Six mismatches, and the eye finds them all

Wrong before it is noticed

Show somebody a bad composite and they will say it looks fake before they can say why. That is not vagueness; it is how visual processing works. Consistency of light, perspective and focus is computed early and automatically, long before anything reaches conscious attention, and an inconsistency registers as wrongness with no accompanying explanation.

This is useful, because it means the fix is never “make it look better”. It is a checklist of six specific things, each of which can be measured and corrected.

The six

1. Perspective. Was the camera at the same height and the same distance relative to the subject in both photographs? A product shot from slightly above, dropped onto a table shot from eye level, sits wrong no matter how good the cut-out is. This is geometry and it is the next lesson.

2. Scale. Is the element the right size relative to everything else on its ground plane? Scale errors are surprisingly hard to see directly and obvious via shadows and contact points.

3. Light direction. Which way do the shadows fall on the element, and which way do they fall in the plate? A subject lit from the left placed in a scene lit from the right is the commonest failure and the easiest to check: look at the noses, the shadows under objects, and the shaded side of anything round.

4. Light quality and colour. Hard or soft, warm or cool. A subject shot with a small flash has hard-edged shadows with sharp boundaries; an overcast street has shadows so soft they are barely shadows. Mixing them fails even when the direction agrees.

5. Edge quality. Does the element’s outline have the same softness as everything else at that distance? A razor-sharp cut-out in a photograph shot at f/2 announces itself immediately, and so does a soft element in a deep-focus scene.

6. Grain, noise and texture. A clean studio element in a grainy night scene reads as pasted, because the surface texture of the two images differs. So does a JPEG-compressed element in a clean raw plate.

There is a seventh that is not about the element at all: occlusion and contact. Does the object touch anything? Does anything pass in front of it? An element floating with nothing overlapping it looks pasted even when the six are right, because real objects in a scene are partly hidden by other objects.

The order to work in

Do them in the order that avoids redoing work.

1. Geometry first: perspective, scale, position. Everything else depends on where the thing is.
2. Colour and exposure, roughly, so you are judging the rest against something plausible.
3. Light direction and quality, by relighting the element.
4. Shadow and contact, which fixes both the floating problem and most remaining scale ambiguity.
5. Edge quality, matched to the depth of field.
6. Grain and noise, last, applied over the whole composite rather than to the element.

That last point matters: adding grain to the element alone leaves two grain patterns. Adding it to a flattened copy of the whole composite unifies them.

Choosing the plate first

The single largest determinant of whether a composite will work is whether the two photographs were compatible to begin with.

When you can choose, choose:

• Overcast or open shade for both, because soft light has no hard shadows to match.
• Similar focal length, because that sets the perspective relationship.
• Similar camera height.
• Neutral light colour in both, which you can then grade together.

When you cannot choose (a client supplies two photographs) the honest assessment comes first. A subject shot with direct sun from the right cannot go into a scene with direct sun from the left. Reversing the element horizontally sometimes solves it, and then any text or asymmetric detail on the subject is mirrored. Say what is achievable before the work starts.

The checking method

When the composite is done, three checks:

Flip it horizontally. Your eye has habituated to the image. Mirrored, it is unfamiliar again and the errors reappear. Flip back afterwards.

Look at a small version. Shrink it to 300 pixels. Errors of tone, contrast and light direction survive; errors of edge detail disappear. If it reads as wrong at 300 pixels, the problem is structural rather than in the mask.

Blur it heavily. A 20-pixel Gaussian on a flattened copy leaves only the broad tonal and colour structure. If the element still stands out as a differently-toned lump, the exposure or colour match is wrong regardless of how good the edges are.

All three take seconds, and all three find different classes of error.

The honest note on effort

A convincing composite is slow. A simple element into a simple plate is an hour; a person into a complex scene with matched light is half a day. Nearly all of that time is items 3 to 6 on the list, not the cut-out, which is the part beginners expect to be hard.

If the schedule does not allow it, the better decision is usually a composite that is visibly a composite: a cut-out on a flat or graphic background, which is honest design rather than a failed illusion. The failure mode to avoid is the half-finished attempt at realism, which reads as incompetence where a graphic treatment would have read as a choice.