Hey all, and happy Fridizzle.

It's cold as hell in NYC. The kind of cold where you question every life decision that led you not to live somewhere with a palm tree.

I went to college in Maine, though, so I can handle it.

Lol jk. Four years in Lewiston did absolutely nothing to prepare me for anything except drinking Shipyard and pretending I like skiing on ice. I am not built for this, and have clearly gotten soft.

Got drinks last night with Colin from Taikun and Tris from Webtopia. We talked about being business owners right now, the post-BFCM hangover (it's real, everyone's exhausted and over it), and whether any of us actually remember what a "weekend" is.

Also, post-Thanksgiving update nobody asked for: baby James has discovered mashed potatoes. Not just discovered. Obsessed. Butter, butter, and more butter. The kid sees a potato and loses his entire mind. Who would've thought our 10-month-old would become a carb enthusiast before he could walk?

Maddie has been adding egg yolks to the mashed potatoes for some extra protein. Kid is about to get swole. 💪

Anyway. I promised you a deeper dive into Nano Banana 2, and I actually delivered for once. Did a whole training call this week with Foxwell Founders, walking through everything. Here's the breakdown I wish I had two weeks ago:

I used to say "always start in Midjourney" when creating statics. Generate your base image there, then bring it into Nano Banana for product insertion and edits.

I'm changing my stance.

You can now create entire statics directly in Nano Banana 2. Start to finish. The image generation, the product placement, the copy—all of it. We've been testing this with clients and it's kind of absurd how well it works.

Let me walk you through one we did for Easy Man.

Easy Man is a THC/CBD beverage brand with a late-90s, Ferris Bueller's Day Off kind of vibe. Here's how we built a full static from scratch:

Step 1: Clean the product shot

Their can has THC and CBD plastered all over it. Can't run that on Meta. So first prompt was basically: "remove anything THC or CBD on the can while keeping everything else."

Prompt: Using the provided can image, remove all references to THC, CBD, dosing, or cannabis-related symbols. This includes removing or replacing text such as “THC”, “CBD”, “2mg THC”, “2mg CBD”, “THC-INFUSED”, and the small THC warning diamond icon. Preserve all other design elements exactly as they are: colors, typography, layout, shadows, highlights, texture, metallic reflections, and overall style. Do not alter the EASY MAN branding, the word “Lite”, the central badge, or any decorative elements. Where text or icons are removed, fill the space with clean, natural-looking background graphics that match the original can design so the edit is seamless and unnoticeable. Do not add any new claims or wording — only remove THC/CBD references.

One generation. Done. And here's the crazy part: it didn't just delete the text and leave weird gaps. It actually adjusted everything so the can still made sense. The layout, the spacing, all of it. I don't know how. I don't ask questions anymore.

Step 2: Generate the scene

This is where it gets fun. I went to ChatGPT with their brand vibe (copied and pasted from their website) and asked for scene ideas for someone holding the can. Then I added my secret weapon prompt:

BOOKMARK THIS: "A flash photography snapshot taken on a disposable camera in 1998, grainy, chaotic, red eye effect, harsh shadows, motion blur, film grain. The composition is slightly tilted and messy."

Why? Because AI-generated people still look slightly off. The hands, the skin, the eyes, something's always a little uncanny. But add that disposable camera aesthetic and suddenly it reads as intentional. The imperfections become stylistic. “Human.” “Real.”

We generated a guy at a bar with the Easy Man can and it looked like a photo your friend took at 2am in 1999. Perfect.

Final Prompt: Use the uploaded can image and place it naturally in someone’s hand. Create a candid 1998 disposable-camera flash photo where part of the photographer’s finger is covering the lens. Capture:

harsh direct flash

red-eye on the person

film grain, dust, slight motion blur

blown-out highlights + hard shadows

a tilted, messy, unposed composition

Keep the can intact and unchanged. No added text.

Step 3: Add copy with their actual font

Identified their brand font (Century Gothic-ish), grabbed their hex codes, and prompted Nano Banana to add the headline.

The font matching is genuinely unhinged. It just... figures it out.

Now, granted, this isn’t the sexiest output, but hell, it does the trick pretty well.

Also, I wanted to add some USPs, so I did. I had Claude scan their website for the best USPs, and then had it added to the creative.

Listen, again, I acknowledge this isn’t the sexiest ad, but I am showing this for illustrative purposes. And honestly, I bet it fares better than some of the more polished shit we put out.

Prompt for just copy: Use the uploaded image as the base. Add the headline: “Proof you don’t need alcohol to have a night.” Place the text in a clean, readable position near the top or bottom of the image without covering the can or the main subject. Formatting requirements: Font: Century Gothic Font color: #fac235 (yellow) Add a subtle dark drop shadow or soft outline to ensure the yellow text is readable against the background. Keep the overall 1998 disposable-camera vibe intact. Do not alter the underlying image aside from adding the text.

Prompt for adding USPs: Use the uploaded image (person holding the can) as base. Below (or to the side of) the main subject, add a row (or cluster) of four simple icons + text labels representing these USPs: Gluten-free Low Calories 3 MG–5 MG THC Onset in < 30 min Requirements: Icons should be minimalist, clean line or simple glyph style (not photo-realistic), matching the overall gritty flash-photo vibe. Text for each USP should be clear and legible, placed near its icon. Icon/text color should complement the image — use a color that stands out but fits the aesthetic (e.g. off-white or soft yellow). Include subtle shadow or outline if needed for readability. Layout must feel balanced and not overpower the main image; USPs should support, not distract. Maintain the 1998 disposable–camera flash look for the overall ad (grain, flash contrast, tilt/imperfections). No other copy or branding should be added.

Step 4: Resize without losing your mind

Had a 1:1, needed a 9:16 for Stories.

Now: one prompt, change the aspect ratio, Nano Banana extends the image while maintaining integrity. Safe zones looked fine. Done.

Universal Prompt: Take the provided image and convert it to a 9:16 vertical aspect ratio. Preserve the full integrity of the original scene, product, colors, shadows, and composition. Do not crop, distort, or alter any existing objects. Extend the canvas naturally above, below, or to the sides as needed to achieve 9:16. Match the existing lighting, shadows, texture, and depth of field so the extended areas blend seamlessly with the real photo. No added objects, no retouching — just a natural, realistic extension of the original environment. Keep the sticky note text exactly as shown.

Total time: maybe 15 minutes. Total cost: probably 60 cents.

Here's what most people actually want to do: see an ad that's working for another brand and make a version for their own client, with their own fonts, colors, brand voice, etc.

So we grabbed a Starbucks ad, threw it into Claude, and said, "recreate this for Easy Man with their brand voice, fonts, colors etc."

Prompt: Use the Starbucks ad layout as the base reference, but convert it into an Easy Man Lite ad. Replace the Starbucks drink with the uploaded Easy Man Lite can, matching the original drink’s placement, scale, lighting, angle, and shadow.

Update the copy to:

Headline: “Find your easy side”

Subhead (under the can): “Easy Man Lite”

Now restyle the entire ad to match Easy Man’s brand colors, replacing Starbucks green with Easy Man’s palette:

Deep blue (#1e3264 or match the can’s blue)

Bright red accents (#d7373f or match the can’s red trim)

Silver/white background elements

Black or dark navy for text when needed

Design guidelines:

Maintain the clean, minimal, modern composition of the Starbucks layout.

Use smooth, bold color blocking inspired by Easy Man’s branding (not gradients).

Keep typography simple and clean, similar in feel to Starbucks’ sans-serif styling.

Ensure the can remains crisp and high-resolution.

No additional icons or elements unless needed for balance.

The final ad should look like an official Easy Man brand asset created in the style of the Starbucks reference — but fully recolored and rebranded.

It matched the layout. It matched the vibe. It swapped in Easy Man's brand elements. Was it the prettiest static I've ever seen? No. But it was damn good, and I am confident most brand owners would be happy with it.

Is this ethical? I think that’s a larger discussion as a whole but I mean, you're not stealing creative, you're using a reference for structure. Designers have done this forever. Now you’re the designer.

Nano Banana 2 is 15 cents per image (vs 4 cents for the original). Yes, it's more expensive. Yes, it's worth it.

But you can also do it directly in Gemini fo freeeee.

The product fidelity is insane. The font matching is insane. The hands actually look like hands now. Do you understand how big a deal that is? We spent two years making excuses for AI hands. That era is over.

I'm not saying fire your designer. 😅

I'm saying: when you need to go from concept to static in 15 minutes, when you need to test 10 variations without briefing anyone, when it's Saturday, and something needs to change...

This is how you survive.

More soon.

Will