The comparison is almost always set up like this: a retoucher's salary on one side, a supplier's per-image rate on the other. Then someone divides the salary by a guess at monthly output and the argument starts.
It starts in the wrong place. A salary is a fixed cost — it is paid in a quiet month and a busy one. A per-image rate is a variable cost — it is paid only when frames are sent. Until the fixed side has been converted into a cost per delivered image, there is nothing to compare.
This page is the conversion. It does not tell you what your in-house retouching costs, because that depends on your salaries, your city and your volume. It tells you which lines belong in the number, where the estimate usually breaks, and what the outsourced side costs at published rates.
A retoucher's salary is the largest line, and it is rarely the only one. The lines below are the ones most often left out of the first estimate.
| Cost line | Often forgotten because |
|---|---|
| Salary and employer costs | The gross salary is used rather than the loaded cost: taxes, benefits, insurance, bonuses |
| Supervision | Someone senior briefs, answers questions and sets the standard. That time belongs to the image |
| Quality checking | Either a second person checks, or the retoucher checks their own work and the error rate rises |
| Rework | Images that come back for a second pass are paid for twice and usually counted once |
| Software and hardware | Licences, calibrated displays, machines, storage, and the replacement cycle on all of it |
| Space | Desk, light control, power. Small per head, not nil |
| Recruitment and training | A retoucher is productive some weeks after they start, and the seat has to be refilled when they leave |
| Cover | Holiday, illness and notice periods still have to produce images, or the deadline moves |
None of these are arguments against an in-house team. They are simply the lines that have to be in the number before it can be set against a rate card.
The arithmetic is straightforward:
Fully loaded monthly cost of the retouching function ÷ images delivered and approved that month = your true cost per image
The numerator is tedious but honest. The denominator is where almost every estimate goes wrong, because people divide by capacity rather than by delivered images.
Capacity assumes every working hour becomes a finished frame. In practice the month also contains briefing, waiting for product or for sign-off, rework, holiday, training, meetings and the days when nothing arrives from the studio. A team that could in theory produce a given number of images will deliver fewer, and the fixed cost is spread over the smaller figure.
This is why the same team can look cheap in one month and expensive in the next without anyone doing anything differently. It is also the single biggest reason an in-house estimate and an outsourced quote disagree: one is measured against capacity, the other against work actually done.
Our rates are published rather than held behind a quote form, which makes the comparison possible without a sales call. These are per image.
| Service | Rate per image (USD) |
|---|---|
| Clipping path / background removal | $0.50 – $2 |
| Colour correction | $0.50 – $2 |
| Ghost mannequin, neck joint | $1 – $3 |
| Product retouching, e-commerce | $1 – $4 |
| Jewellery and watch retouching | $2 – $8 |
| High-end beauty and editorial | $10 – $30 |
The band exists because the work inside one category varies: a single-path trainer and a multi-path handbag with straps and holes are both clipping path, and they are not the same job.
Volume discounts are published on the same page: 10% from 1,000 images a month, 15% from 3,000, 20% from 5,000, and above 10,000 a rate agreed on a call rather than from a table. A fuller breakdown sits in the cost per image guide and the price index.
There is no universal break-even volume, and anyone who quotes one is guessing at your salaries. What can be said is which way each variable pushes.
| If this is true of your volume | It favours | Why |
|---|---|---|
| Steady, every working day, all year | In-house | The fixed cost is spread thinly and never idles |
| Arrives in peaks — drops, seasons, campaigns | A supplier | Quiet months are not paid for, and peaks do not need hiring |
| Growing faster than you can hire | A supplier | Capacity is available before the recruitment finishes |
| Small and highly creative | In-house | Art direction in the room beats a brief sent abroad |
| Large and repetitive | A supplier | Repetition at volume is where a production floor is built to win |
| Confidential beyond what a contract can cover | In-house | Some material genuinely cannot leave the building |
Most studios find they are not choosing one. The steady baseline stays in-house and the peak goes out, which is why white-label overflow is the common arrangement rather than wholesale replacement.
It is worth saying plainly, because a page written by a studio has an obvious incentive not to.
Keep it in-house when the retouching is creative rather than repetitive and needs someone to lean over the screen and change their mind. Keep it in-house when volume is steady enough that the team is busy every day. Keep it in-house when the images genuinely cannot leave, or when the retouching is itself the thing you sell and the skill is the business.
In those cases the fixed cost is buying something a per-image rate does not: presence, iteration and control. Paying more per image for that is not an error.
Three things usually decide the outcome, and none of them appear in either number.
A supplier whose rate is slightly higher and whose batch arrives whole, on time and matching is usually cheaper once rework and delay are counted.
An estimate built from someone else's assumptions will not end the argument inside your business. One real batch will.
We return five test images free within six hours, which is enough to judge edge quality and consistency before anything is agreed. If you would rather measure the supplier you already have, the retouching audit does that instead.
Take the fully loaded monthly cost of the retouching function and divide it by the images actually delivered and approved that month. Fully loaded means salary plus employer costs, the share of a supervisor's and a QC checker's time, software licences, hardware and its replacement cycle, desk space, recruitment and training, and cover for absence. The mistake is dividing by capacity rather than by delivered images.
They are different units. A salary is a fixed monthly cost that is paid whether or not images arrive. A per-image rate is a variable cost that is only paid when work is sent. Until the salary is converted into a cost per delivered image, the two numbers cannot be set against each other at all.
Retouch Atelier publishes its rates. Clipping path and colour correction run $0.50 to $2 an image, ghost mannequin and neck joint $1 to $3, e-commerce product retouching $1 to $4, jewellery and watches $2 to $8, and high-end beauty or editorial $10 to $30. Volume discounts are published too: 10% from 1,000 images a month, 15% from 3,000, 20% from 5,000, and a custom rate agreed on a call above 10,000.
There is no universal threshold, because it depends on how steady the volume is rather than how large it is. A team that is busy every working day of the month spreads its fixed cost thinly; the same team at half capacity doubles the cost of every image it does deliver. Steady, predictable volume favours in-house. Volume that arrives in peaks favours a supplier, because the quiet months are not paid for.
When the work is creative rather than repetitive and needs an art director in the room, when volume is steady enough to keep the team busy, when the images cannot leave the building for contractual or regulatory reasons, or when the retouching is itself the product you sell. In those cases the fixed cost buys something a per-image rate does not.
Three things that usually decide the outcome: how quickly a batch comes back, whether image 4,000 still matches image 1, and what happens when volume triples for a month. A cheaper image that arrives late, drifts across a batch or cannot scale for a campaign costs more than its rate suggests.
Take one real batch you have already produced in-house, note what it cost and how long it took end to end, then send the same frames to a supplier and compare like for like. Retouch Atelier returns five test images free within six hours, which is enough to judge edge quality and consistency before any commitment.
We'll tell you which technique it needs — and show you the edge quality.