All posts by jnack

Fun with Omni: Changing cams, plus Wolverine

AI/ML, Google Omni

Karen X. Cheng shows a subtle but powerful model capability:

 
 
 
 
 
View this post on Instagram
 
 
 
 
 
 
 
 
 
 
 

A post shared by Karen X (@karenxcheng)

Meanwhile Christian Cantrell is looking sharp, is not downright superheroic:

Come build with Gemini Omni Flash!

AI/ML, Google Omni

I’m thrilled to say that the first big launch of my Chapter 2 at Google is here! You can now build on Gemini Omni Flash in Google Enterprise Agent Platform (aka Vertex); see docs.

TBH I was so busy helping get the release out the door, and then taking some much needed rest over the Fourth of July break, that I’ve hardly had a chance to post useful info. I’ll fix that soon! In the meantime, here’s our little intro sizzle reel:

Of beat labs & photo shoots

AI/ML, Photography

It’s always cool to see how creators are embracing new tools:

Quick tour: Creating Flow tools with natural language

AI/ML, Google Flow

My teammate Anika & I got to meet the other day with a really big creative brand the other day (more details to share soon, I hope), and they got excited about delivering super focused, relevant experiences for their customers by building on the Flow agent & apps. Here’s Anika offering a concise tour of how to create, share, and remix the latter:

Behind the scenes: How Skydio builds drones

Design

“Everyone loves love,” as they say. I’ve noted many times just how much I enjoy people enjoying their work—and that pride & purpose clearly come through in this brief glimpse into the making of Skydio drones.

Skydio makes more drones than any other company in America, and it builds all of them inside one factory in Hayward, California. In this episode we go onto the floor with co-founder and CEO Adam Bry to see exactly how the Skydio X10 comes together, from a pile of thousands of parts to a finished autonomous drone that thinks more like a flying robot than a traditional quadcopter.

You will see the steps most companies never show. Skydio waterproofs the critical electronics with a nanocoating process so the X10 can fly in the rain, fuses the drone arms together using high frequency ultrasonic welding, and hand solders the motor wires onto the power board. Inside the chassis sit an NVIDIA CPU and GPU plus a Qualcomm chip running the camera feeds, all stabilized by a custom gimbal that keeps the camera locked while the drone fights wind in the air. Then comes the part that matters most. Every single drone gets pushed through a brutal burn-in stress test, flown by hand, and run through a calibration robot before it is ever allowed to ship.

“Google Just Turned Street View Into a Video Game”

3D, AI/ML

As Bilawal puts it,

At Google I/O 2026, DeepMind shipped Maps Imagery Grounding for Genie 3 — their real-time world model can now generate interactive 3D worlds conditioned to any of the 280 billion Street View images Google has captured over 20 years. Pick a location on Google Maps, choose a style, drop in a character, and walk around.

Check out his accessible & illuminating tour of the new tech:

Omni Teapot

3D, AI/ML, Google Omni

My 16yo is lowkey impressed that at Adobe I got to work with Utah Teapot creator Martin Newell. At this point, anything that impresses a teen is very welcome. 🙂

I wonder what he’d think of Gemini Omni turning real teapots into geometry just by saying the word:

Speedy document scanning via Drive

Mobile

Approximately 2000 years ago (give or take a couple orders of magnitude), we shipped a cool Crop & Straighten feature in Photoshop. It enabled putting a bunch of images into a flatbed scanner (talk about words I haven’t typed in decades) and automatically turn them into individual images (“Lift & Separate,” as PM Karen Gauthier cheekily dubbed it).

Along those lines, but with radically more smarts & speed, check out what you can now do in the Google Drive app (who knew?):

Puppetry + AI FTW: Behind the Scenes with Timmy TPU

AI/ML, Design, Google Omni

I love the blend old-school puppetry, 3D animation, Gemini Omni, and the latest experimental video tools that went into creating TPU Training Day, the short film that debuted during Google I/O 2026.

I know you’ve heard it a million times, but it bears repeating: AI isn’t a substitute for human creativity, or in many cases even for traditional techniques. It’s just a whole new toolbox that can multiply our expressive powers.

And here’s the film itself:

Check out Google Flow Agent

AI/ML, Google Flow, Google Omni

Did I have “Google makes cool, extensible, AI-powered creative tools” on my 2026 Bingo card? I did not—and I’m happy to be wrong! Check this out:

According to the docs, you can use the Agent to:

  • Brainstorm and plan: Chat with the Agent to outline storyboards, develop visual mood boards, and turn high-level concepts into actionable prompts.
  • Generate new media: Ask the Agent to generate videos or images and select the best model to generate with.
  • Edit assets directly: Ask the Agent to edit selected media from your project.
  • Batch generate: Ask the Agent to create multiple variations of an asset at once.
  • Organize your assets: Ask the Agent to rename specific files, group selected media into a new Collection, or archive unused assets.
  • Add context & references: Drag media into the Agent prompt box from your device or project. You can also select multiple assets and let the agent know which media you are referring to.

