Books made from your child's photo: how to tell the three kinds apart

This is the largest category by far and the one you will meet first. Shutterfly is the obvious example: you upload your own pictures and personalise "each page with layouts, text, backgrounds, and embellishments", and what arrives is your real photographs, printed well.

It is an excellent product and it is not a storybook. There is no plot and your child is not a character - it is an album with a nice cover. If what you wanted was a record of the year, stop here, because this kind is cheaper and better at that job than anything below.

This is the one that catches people out, because it is sold with the same words as the third kind. I See Me!'s "My Very Own Name" includes "a printed message from you and a photo of the child on the first page" - but the story illustrations themselves are fixed art, and are not customised to look like your child.

So the photo is real and the child in the pictures is a stranger. That is worth knowing before it arrives, especially if the reason you were buying it was so that a child could see themselves.

A close relative of this kind has no photo at all: you build a character from menus. Wonderbly's character books let you pick "from braids to curls, and dark skin to light"; Hooray Heroes uses "a preset avatar builder"; Librio "personalizes by configuration, not photo". These are good books. They just are not made from a photograph, and no amount of configuring produces a likeness.

Here the photograph is the input to the artwork rather than something printed alongside it. You upload one picture, and the child is drawn as a character - face shape, hair and skin tone carried through - and stays recognisable from page to page.

This is what we make, and it is worth saying plainly that we are one of several. ToonyStory, Custom Heroes, TellMyTale, Childbook.ai and Magic Story all work from an uploaded photo too. Anybody in this category telling you that working from a photo is their unique idea is overstating it, ourselves included.

What separates the products inside this third kind is not whether they use a photo. It is whether the likeness holds up on faces that are not easy, and whether you can see it before you pay.

Can you see it before you pay? A free preview of real output from your own photo is the only honest test, because every company in this category shows beautiful examples on its homepage. Sample art is a claim; your own child on the cover is evidence.

Do they show the hard cases? Likeness engines fail in predictable places - deep skin tones with coiled hair, a baby with almost no hair, a heavy fringe, freckles. A company that only publishes its easy results has told you nothing. Ours are on the likeness page, failures included.

How many photos, and what happens to them afterwards? One clear photo should be enough; if a service wants a dozen, ask why. And the retention policy should be findable without emailing anyone.

Is the child in the story, or on one page of it? Ask which pages the child actually appears on. That single question separates kind two from kind three faster than any amount of marketing copy.

One clear, well-lit, forward-facing photograph beats a beautiful one every time. A gorgeous side-on shot in low light produces a weaker likeness than a plain straight-on picture taken by a window, because the illustrator - human or otherwise - can only work from the face it can actually see.

Do not use the most flattering photo. Use the one that looks most like them on an ordinary day, because that is the face the child will be looking for when they open the book.