Clean product listing images, generated by your AI agent

The photos your supplier sends are costing you sales

Most dropship suppliers hand you the same tired photos everyone else is using. Harsh lighting, a cluttered table, a watermark in the corner, and a background that changes from shot to shot. Shoppers scroll straight past listings that look like this because nothing signals that the product is worth trusting.

Fixing it used to mean a photographer, a lightbox, and a day of editing per product. For a catalog of fifty items that is not realistic, so most sellers upload whatever the supplier gave them and hope.

There is a third option now. You describe the clean image you want, and an AI agent generates it from a reference photo. No studio, no Photoshop, no waiting.

What a clean listing image actually is

Clean does not mean fancy. It means the product is the only thing the eye lands on: a pure white or soft neutral background, even lighting with no hard shadows, the item centered and filling most of the frame, and no text or logos baked into the picture.

Consistency matters as much as any single shot. When every image in your catalog uses the same background and framing, your store looks like a brand instead of a bin of random products.

That consistency is exactly what an agent is good at, because you can give it the same instructions for every product and get the same look back every time.

How the generation works

Figgshield is an MCP connector. Your AI agent talks to it at https://api.figgshield.ai/mcp and asks it to make an image. It works with Claude on desktop and in Claude Code, with Cursor, with Windsurf, with ChatGPT in Developer Mode on a paid plan, and with any other MCP agent.

The image model is Nano Banana 2 from Google. You send a reference photo of the product and a short description of the background and framing you want, and it returns a finished image. A 1K or 2K image costs 8 credits, and a 4K image costs 16.

Your reference photo is used in memory to guide the result and is never stored. The finished file sits on a private link for seven days and is then deleted, so you download it and add it to your listing.

A prompt recipe you can reuse

The trick to a consistent catalog is writing the instruction once and reusing it for every product. Ask your agent to generate the image and keep this structure:

1. The product, described plainly, for example "a matte black stainless steel water bottle"
2. The background, for example "pure white studio background, no props"
3. The lighting, for example "soft even lighting, no harsh shadows"
4. The framing, for example "centered, filling about 85 percent of the frame"
5. The output, for example "square, 2K resolution"

Turn the winning image into a short clip

A short video on a listing or an ad often holds attention better than a still. Once you have an image you like, you can animate it. Pass the id of the image you already generated and the agent starts a video from it, so the product stays exactly as it looked in the still.

The video models are Kling 3 and a faster Kling 3 Turbo tier. A five-second clip costs 35 credits, which is enough for a slow rotation, a gentle push in, or a subtle light sweep across the product.

The same credit balance covers both images and video, so you do not manage two accounts or two budgets.

What it costs and what happens to your files

Figgshield charges at cost. The model providers set the price and there is no markup added on top. The only money the service makes is on credits that people buy and never spend, which means the incentive is to price the generation itself fairly.

You start with a $5 seven-day Pro trial that includes 150 credits. There is no free tier. If you cancel before day seven you pay nothing more. At 8 credits per image, that trial is enough to redo the photos on a small catalog and see whether cleaner images move your numbers.

Every generated file lives on its private link for seven days and is then removed, and any reference photo you send is used only in memory. See how credits work if you want to plan a larger batch.