A house look for work that is never the same twice

A florist photographs Tuesday's arrangement, and by Thursday that arrangement is on somebody's kitchen table dropping petals. There is no reshoot. Next week's bouquet uses different stems, because the market had different stems, and the week after that the whole palette has moved because the season has. Run that for a year and the feed looks like twelve different people took it.

This is the situation for anyone whose inventory is one of a kind. Florists have it most acutely, and it applies just as much to a potter firing small batches, a maker working with salvaged fabric, or anybody selling what they happened to make this week. Most product photography advice assumes a fixed item photographed once and used for two years. Very little of it survives contact with work that does not repeat.

The constant is the set, not the subject

You cannot make the product consistent. You can make everything around it consistent, and everything around it is most of the picture.

Six things are worth fixing once and then never deciding again: the surface the work sits on, the backdrop and how light or dark it is, the height of the camera relative to the work, how much of the frame the work fills, the direction the light comes from, and the crop ratio. Write them down. Better than writing them down, mark them. A strip of tape on the bench where the vase goes and a second strip on the floor where you stand removes the two variables that drift most, and it turns the whole thing into a thirty second job instead of a small creative decision you have to make while a customer waits.

Here is the part that makes this worth the discipline. Nobody browsing your work ever sees Tuesday's bouquet and Friday's bouquet side by side at full size. What they see, scrolling, is a run of images in which the ground and the light either agree or do not. Those two things are what a stranger reads as a house style, and they are the two things entirely under your control even when the product is not.

Contrast in value, not in hue

The most common failure in flower photography is pale blooms photographed against a pale ground.

Value means how light or dark something is, independently of what colour it is. Cream peonies on a white wall in soft daylight are two nearly identical values, so every petal edge is a boundary between two tones that are almost the same, the silhouette dissolves, and the arrangement becomes a bright smear with some shapes in it. The photograph is not badly exposed and the flowers are not badly arranged. There is simply nothing separating one from the other.

The test takes five seconds and works on any phone. Open the photo, turn the saturation all the way down, and look at the black and white version. If the arrangement and the background have collapsed into one shape, the picture will not read at thumbnail size and no amount of colour grading will rescue it.

The rule that follows is short. Pale blooms want a mid tone or deep ground. Deep blooms want a light one. Grounds that flatter almost everything: warm bone, soft clay, sage, dusty blue, deep charcoal. Avoid pure white, which blows out pale petals and is also what every phone photo already looks like, so it gives away nothing about you.

Two grounds cover most of a year's work. One light, one deep. Pick which one you are using by squinting at the flowers before you set up, not after you have shot.

Consistency is not sameness

There is a real tension here worth naming, because the advice above can be followed into a different failure. Twelve posts with identical framing, identical light and identical mood read as machine made, and a viewer will feel that long before they can explain it.

The distinction is that the set stays constant and the shot does not. Same ground, same light direction, same working distance, and then genuine variety in what you do inside that: the whole arrangement square on, one bloom close enough to fill half the frame, hands tying the string, the wrapped bunch on the passenger seat before delivery, a bucket of stems at the back of the shop. A viewer reads the repeated ground as identity and the changing shot type as range, which is exactly the pair of things you want them to conclude.

Get it backwards, twelve different backdrops all shot from the same distance at the same angle, and you manage to look inconsistent and monotonous at the same time.

The photograph is evidence of taste, not a listing image

For a maker with repeatable stock, a product photo says: this is the item, buy this item. For one-off work it cannot say that, because the item in the picture is already sold and, in the case of flowers, is already dying. The photograph is doing a different job. It is a sample of your judgement, offered as proof that the thing you make next week will also be good.

That changes what belongs in the caption. Name the stems. "Ranunculus, spray roses, hellebore" is more persuasive than "beautiful spring bouquet", because a buyer deciding whether you know what you are doing reads specific names as evidence and adjectives as noise. Then state the thing that is genuinely purchasable, which is a size, a price and a deadline rather than a particular set of flowers: market choice, three sizes, order by 2pm for same day, delivery within four miles.

Logistics lines are unusually persuasive in this trade and most florists bury them. Same day, order by, delivery radius, what is in season this fortnight. Set them small and put them somewhere they appear every time.

The occasion outranks the house style

Almost no other category has this property. A florist's picture is usually about a date, and the date decides the register more than any brand decision does.

The sympathy row is a hard rule rather than a preference. A cheerful promotional treatment on funeral flowers is not a style miss, it is an offensive one, and it is the single post on your account that people will remember. If you schedule anything, do not schedule that.

Do not polish it flat

Real flowers are asymmetric. A stem points somewhere unhelpful, one bloom is further open than its neighbours, a petal has a bruise. That irregularity is not a defect in the photograph, it is the evidence a buyer uses to conclude that a person arranged real flowers on a real morning.

Editing until everything is even removes the proof. So does pushing saturation, which is the fastest way to make petals look synthetic, because the eye knows the range a real petal colour lives in and notices immediately when it is exceeded. In a trade whose entire premium is freshness, looking manufactured is the expensive mistake.

The same applies if you drop a photograph of your work into a generated setting. The flowers will usually be fine, because they are your photograph. The room is where it goes wrong, and it goes wrong by being too clean to be a room. The specific tells are worth knowing before you publish one.

The ninety second routine

The reason florists end up with bad photographs is almost never taste. It is that the only moment the work exists is the gap between finishing it and handing it over, and that gap is short and busy. So it has to be a routine rather than a decision.

1. Wrapped and finished, onto the marked spot. Phone on the other mark.
2. Three frames: the whole thing square on, one from slightly above, one close enough that a single bloom fills half the frame.
3. One frame with hands in it, holding or tying. That one does more work than the other three combined and takes ten seconds.
4. All of it before the water, before the card goes in, before the van.

Ninety seconds, done in the same order every time, is the difference between a year of usable images and a year of pictures taken in a hurry in bad light on a doorstep.

Before you post

Desaturate it and check the arrangement is still a separate shape from the ground. Put it next to your last three posts and check the light comes from the same side in all four. Shrink it to thumbnail width and check it still reads as flowers rather than as a blur. Then read your own caption and ask whether it names the stems and says what a stranger could actually order.