you are replaceable. If your value is in the system behind the output, you are not. AI makes weak creative directors redundant. It makes strong ones faster. The question is not whether to integrate it. The question is whether you have something worth accelerating. Tools do not replace thinking. They expose whether thinking was ever there. #CreativeDirection #BrandStrategy #AIinDesign

4

When I tell people Artwrld tracks 800+ galleries in NYC and 300+ in LA, they always ask the same question: how do you get the content? They assume we built some insane AI scraping agent ripping data off the web. I tried that. It failed miserably. The truth? It's a thoughtfully manual process — backed by tools we built to predict which galleries are likely to have upcoming openings. The curation is the product. On the tech side: we started with Supabase, Strapi as a CMS, and Airtable sitting on top of everything. It worked. Until it didn't. Then I vibe-coded the dashboard of my dreams directly in Claude, wired it straight to our Strapi endpoints — and it just clicked. Still iterating. But it's slick, it's scalable, and it's coming to a city near you.

62

11 Comments

👽 Tim Cook vs the AI Slop Apple just revealed how it made the new Apple TV animation — and the secret shocked everyone. No CGI. No VFX. Just glass, light, and pure craftsmanship. The entire sequence was filmed with real reflections and refractions — a love letter to the art of making things by hand. In an era where everything can be generated in seconds, Apple reminds us how much beauty still lives in the manual. Tim Cook just declared war on AI slop. 😄

3

2 Comments

Stop playing with Studio Ghibli avatars. The AI image generation space just took a quantum leap forward. Both ChatGPT and Gemini have recently released major updates to their image generation capabilities, and the results are nothing short of astounding. If you haven't tried them yet, do so now. They're available in ChatGPT 4o to all accounts or via the Google AI Studio using the "image generation" Flash 2.0 model. These tools are so powerful that OpenAI CEO Sam Altman recently tweeted their "GPUs are melting" from the overwhelming usage. While the internet obsesses over turning selfies into Ghibli characters, the true power of these advancements isn't in creating whimsical profile pictures. It's in transforming how your business works. **Product Visualizations** With their massively improved text rendering capabilities, you can now create high-quality product mockups, diagrams, and visual prototypes in minutes instead of days. Think custom packaging designs, product concept visualizations, and promotional materials - all without graphic design skills or expensive contractors. **Marketing Asset Creation** Generate compelling images for social media, email campaigns, and website content that perfectly align with your brand voice. The new models excel at creating cohesive visual narratives that would previously require substantial design resources. **Visual Explainers** ChatGPT and Gemini's image generation can now create clear, readable infographics, process diagrams, and visual guides with properly rendered text - perfect for internal documentation, customer education, or sales materials. **Menu, Catalog & Brochure Design** Their ability to generate beautiful menus with perfect text rendering opens up possibilities for SMBs to quickly create and iterate on visual collateral without specialized design skills. Many SMBs are already quietly transforming their operations while competitors debate whether to dip a toe in the water. They're replacing expensive design agencies, reducing time-to-market, and rapidly testing visual concepts that previously would have taken weeks or months. The gap between businesses using these tools and those who aren't is widening daily. What's your business doing with these new capabilities? Are you still playing with anime avatars, or are you transforming your operations? Drop a comment with how you're using these new image capabilities. ✨ ✌🏻 ✨ If this kind of advice is helpful, then you'll love my AI for SMBs Weekly newsletter. Click "view my newsletter" next to my name or a subscribe link is in the comments . hashtag#GenerativeAI hashtag#SMB hashtag#VisualMarketing

3

1 Comment

Doc Williams invited me to go live on The Brand Factory to build an AI-powered doll generator — in real time. I said yes immediately. I figured: We’ve got plenty of visual references. Modular doll design? Easy. Drag-and-drop actions? Even easier. But what I learned on air is this: 🚫 Prompting AI to build a doll generator — even with clear instructions — is not that simple! 🧠 It revealed how much design logic, image structure, and interactivity mapping we take for granted as humans, who clearly understand human bodies.

7

Liquid Glass isn’t a catastrophe. It’s a countdown. Some see Apple’s new design direction as a break from clarity, simplicity, and the legacy of Jobs. I get it. As someone who works in visual storytelling, I understand the discomfort when familiar patterns shift. But I also think this shift is intentional — not just aesthetic, but strategic. Liquid Glass feels like a step toward interfaces that don’t just live on screens, but in space — responsive, adaptive, and aware of context. In a world inching toward spatial computing and ambient interaction, this design language might be preparing us for what comes after smartphones. There’s a lot still to refine. And the skepticism? It’s not just valid — it’s necessary. It keeps the conversation grounded in usability, accessibility, and human experience. But maybe this isn’t Apple losing its way. Maybe it’s Apple slowly redefining the map. #LiquidGlass #VisualDesign #SpatialUI #AppleDesign #DesignEvolution #PostSmartphoneFuture #HumanCenteredDesign

