How to Combine Photos: 3 Ways That Look Real

Three completely different jobs share the phrase, and picking the wrong one is why so many attempts go badly. Decide which of these you want before you open anything.

• **A layout.** Both photos stay whole and sit side by side, in a grid or a strip. Nothing is invented, nothing can go wrong technically, and the only craft is in the alignment.
• **A blend.** The two images overlap and share the same pixels, as in a double exposure. This is a graphic effect rather than a claim about reality.
• **A composite.** Subjects from separate photographs end up in one scene, as if they had been photographed together. This is the hard one and the only one that has to survive being believed.

It also changes what honesty requires of you. Nobody needs a disclaimer on a grid of holiday photos, and nobody is deceived by a double exposure. A composite is different in kind, because it produces an image of a moment that did not happen, and the better the tool gets the more that distinction matters.

The failure modes differ too. A bad layout looks untidy. A bad blend looks muddy. A bad composite looks like a lie, which is a bigger problem when the picture is going to a family group chat that will study it closely.

The simplest job and the one people overthink. Border style is almost irrelevant. What decides whether a pair reads well is geometry.

1. 1

   Crop both to the same ratio first

   Two frames of different shapes forced into one row will be scaled unevenly, and one subject ends up bigger than the other for no reason.

2. 2

   Align the horizontals

   If both photos have a horizon, a table edge or a shoulder line, get those at the same height. That single adjustment does more than any gutter width.

3. 3

   Match brightness before you match colour

   A dark frame next to a bright one reads as a mistake even when the colours agree. Equalise exposure, then worry about warmth.

4. 4

   Keep the gutter consistent

   Equal spacing everywhere, including the outer margin. Uneven white space is the thing the eye catches first and the thing nobody can name.

How to put two pictures together is a layout question, and for a layout you do not need generative anything. A phone gallery, a slide, or any collage tool will do it, and the result is honest by construction because both originals stay intact. For layouts beyond a pair, our photo collage ideas post lists options by occasion.

A multiple exposure (2026) was originally one frame of film exposed twice, and the digital version imitates the same behaviour: bright areas of one image survive into the dark areas of the other.

Which means the choice of base image decides everything. A high-contrast silhouette works because it has genuinely dark regions for the second image to occupy. A flat, evenly lit portrait does not, and no amount of adjustment rescues it.

• **Pick a base with real blacks.** A profile against a bright sky, a body shape against a white wall.
• **Pick a partner with structure, not detail.** Tree branches, a skyline, a mountain ridge. Faces inside faces almost always read as a mess.
• **Decide where the second image should be legible.** State it. "Trees visible inside the hair and shoulders, face left clear" produces a usable frame; "blend these" produces grey.
• **Desaturate one of the two.** Two full-colour images fighting for the same pixels is the most common reason a blend looks muddy.

When you blend two photos this way you are making a graphic, not a record, so accuracy stops mattering and composition takes over. That is a relief after the composite section below, where the opposite is true. An instant-print frame is another way to present two photos as one graphic, and our Polaroid filter guide covers the correct border and dimensions.

The composite. Two people photographed separately, standing together as if they were not. Four things have to agree, and the tool will get two of them right on its own.

Here is the job running in the image combiner, which takes two uploads rather than one: a Subject box and a Scene box. Both studio portraits were shot separately against plain backgrounds.

Run one put them in the same place with the same light and got the scale wrong, which is the classic result. Heads at different heights, feet on different ground lines. Nothing about the instruction was incorrect; it simply said nothing about size.

Same coastal path, but put their heads at the same height and their feet on the same ground line so the scale matches; same faces and clothes.

Once the scale is right, four physical properties are left, and all four are cheap to specify and expensive to ignore.

None of that is model-specific advice, but it is worth knowing what you are instructing. The combiner runs on GPT Image 2, the same engine as the background and repair tools, which is why an instruction that works in one of them tends to work in the others. Light direction, camera height and contact shadow are the vocabulary that carries furthest. The repair side is covered in our guide on how to restore old photos, from scanning the print to fixing fading and tears.

The flip test is still the fastest check there is. Mirror the finished image and look at it fresh: your eye stops reading faces and starts reading shapes, and a light-direction error that was invisible thirty seconds ago becomes obvious.

A composite fails on physics, not on cutting. Get the light and the ground line right and nobody inspects the edges.

— Maya Chen

For layouts, the phone is genuinely the best tool: the photos are already there and the gallery app will build a grid in seconds. For composites it is a weaker choice, and the reason is output size rather than capability.

People search for how to combine two photos on iphone and get pointed at the Photos app, which is right for the layout case and silent about the rest. The composite case has no built-in answer on either platform, which is why browser tools keep turning up in the results.

For layouts, easily and everywhere. For composites, the free options thin out fast. Everything below came from the supplier's own page on 7 September 2026, and where a supplier publishes nothing this says so.

A composite usually needs two runs rather than one, as the section above shows, so budget accordingly. Credits in the image generator are shared across tools, which is what makes the run-then-correct pattern affordable on a new account.

Five tells, in the order people notice them. None is about the join.

• **Two light sources.** Shadows disagreeing about where the sun is. The fastest and most damning error.
• **Wrong scale.** A person who is subtly too large for the doorway behind them. This is what run one above got wrong.
• **Floating feet.** No contact shadow, so nobody is standing on the ground.
• **Grain mismatch.** One figure noisy and one clean, which happens when the two sources were shot at different ISO settings.
• **A shared edge that is too clean.** Real photographs have a little softness where objects meet. A perfectly crisp boundary between a person and the hillside behind them reads as pasted.

The honest coda: if the picture will be presented as a record of a moment that happened, say that it is a composite. Photo merger tools are good enough now that the picture will not announce itself, and a one-line caption costs nothing next to the awkwardness of being asked later.

That is the whole discipline of learning how to combine photos well. Pick the right one of the three jobs, spend your effort on physics rather than on edges, and be straightforward about which pictures are records and which are constructions.