7 min read
About three and a half cents. That is roughly what Google lists for one standard resolution image from its new Nano Banana 2.1 model, according to the Gemini API pricing page. A single product shot from a hired photographer can cost more than a thousand of those. For a solo seller with forty listings, a seasonal catalog and a social feed that never stops asking for more pictures, that gap is the whole story of this month.
Three releases landed in the first half of October that change what a one-person shop can do with AI product photography: Google’s Nano Banana 2.1, Black Forest Labs’ FLUX 3 Image, and a free, open source photo editor called PhotoCraft. None of them is a magic button. Together they cover the three jobs most small sellers actually need: making a product image, placing it in a scene you control, and polishing the final file without paying for a subscription.
Nano Banana 2.1: the cheap, fast workhorse
Google released Nano Banana 2.1 on October 6 as an update to Nano Banana 2, its image generation and editing model built on Gemini’s Flash image engine. Think of it as the everyday model in Google’s lineup, sitting next to the more expensive Nano Banana Pro rather than replacing it.
What improved, according to Google’s documentation and early coverage:
- Better prompt following. If you ask for a mug on a marble counter with morning light from the left, you are more likely to get exactly that.
- Cleaner text in images. Useful for a simple promo tile, though you should still check every letter.
- Multi-image fusion with up to 14 reference images. In plain English: you can feed it several photos of your real product from different angles so the result looks like your item, not a lookalike.
- Output at 1K, 2K and 4K, with a fix for the odd tiling artifacts older versions produced on wide banner shapes.
It is available through Google AI Studio and Flow, so you do not need to write code to try it. The listed price is about $0.0336 per standard image and about $0.0756 per 4K image, with batch processing halving those figures. One practical note: trackers report the older Nano Banana 2 model is scheduled to shut down on October 29, so if a plugin or app you use relies on it, expect a switch soon.
Why it matters for a solopreneur: this is the model to reach for when you need volume. Lifestyle variations of one product, seasonal backgrounds for the holidays, or ten versions of an ad image to test. At these prices, experimenting costs less than your coffee.
FLUX 3 Image: when you need the picture laid out exactly
Black Forest Labs, the team behind the popular FLUX models, released FLUX 3 Image as the image part of its new FLUX 3 family. Its headline feature is something small sellers have wanted for a long time: bounding box layouts.
Instead of hoping the AI puts your product in the right spot, you draw boxes on a grid and describe what goes in each one. Product bottle in the lower left. Sprig of rosemary on the right. Empty space across the top for your headline. FLUX 3 renders each element inside its box. If you have ever fought an image tool that kept cropping your product or covering the spot where your sale text needed to go, you will understand why this matters.
Other features worth knowing:
- Up to 10 reference images in one composition, so your real product and your brand props can appear together.
- Native 2K and 4K output, sharp enough for a website hero image or a printed flyer.
- Precise region editing, which lets you change one area (swap the background color, remove a stray object) while leaving the rest untouched, and make several edits at once.
There is a browser playground to try it, plus an API for apps that build on it. Black Forest Labs does not publish a free tier on the model page, so check current pricing before you plan a big batch.
Why it matters for a solopreneur: FLUX 3 is the model for designed images rather than just pretty ones. Banner ads, email headers, marketplace main images with strict layout rules. It is the closest thing yet to giving an AI a sketch and getting back a finished layout.
PhotoCraft: a free editor for the last ten percent
The third release is not an AI generator at all, and that is exactly why it belongs here. PhotoCraft is a free, open source photo editor from the ArtCraft team. Open source means the code is public and the app costs nothing; it is released under the permissive MIT or Apache 2.0 licenses.
It covers the tools most sellers use Photoshop for: layers, masks, adjustment layers like Curves and Levels, drop shadows, text, and layered PSD files. It runs on Windows, Mac and Linux, and even in a web browser. Its subject selection tools run locally on your computer, with no cloud account required. It also includes a command line mode and an MCP server, which is a standard connector that lets AI assistants drive the app, so in the future you could ask an assistant to resize and sharpen a whole folder of product shots.
Two honest caveats. The project describes itself as early alpha and says plainly that it is not yet a Photoshop replacement for daily professional work. And it has no generative AI features of its own. Treat it as a free finishing tool: crop, color correct, add a soft shadow under the product generated elsewhere, export.
Why it matters for a solopreneur: AI images almost always need a final touch. Paying a monthly fee only for that last step never made much sense for a small shop. PhotoCraft is worth installing and testing on a few images now, even if you keep your current editor for the busy season.
A simple workflow that uses all three
Here is how a solo seller could combine these tools this week, with no design background:
- Shoot your real product on a plain background with your phone, from three or four angles. These become your reference images, and they keep the AI honest about what you actually sell.
- Generate lifestyle variations with Nano Banana 2.1 in Google AI Studio. Ask for your product in two or three seasonal scenes. At a few cents each, make plenty and keep the best.
- Build your layout images with FLUX 3 when placement matters, such as a holiday banner with clear space for your offer.
- Finish in PhotoCraft: crop to each platform’s size, fix color so the product matches reality, and export.
- Compare against the real item before publishing. Color, size and texture must match what the customer receives, or you are buying yourself returns and bad reviews.
That last step is not optional. Marketplaces and shoppers punish images that misrepresent a product, and an AI scene that makes a small candle look like a large one is a refund waiting to happen.
What to watch before you go all in
The excitement is real, but a few things deserve attention:
- Prices and access change fast. The figures above are list prices at launch. Confirm them on each provider’s own page before committing a budget.
- Marketplace rules. Some platforms require main product images to show the actual item on a plain background. Use AI scenes for secondary images and social posts unless the rules clearly allow otherwise.
- Disclosure and trust. Customers increasingly assume photos might be generated. Keeping at least one real, unedited photo of every product builds the kind of trust that AI scenes cannot.
If you are just starting with image tools, our walkthrough on making a flyer and product photo with ChatGPT Images 2.5 is a gentle first step, and the story of a solo jewelry maker who tripled her Etsy listings shows what a full workflow looks like in practice.
Your move this week
Pick one product that sells well but has tired photos. Run it through the five steps above, publish the best new image as a secondary photo, and compare clicks for two weeks. That small test will tell you more than any launch announcement. Which product would you try first? Tell us in the comments.