10

13 Comments

Click a selection in a photo or video, and get it as a photorealistic, textured 3D mesh, automagically! SAM 3D: Powerful 3D Reconstruction for Physical World Images Select an image from a photo or video still, and the AI will extract it into 3D mesh geometry, with textures, even if the object is occluded and only seen from one angle. It's like magic, but with code. Some useful aspects to this: - #videogame / #XR assets - #digitaltwins - Training for #robotics - Simulation primitives for AI agents (training/learning) - Production ready models for #ads + #ecommerce (meh, closer-r, not quite there yet, IMHO, but that this exists at all, for free, is rather amazing) What's concerning is that as this gets better, and it will, ever faster, it collapses an entire department and pipeline into several minutes. This won't fully strip away efforts done with scanners and #photogrammetry, but it will, in time. Consider a tool such as this along the lines of not just utility for a specific function (like 3D modeling, texturing, etc.). This becomes data leverage. Every photo becomes spatial, just like every video. Every "thing" is a 3D primitive. It is the foundation for understanding the world like a human, with greater context and physics. Photo to 3D to video to simulation to agent interaction is now a pipeline unto itself. Stay tuned, more to come. The #MetaAI blog on the topic: https://lnkd.in/gQY5p3tW Download the models (#ComfyUI nodes for this are already live): https://lnkd.in/gEsFd85Z Play with the demo https://lnkd.in/gAPi5kJd _ #SegmentAnything #SegmentAnything3D #SAM3D

16

4 Comments

Investigating more physical interactions on the web. I took one theme " liquids " and tried to build a small UI system around it. Gum that stretches toward your cursor. Ink that floods the button from inside. Wax that melts and drips with actual gravity. Oil that blooms iridescent under your mouse. Sure, buttons are the easy starting point. But what I really want to know is how far this language can expand🤔 build with #ClaudeCode an Figma (for more direct visual references)

582

19 Comments

🤖 OmniHuman 1.5: a new generation of expressive AI avatars BytePlus just released OmniHuman 1.5, its most expressive and cinematic AI avatar yet. From a single image and a short voice clip, it now creates film-grade expressive videos that feel alive, with natural gestures, perfect lip-sync, and fluid motion. ✨ What’s new: 🔹 Cinematic lip-sync and human-like body language 🔹 Multi-character scenes with smooth camera motion 🔹 Music and emotion synchronization 🔹 Gestures that match tone, rhythm, and intent 🔹 Full avatar control: move the camera, direct emotions, create performance 🎭 With this update, Dreamina by CapCut also integrates OmniHuman 1.5, giving creators real-time direction over AI avatars. A new visual storytelling era has begun. 🔔 Follow me for the last AI updates! #AInews #OmniHuman #BytePlus #Dreamina #AIavatar

5

1 Comment

Yesterday, I visited Meta's NY headquarters to try their new glasses. Meta’s new glasses are more than just hardware upgrades— they’re a glimpse at how mixed reality naturally integrates into everyday life. 🔹 Style + seamless control → glasses that look good and feel natural to wear 🔹 Neural Band → tracking natural movement, making interaction effortless 🔹 Still in the real world → not pulling you into VR, but layering digital content into your environment This is the milestone: mixed reality that feels less like a gadget, and more like something meant to exist in your world. This is a huge step toward the inevitable future where AR media and mixed reality seamlessly blend into everyday life—first through the phone that’s already in your hand, and soon through glasses. Watch the 4min recap below 👇

8

Full video can be found on my website: https://lnkd.in/efCxWeEM In this video, I break down the full process of reimagining these classic 90's cartoon character for my interpretation of what 3D spin-off of Hey Arnold! would look like, called 'Ayo, Gerald!' where Arnold's best friend Gerald Johanssen is the star of the show. The concept for this "trailer" takes place a few years after the time of the original show when Gerald and Arnold are in middle school. Throughout the video Gerald goes on different adventures, from winning the back-to-school talent show, skateboarding around town and hitting the game winning home run. 🛹 For the process I started by experimenting with hand-drawn textures to create my own spin on Arnold and Gerald in a 3D world. 🛹 You’ll see each scene, including Gerald skateboarding, heading to the store, hanging out with Arnold, making music in his room, and performing at a talent show. I’ll walk you through the steps I use: 🛹 Blender for modeling, lighting and rendering 🛹 iClone for animation 🛹 Substance Painter for Texturing 🛹 Ableton Live for music and sounds 🛹 After Effects for editing everything together

15

7 Comments

Traditionnal design processes were built for a different time. The AI-native era needs a different kind of design partner. One that thinks at the speed you build, integrates into the workflow you already have. The way I think about design has always been applied, not conceptual. Every output needs to be immediately usable. Every decision needs to connect directly to what the product ships next. In practice that means: • Clear actionable UX direction, not a full research report. • Interfaces delivered in a way your stack can consume directly. • A design system crafted at the product level, not the entire company. • AI-friendly guardrails that keep everything coherent as your product scales. Everything compressed to what is most impactful now, not in 3 months. That's what Ghostudio is here for. If you want to learn more about it: ghostudio.design/services See you around! ✌️ G

