Monthly Archives: July 2026

From Nano Banana to Beast Mode

AI/ML, Illustration, Nano Banana

As I walked the dogs past our new camper van a few weeks back, Pinterest happened to send me some vintage World War II aircraft art. I thought, heh, how nuts would it be to give the van shark teeth? I snapped a quick pic of the van, popped it plus one of the shark mouth designs into the Gemini app, and had Nano Banana mock up the possibilities:

This helped me bring the inimitable & indulgent Margot onboard with the idea, and soon enough I found myself in Illustrator, recreating the art the old fashioned, point-by-point way. Upon seeing the design, the local graphics shop advised me on some needed mods, so I hopped into Photoshop to oblige, invoking a dash of Generative Fill. Check out the (very real!) results.

To me this is just how AI ought to work—not as a creative replacement, but rather as an accelerant that lets us try more things & communicate ideas better.

Flow to the Upside Down

AI/ML, Google Flow, Google Omni

Fill your head with sweet 80’s synth chords, using Google Omni Flash to reimagine suburbia as a sci-fi dreamscape:

Thank you, Dr. Cerf

Miscellaneous

“Correlation is not causation,” and I hope that Internet pioneer Vint Cerf announcing his retirement from Google has nothing to do with my return. 😉

The news brings to mind a couple of fun memories:

“Taste” sucks

Idle Philosophizing

I’ve been trying to put my finger on exactly why the AI-related buzzword du jour gets under my skin. “Taste” is anything but new, yet I can’t stand the way it gets bandied about, like some mystic shibboleth, by grasping little growth hackers.

I’ll have more to say about this (aren’t you lucky!), but I’m reminded of my all-time favorite Steve Jobs clip. He gets at “taste” being a function of work, curiosity, and hard-won love of the excellent.

If a YouTube clip could be worn out like vinyl, I’d have done so long ago, given how many times I played it for teammates at Microsoft. Steve was right 30 years ago & he’s right now. Just buying some Aeron chairs ain’t gonna flip that switch…

Joy-scrolling 3D

3D, AI/ML

I love the art direction on this site, where the pace of animation is controlled by your scrolling:

Meanwhile the “Scroll World” Claude skill promises to interview you (!), then whip something up with the help of generative text-to-video tech (gotta get Omni in there!).

 

How it works is intriguing enough to quote at length from the GitHub page:

—–

It leans on Higgsfield for the art: cohesive isometric diorama scenes (GPT Image 2 — via Higgsfield, or the Codex CLI on a ChatGPT subscription) and the camera flights themselves (Seedance or Kling image-to-video — only models that can frame-lock a seam), scrubbed by scroll position — the same technique behind Apple’s scroll-through product pages. The camera genuinely moves; scroll only drives time. It’s framework-agnostic: you get the Higgsfield pipeline, the prompt templates, and a portable vanilla-JS scrub engine that drops into plain HTML, Next.js, Vue, or a Python-served page — nothing assumes a stack.

When invoked, the skill:

  1. Interviews you — the subject/industry + pitch, a brand kit (import from a URL, hand it over, or have it proposed), art direction, the ordered scenes the camera visits, whether you want the mobile version (a second chain rendered natively in 9:16 portrait — composed for phones, not a crop of the landscape film), and the budget — render tiers and stills source shown with estimated credit costs, approved before anything generates.
  2. Generates the assets — one still per scene, one “dive-in” camera clip per scene, and the connector clips that join consecutive scenes, generated from the actual rendered frames of their neighbours so every seam is frame-identical. Mobile opt-in renders a parallel portrait chain the same way, frame-locked against its own 9:16 renders.
  3. Wires it up — a config-driven scroll engine that plays the whole chain as one flight, serving the portrait clips and posters automatically on phones.

Keeping It Real: How I wished we could’ve made Photoshop feel

Design, User Interface

jnack.com/BlowingYourMindClearOutYourAss—that’s the URL I picked, back circa 2005 (when men were men & we had to self-host all our videos!), to express my admiration for Nelson Chu’s Moxi watercolor app. Harnessing graphics processors to create realtime natural-media simulations has been a passion of his for the better part of 30 years.

We hosted Nelson as a summer intern on Photoshop, but we could never quite manage to marry this tech to the, ah, somewhat vintage underpinnings of PS. Later he launched a refined version as Expresii:


It remains, to my eye, just incredibly cool—never mind that its target audience will seemingly forever be limited to those with serious painting ambitions and abilities.

I thought of all this upon seeing the procedural Damage Control. Here too I dreamed of offering godlike command of realistic materials & motion:

All this stuff has remained so niche, but I think a day is coming—very soon, I hope—when realtime world models (think Google Genie, which is the realtime 3D cousin of our new Gemini Omni) make this kind of grounded manipulation absolutely ubiquitous. We shall see!

Quick tutorial: Create with Omni inside Adobe Boards

AI/ML, Google Omni

Companies are imaginary. Creativity is real.

I love seeing my old friend & Adobe design partner Dave (mentioned innumerable times here over the years) showing off the expressive power of the collaboration so many Adobe & Google friends have been nurturing over the last several months. We have the first fruits showing now in Firefly Boards (try it now!), and there’s so much more we have in mind.

I’ve never cared much about the company name in my email address. All that stuff is always just a means to an end—a way of aligning incentives so we can help people make the world more fun and beautiful. I feel so grateful to have found a spot to stand among friends & continue this work—and we’re just getting started.

“All of this has happened before, and all of this will happen again”

AI/ML, Nano Banana

…So says my friend Chris Perry, who welcomed me to Google 12 years ago and who was essential in shipping the first feature I worked on there—face painting for the 2014 World Cup:

Smash cut to 2026: Chris has just left Big G to start his own company, but we’re still inviting people to paint their faces—and to do so much more—for the World Cup. Opening Gemini today, I saw this smorgasbord of Nano Banana-powered templates:

Finn & Henry (pictured up top) are now far too old and, critically, too cool to abide my applying any patriotic AI to them—but you can give it a try with pics of you & yours. 🙂

“Loser,” baby

Design

“When it’s great, there’s no debate”—ah, what a crock of crap that is.

In this fun behind-the-scenes video, Beck explains how his still-bizarre iconic rap/sitar combo came to be, and how it sat nearly forgotten for years (including by him) before having its breakout moment.

Talk about conviction & commitment from the ol’ perdedor.

Remix your photos using Gemini Omni, right inside Google Photos

AI/ML, Google Omni

The fun has begun:

Located in the Create tab — your central hub for creativity in Google Photos — Video Remix helps you make inspired content in seconds. Instantly apply cinematic relighting to spruce up a dark clip, swap out a plain background for something fun, or add artistic treatments, such as watercolor, raw sketchbook and oil painting effects.

Video Remix starts rolling out today to eligible Google AI Plus, Pro and Ultra subscribers in select countries.

Use Google video models right within Premiere Pro!

AI/ML, Google Omni, Google Veo

I love seeing a plan come together. 😀 Check out this new Premiere Pro integration with Google Veo:

The Google Omni Flash model isn’t there yet, but we’ll work on connecting the dots. As Mark Twain might’ve said, “Some Adobe menus are so long, they have perspective,” and I’d love to get back to contributing to that phenomenon. 😉

In the meantime, my old partner Dave Werner is cooking with Omni inside Firefly Boards—where you can try it yourself:

[Via Bill Hensler]

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: