Skip to main content

About

Who is behind these tools

A short, checkable account of what this site is, how it decides what to build, and what it refuses to build. If you are about to upload your photographs to something, this is the page you should be able to read first.

What this is

The AI Image Agent is a small set of image tools that share one account and one pool of credits. It exists because the alternative — a browser with nine tabs open, each a different single-purpose site, each wanting its own sign-up — is how most people currently do this work.

The entry point is the AI image agent: describe the finished image and inspect the plan before anything runs. Around it sit the direct operations — image to prompt, remove background from image, upscale image, restore a scanned print, studio scenes for a product shot, and anime image generator — for people who already know the exact operation they need.

How we decide what a tool does

Every tool page starts from watching what the existing options make people do, not from a feature list. The background remover shows a chequerboard preview because that is the only way to judge whether an edge was cut cleanly. The upscaler has separate models for photographs and line art because using one for both is the single most common reason an upscale looks wrong.

When we cannot do something well, the page says so on the page rather than in the fine print. The photo enhancer warns you next to the switch that face repair can make a low-resolution face stop looking like the person — before you use it, not after.

How the models are chosen

No single image model is best at everything, so we route per operation rather than committing to one provider. A model is picked on three things: whether it holds the parts of the image you asked to keep unchanged, what one run actually costs, and whether it can return the exact output size a marketplace requires.

The routing is ours, the models are not: we use commercial APIs and say so. What that means for your files is set out in full on the privacy policy, including which processors are involved and how long anything is kept.

What we will not build

This is a deliberate list, not an oversight. Several of these are the highest-traffic queries in this category, and skipping them costs us real volume.

  • Face swaps and deepfakes of real people.
  • Sexual or suggestive imagery, and anything involving minors in such a context.
  • Generating identifiable celebrities.
  • Watermark removal.
  • Anything designed to pass an AI image off as a photograph of a real event.

How we write about tools

The guides are written by the same people building the tools, and each one names the tool it is about. We do not publish comparison content about categories we have not used, we do not accept payment for placement in any list, and when a guide is materially wrong we correct it and change the date on it rather than quietly rewriting history.

Sample and proof images are our own output. Where we do not yet have real output — the reproduction proof on the image to prompt page, for instance — the section stays empty and says so. A borrowed sample would answer the exact question the section exists to answer, dishonestly.

How we make money

Free tiers are free because they cost us little and prove the tool works. The generation and editing tools consume credits, which come with a subscription. Nothing runs before you have seen what it will cost, failed and blocked runs are refunded automatically, and credit packs you paid for do not expire. The full breakdown, including what we do not charge for, is on the pricing page.