2

In partnership with Luma After 8 years in design, few tools surprise me. In one minute, Luma Uni-1 did a brand audit that usually takes days. I dropped a URL—[theivydotso]. Got back 11 visual assets in my own brand language. No prompts. No references. No hex codes. Just the URL. Then I tried it on [ceroflodotcom]—a medical device company. Different industry, different tone of voice. One minute later: a full visual pack with medical terminology, the right stylistic direction, and a Brand Language Guide. This isn't "images from a prompt." This is a brand intelligence audit. Uni-1 catches issues in its own output at the senior-designer level: "edges slightly digital vs organic." I type "fix this issue," and it knows which issue, because it flagged it itself. This is what a senior practitioner notices: a tool exceeds your expectations when it thinks with you, not after you. This isn't a tool. It's the presence of a designer in my canvas—one who reads context and catches mistakes before I do. But here's the part that matters more, and where I shift hats—I also build products. Right now, I'm shipping a SaaS on Next.js, Node, and Vercel. So when Luma releases Uni-1 as an API, I read it differently than a designer would. Until now, Uni-1 was a model. With the API release, it becomes a foundation, something you build on, not just something you use. You can replace an old model in an existing pipeline with Uni-1. You can wire it into a node-based creative workflow as the reasoning core. Or you can build something new entirely—an app, a tool, an agent—with Uni-1 as the engine, generating outputs that stay true to a brand through natural language alone. The people who benefit from this aren't individual creators experimenting on a Sunday. It's builders, engineers, production teams, and agencies running real pipelines. Companies shipping at scale. Uni-1 isn't a toy anymore. It's infrastructure. A dependency. Try Uni-1 API access here: https://lnkd.in/dpDwMF9v Luma Has anyone else tried it? What caught you off guard?

6

1 Comment

When people think of Generative AI, they usually think of an open text prompt. The new Rotate Object feature in Photoshop takes a 2D layer, turns it temporarily into a 3D object that you can reposition, and then harmonizes it back into your composition. This is professional level control with AI.

57

1 Comment

Reinvention is everything in this AI-driven era. (It is also very human.) Just came across Dropbox’s interactive brand guideline, what a breath of fresh air. Smart UX, thoughtful interaction, and a seamless blend of design, code, and creativity. AI can resemble and assemble, leveling the game. But it’s still the human touch that goes beyond the box. #BrandDesign #UXInspiration #InteractiveDesign #CreativeTechnology #AIandDesign

4

Would you guess this is AI-generated or shot for real? This entire Sun Cola stadium sequence I generated in 10 mins using AI workflows with only minimal graphic overlays added in After Effects. What’s becoming scary good is the consistency: - Natural crowd reactions in high quality - Stable product appearance without morphing - Realistic and natural camera movement - Facial emotions that feel surprisingly real - All sound effects generated from the same video prompt A few months ago, shots like this would completely fall apart. Now, AI video is getting dangerously close to real commercial production quality — and it’s becoming easier to make bold AI production decisions without the same fear of morphing, broken frames, or unstable results that held productions back before. The tools are evolving fast, but the workflow and creative direction still make the biggest difference. Open to collaborations for cinematic AI ads and branded content.

42

7 Comments

Let's rebrand "Throwing spaghetti against the wall and see what sticks" as "Desire path theory." Probly not gonna make fetch happen here, but there's an interesting connection for me between the two ideas and how they're viewed as a creative modus operandi, esp. in light of my takeaways from Adobe Summit 2026. "Throwing spaghetti against the wall to see what sticks" (great visual btw) has always had a comical, but negative connotation. We use it either as an accusation of somebody else's lack of courage / decision-making, or as an admission to ourselves that, out of time and answers, we have no other choice but to sling it and hope. But that precise mentality, endorsed by our mutual friend Gary V. and others, is dominating the TikTok algorithm content game. It's volume and versions. It's "clipping" multitudes w/ slight variations of the same basic content, all with the express intent of seeing what will stick. It comes with the humility that, as smart as you are (ACD, CD, Motion Designer, Account Supervisor, CMO, whoever) you actually have no idea which version of the video will get 4k views, and which will get 4MM. The people will decide. Now, on to "desire paths," a phenomenon familiar to architects and urban planners. It was famously studied and employed in college campus planning in the early 1900s where, instead builders paving sidewalks based on blueprints, first they waited first to see where students walked. Ohio State, Michigan State, and Illinois Institute of Technology are some notable examples. It's the opposite of lazy design. It's superior design because it leads w/empathy. "Paths that humans prefer, rather than the paths that humans create." Now made possible in marketing by creating content in volume with versioning driven by curiosity. Which path to our brand will the people choose? This is where MERGE is headed with #wholehumanmarketing. [ Photo credit https://lnkd.in/gUJJSZu5 ]

10