3D typography using Omni + Flow

3D, AI/ML, Google Flow, Google Omni, Typography

Check out this cool little technique:

This is especially wild when you consider where typography stood just a couple of years ago—for which I’ll forever be kinda nostalgic. 🙂

A beautiful moment of expressivity unlocked

AI/ML, Shit That Actually Matters

Despite—or maybe because—of my line of work, I have some genuinely mixed emotions about AI. Is it about empowerment, devaluation, theft, magic? Yes. It is all, as my wife would say of me, A Lot™.

Alongside whatever else it may be, however, the tech can be a genuine enabler of human expressivity. If you don’t believe me, just take 90 seconds to read & watch this heartfelt moment:

Vibe-code your own VFX apps & more Google Flow Tools

AI/ML, Google Flow, Google Omni

Democratize all the apps!!

I think this new platform will be a major sleeper hit:

With Google Flow Tools, you can build creative workflows customized to fit your creative process. Explore a gallery of premade Tools built by creatives, remix existing ones to fit your needs, or create your own from scratch by just typing a description of what you want to create. You can shape and iterate on these Tools fluidly, adjusting them for individual projects, or singular clips and images. All users can explore Tools in Google Flow, and Google AI subscribers can create custom Tools from scratch or remix existing ones.

Check out the ways some artists have been spinning up tools & putting them to work:

Google Earth + Omni = Drone magic

3D, AI/ML, Google Omni

My friend Bilawal, who used to work in Google’s Geo group (Earth, Maps, and more), has created an eye-popping faux-drone video using Omni Flash:

Here’s another exploration, inspired by Bilawal’s:

A soldier’s sketches of World War II

Illustration

Before commencing his long & distinguished career as an architect, 19yo Victor Lundy captured life at an American GI.

He drew out his experiences from training at Fort Jackson (May 1944) to his journey across the Atlantic and then his time in France. In total, he produced a visual diary with 158 pencil sketches brings to life the wartime experience. Lundy applied his drawing skills to what was around him—training at Fort Jackson, South Carolina; forced marches; men at rest; the PX and tents; New York Harbor; aboard ship in the Atlantic crossing; Cherbourg Harbor; and French villages. Many vivid portraits of fellow soldiers and frontline danger also fill the pages. The sketches cover May to November 1944 when Lundy was wounded, with some gaps where notebooks were lost.

The eight surviving sketchbooks are spiral bound and 3 x 5 inches—small enough to fit in a breast pocket. Lundy used black Hardtmuth leads (a drawing pencil) and sketched quickly. “For me, drawing is sort of synonymous with thinking.”

He later donated his sketches to the Library of Congress. On this Memorial Day, it’s well worth taking some time to dwell with them.

Awesome examples of Omni video transformation

AI/ML, Google Omni

This is such a wild, game-changing feature:

I think Carlos gets it exactly right: “I think many are focusing on the wrong aspect of the Gemini Omni model when comparing it to Seedance 2.0, since conceptually they are entirely different things. This is a model for editing videos (like Nano Banana) like we’ve never had before!“

“Nano Banana for video” is here!

AI/ML, Google Omni, Nano Banana

I’m so pleased to be playing a very small role in bringing breakthrough video transformation to the world. Check out the new Gemini Omni:

The team writes,

We’re introducing Gemini Omni, where Gemini’s ability to reason meets the ability to create. Omni is our new model that can create anything from any input — starting with video. With Omni, you can combine images, audio, video and text as input and generate high-quality videos grounded in Gemini’s real-world knowledge. You can also easily edit your videos through conversation.

Today, we’re rolling out the first model in the Omni family: Gemini Omni Flash, to the Gemini app, Google Flow and YouTube Shorts. In time we will support output modalities like image and audio.

Conversational video editing is the real breakthrough:

Check it out & let us know what you think!

Putting in the mental reps

AI/ML, Idle Philosophizing

I keep finding myself thinking of this observation from Paul Graham:

“In preindustrial times most people’s jobs made them strong. Now if you want to be strong, you work out. So there are still strong people, but only those who choose to be. It will be the same with writing. There will still be smart people, but only those who choose to be.“

To reiterate from a previous post, quoting Keep the Robots Out of the Gym:

Think very carefully about where you get help from AI.

I think of it as Job vs. Gym.

  • If we’re working a manual labor job, it’s fine to have AI lift heavy things for us because the actual goal is to move the thing, not to lift it.
  • This is the exact opposite of going to the gym, where the goal is to lift the weight, not to move it.

He argues for identifying gym tasks (e.g. critical thinking, problem solving), and for those use just your brain (with minimal AI assistance, if any).

My primary metric for this is whether or not I am getting sharper at the skills that are closest to my identity.

Try personalized image creation via Gemini

AI/ML, Google Photos, Nano Banana

As I often said back in the day, Google’s longstanding mission is to “organize the world’s information and make it useful.” A lot of that information is photographic, and a lot of that information is private; hence the value and power of Google Photos. It knows (with your blessings) who’s who, what places are important, and so on.

Now Nano Banana can leverage that info to make fun and beautiful things on your behalf.

Since you can already organize and label groups of people and pets in your library, those labels provide the context that Gemini needs to make your images feel truly yours…

With those labels in place, you can simply ask Gemini to “create a claymation image of me and my family enjoying our favorite activity” and Gemini can generate that specific image for you automatically. You can also experiment with different styles like watercolors, charcoal sketches or oil paintings. You can turn a quick idea into a custom creation, saving you the trouble of searching for, downloading and re-uploading files just to see a concept come to life.

3D pets, now & then

3D, User Interface

Check out this charming & revealing image->3D creation from Gábor Pribék:

Folks in the replies fondly remembered back to the Cat Explorer demo for Leap Motion (rest in power):

Google Earth + Nano Banana? Go Go Godzilla!

3D, AI/ML, Nano Banana

I love this kind of simple, scrappy creativity,:

Here’s the Chrome extension:

  • Capture any Google Earth 3D view
  • Transform with AI (Nano Banana Pro) into cinematic shots
  • Generate videos (Veo 3.1) with customizable duration and audio

GenFill + Vividon = Magic

AI/ML, Photography, Relighting

It’s insane what we can do now—from object removal to lighting changes—that was simply out of the question even a year ago.

Check out this little progression of edits, starting with the newly enhanced Generative Fill in the Photoshop beta, followed by a couple of steps of Remove, followed by a pass with Vividon & a few tweaks in Camera Raw (running inside PS):

Nutty & I’m here for it. Per PetaPixel,

Co-founder and Chief Innovation Officer Marcus Kurn adds that the ability to deliver two or three lighting variations alongside every final image is a real differentiator: “once you start delivering two or three lighting variations with every final image, your clients will never want to go back.”

Vividon relighting comes to Photoshop

AI/ML, Relighting

“No prompting, no friction. Just incredible results.”

As I mentioned back in January, Vividon offers new generative relighting tech that promises amazing realism & identity preservation:

Vividon places every relight on its own Photoshop layer. Adjust opacity, change blend modes, paint in or out exactly what you want, or remove it entirely. Your original always stays untouched.

Check out a 10s demo below, and visit their site for a more interactive preview:

And here’s a full 2-minute tour:

“A vehicle that cares back”

AI/ML, Idle Philosophizing

“People will forget you said, people will forget what you did, but people will never forget how you made them feel.” — Maya Angelou

I’ve reflected on this maxim countless times over the last couple of years, as I’ve considered the relationships I want with AI—particularly with notional creative partners. I want a partner who cares—who (which?) actually takes the time to get to know me, asking thoughtful questions, noodling on answers, and genuinely taking my feedback to heart.

I thought of this while listening to Stewart Brand talking to Ezra Klein the other day. Check out this poetic & provocative passage:

Well, it wound up that, basically, most of the book is Chapter 2, “Vehicles.” And the land vehicle that humans have used for 6,000 years is a horse, and the horse takes a lot of maintenance.

I’ll read something here from the book, if I may. There’s this philosopher named Albert Borgmann who wrote:

You cannot remain unmoved by the gentleness and conformation of a well-bred and well-trained horse — more than a thousand pounds of big-boned, well-muscled animal, slick of coat and sweet of smell, obedient and mannerly, and yet forever a menace with its innocent power and ineradicable inclination to seek refuge in flight, and always a burden with its need to be fed, wormed and shod, with its liability to cuts and infections, to laming and heaves. But when it greets you with a nicker, nuzzles your chest and regards you with a large and liquid eye, the question of where you want to be and what you want to do has been answered.

And I end with: “I wonder if that might come again someday — a vehicle that cares back.”

The scarily beautiful animation of Sincitium

AI/ML, Illustration

Side note: “Macrófago” is 100% the best word I’ve learned all week.

AI filmmaking turns a (creepy, fun) corner

AI/ML

This is the first time I can recall watching a genuine narrative (not a handful of gee-whiz demo shots) made with AI & not really caring about the production details. We’re turning the inevitable corner where it’s just the quality of ideas & narrative that’ll matter—not so much how the proverbial sausage was made.

See yourself from a new angle in Google Photos

3D, AI/ML, Google Photos, Photography

Get some fresh perspective from our amazing teammates in research:

Today we are announcing a new approach to fix scene alignment after a photo was taken. Our method, now available as part of the Auto frame feature in Google Photos, uses machine learning (ML) models to understand the scene and its spatial layout and uses generative AI to imagine the photo from that new perspective. In contrast to classical photo editing, our method interprets a photo as a 3D scene — think of a real moment frozen in time — and change the camera position automatically within that space.

How to get the most from Nano Banana

AI/ML, Nano Banana

My new teammates have posted a series of detailed tips & tech specs (e.g. you can upload as many as 14 images together with a prompt). Check it out!

1. Introduction to Nano Banana

  • The Models: Overview of Nano Banana 2 (powered by real-time web search) and Nano Banana Pro (built for high-end reasoning).
  • Core Strengths: Deep reasoning capabilities, accurate visual rendering, and premium features like text rendering and upscaling (2K/4K).

2. Technical Specs at a Glance

  • Context Windows: Up to 131,072 input tokens for Nano Banana 2.
  • Versatility: Supports multiple aspect ratios (from 1:1 to 21:9) and up to 14 reference images in a single prompt.
  • Safety: Built-in SynthID watermarking and C2PA credentials for responsible AI use.

3. Best Practices for Prompting

  • Be Specific: Focus on concrete details regarding subject, lighting, and composition.
  • Positive Framing: Describe what should be there (e.g., “empty street”) rather than what shouldn’t.
  • Director’s Perspective: Use cinematic terms like “low angle,” “bokeh,” or “aerial view.”

4. Five Powerful Prompting Frameworks

  • Image Generation: Using the [Subject] + [Action] + [Context] + [Style] formula.
  • Image Editing: Utilizing “Semantic Masking” to change specific parts of an image via text.
  • Real-Time Data: Leveraging web search to create visuals based on current events or weather.
  • Text Rendering: How to get legible, localized text in over 10 languages within your images.
  • Creative Direction: Advanced tips for controlling lighting (e.g., Chiaroscuro), camera hardware (e.g., GoPro vs. Fujifilm), and film stock.

5. The Creative Ecosystem

  • How to combine Nano Banana with other models like Gemini (for prompt engineering), Veo (for video keyframes), and Lyria (for AI soundtracks).

Photoshop, 3D, and redemption

3D, AI/ML

“Being early is the same as being wrong.” — Marc Andreessen, Vol. ~900

We put 3D into Photoshop nearly 20 years ago, and it got used by nearly 20 people total, lol. For many of the past several years, it was on the team’s “gotta throw overboard, as soon as we can find time” list—but happily that time was never found.

I am so glad to see this foundation now finding a meaningful niche, and I have high hopes for its generative future. Posing a person or thing directly is so much more intuitive than trying to precisely describe an outcome via prompt, and simple 3D manipulation + generative rendering could well deliver game-changing best of both worlds.

Canva’s new Magic Layers converter is really impressive

AI/ML, Design

As generative imaging models like Nano Banana get increasingly adept at rendering text-heavy layouts, the ability to convert those layouts into native text/image compositions is of course hugely valuable for editing. Check out Canva’s new Magic Layers feature:

I couldn’t resist trying it out with a silly infographic I made using the new ChatGPT image model, and dang if it didn’t do a pretty a great job:

“LooseRoPE” promises super intuitive illustration & compositing

AI/ML, Illustration

Man, it must be nearly 20 years ago that we started envisioning drag-and-drop-simple composition and compositing in Photoshop—back when gradient-domain painting & blending was the emerging hotness. After plenty of false starts, could these simple interaction patterns finally become mainstream? Maybe! I must know more of this witchcraft:

A love letter to the Chicago lakefront

Design

My mom and her sisters (all with the skin tone 255/255/255) spent way too many days turning themselves into rotisserie chickens here over the years. That’s one of millions of stories, ranging from shipwrecks to chicanery, plucky birds to an African American rodeo scene, that have unfolded along Chicago’s amazing lakefront. Having grown up in Illinois & visited hundreds of times over the years, I really enjoyed this tour from WTTW & Geoffrey Baer:

Recent 3D hotness from Apple & Microsoft

3D, NeRFs & Splats

Apple’s LiTo can generate Gaussian spalts with realistic view-dependent rendering:

Meanwhile Microsoft’s open-source Trellis tech promises super fast 2D-to-3D conversion: