Start with the shot count, not the shoot date
The first number you need is not "how many days in the studio" but "how many finished images have to exist on launch morning". It is derived from three lists: the products in the drop, the variants of each product, and the views each channel requires.
A modest apparel drop looks like this: 24 styles, an average of 3 colourways, and a house standard of 5 views per colourway (front, back, side, detail, on-model). That is 360 finished images. Add a flat-lay for the collection grid and a lifestyle shot per style for email and social, and you are past 400. If Amazon is also a channel, its main-image rules usually mean a separate pure-white hero per variant on top.
| Input | Example | Running total |
|---|---|---|
| Styles in the drop | 24 | 24 |
| Average colourways per style | 3 | 72 variants |
| Views per variant (PDP standard) | 5 | 360 images |
| Collection flat-lay per style | 1 x 24 | 384 |
| Lifestyle / campaign per style | 1 x 24 | 408 |
| Marketplace white hero per variant | 1 x 72 | 480 images |
Teams that plan in shoot-days almost always underestimate because the multiplication happens in the variant column. Six colourways of one hoodie is six times the retouching of one hoodie, even if the capture time is close to identical. The shot count is what the retouch queue and the upload queue actually see, so it is the number the plan has to be built on.
Build the shot count from the buy sheet or PLM export, not from memory. Merchandising already has the definitive list of styles and variants; the photo team's job is to multiply it by the view standard and hand the number back before the launch date is fixed.
Convert images into hours across three stages
Capture, retouch and delivery are three separate capacities with three separate bottlenecks, and a plan that only sizes the first one is half a plan. Rough throughput benchmarks for a small in-house or freelance setup:
| Stage | Typical throughput | 480-image drop |
|---|---|---|
| Capture (product, tabletop) | 60-100 finished frames per day | 5-8 studio days |
| Capture (on-model) | 25-40 looks per day | 2-3 model days |
| Manual retouch (cutout, clean, colour) | 10-20 minutes per image | 80-160 hours |
| QA, naming, resizing, upload | 2-4 minutes per image | 16-32 hours |
The retouch row is the one that surprises people. A five-day shoot produces roughly three weeks of retouching for one person, and retouching cannot start until the frames exist, so it lands entirely inside the window between the last shoot day and the launch. That is the interval where drops slip.
The honest way to plan is to work backwards from launch: delivery needs to finish two working days before the go-live so the landing page and email can be built against real images; retouch needs to finish before delivery starts; capture needs to finish before retouch can run at full rate. Lay those out and you will often find the shoot has to begin earlier than anyone expected, or the retouch stage needs a different approach.
Where the hidden queues sit
Beyond the three main stages there are four queues that rarely appear on a plan but routinely consume the buffer.
Sample arrival. Production samples turn up in waves, and the last wave is usually the colourway that was added late. If the shoot is scheduled as a single block and 20% of the samples miss it, that 20% needs its own capture slot, its own retouch pass and its own upload run. Plan for at least two capture windows, and read our guide to prepping product samples so the ones that do arrive are shoot-ready.
Approval loops. Every image that goes to a brand or merch lead for sign-off comes back with a percentage of rejections. At a 15% rework rate, 480 images produce 72 rounds of "fix the shadow, warm the white". A defined review and approval workflow with one reviewer, one round, and a reference image per category keeps that number from doubling.
Channel variants. The storefront wants 2000px square on a warm off-white; the marketplace wants pure white with the product filling 85% of the frame; the email template wants 600px wide with generous headroom. Each is a derivative of the master, but each is a separate export with its own checks.
Metadata. Filenames, alt text and image order are the last thing done and the first thing forgotten. Position 1 drives the collection grid on most platforms, so a variant that uploads in the wrong order is a visible defect on launch morning.
Decide what is captured and what is derived
The single biggest lever on capacity is the ratio of captured images to derived images. Every view that can be produced from a master frame rather than a new frame removes capture time, sample handling and studio cost, and moves the work into a stage that scales with compute rather than with calendar days.
Candidates for derivation in a typical drop:
- Colourways. Shoot the hero colour in full and derive the remaining colourways from it when the silhouette is identical. One captured hoodie plus five colour flips is one studio slot, not six.
- Backgrounds. Capture once on a clean plate, then place the cutout on the storefront tone, the marketplace white and the campaign scene. Never re-shoot for a background change.
- Flat lays and ghost mannequins. A garment captured on a form can be turned into a flat-lay view, and vice versa, which halves the styling time for the collection grid.
- On-model views. For a launch with fitted samples but no model day booked, an AI on-model render from the product shot fills the PDP slot until campaign photography lands.
Everything captured
- 72 variants x 5 views = 360 studio frames
- Every colourway needs a physical sample on the day
- Late samples force a second shoot
- Retouch scales with frames shot
Hero captured, rest derived
- 24 hero variants x 5 views = 120 studio frames
- Colourways derived from the hero, samples optional
- Late samples only matter if the silhouette changed
- Derivation runs in batches overnight
This is the model we have built Retouchable around: capture a clean hero, then batch the derived views (background, colour, flat-lay, on-model) with consistent framing and plate colour across the whole drop, so a 480-image requirement rests on 120 captured frames. The seasonal collections guide covers how the same split applies across a full calendar rather than one drop.
Build the plan backwards from the launch date
With the shot count, the stage throughputs and the capture-versus-derive split decided, the plan itself is a short backwards schedule. For a launch on a Tuesday:
| Milestone | Working days before launch | Gate |
|---|---|---|
| Line list frozen, shot count signed off | 20 | Merch confirms styles and variants |
| First sample wave in studio | 15 | At least 70% of hero samples present |
| Capture window 1 complete | 12 | All hero frames on the retouch queue |
| Derived views batch complete | 9 | Colourways, backgrounds, flat lays generated |
| Capture window 2 (late samples) | 7 | Remaining heroes captured and derived |
| Review round complete | 4 | Rework rate under 15%, no open rejections |
| Channel exports and metadata done | 3 | Filenames, alt text, position set |
| Images live on unpublished products | 2 | Landing page and email built on real assets |
Two rules make this hold. First, the line list freeze is a real gate: a style added after day 20 goes into capture window 2 and gets derived views only, or it launches with a single hero image and is completed in the week after. Second, the two-day buffer is not for imagery work; it exists so that marketing can build against final assets. Spending it on retouch means the email goes out with placeholders.
If the drop is going to Shopify, the launch imagery checklist lists the per-product items to tick off in that final window, including position ordering and alt text.
Measure the drop so the next one is easier
Capacity planning improves quickly when each drop leaves behind four numbers. Record them in the same sheet as the shot count so the next plan starts from evidence rather than from a guess.
- Frames captured per studio day, split by product and on-model. This is your real throughput, and it is usually lower than the benchmark on the first drop and higher by the third.
- Retouch minutes per image, split by captured versus derived. The derived number should be a small fraction of the captured one; if it is not, the derivation settings need tightening rather than the plan needing more hours.
- Rework rate at the review stage. Above 15% means the reference images or the brief were unclear, not that the retoucher was slow.
- Days late per sample wave. This tells you how far ahead of the shoot to set the sample deadline next time, and which suppliers need it earlier.
Within two or three drops these numbers make the shot-count-to-hours conversion almost mechanical, and the conversation with merchandising changes from "can we make the date" to "here is what the date costs in studio days, and here is what moving three styles to derived views saves". Our guide to product photography turnaround time covers the benchmarks to compare your figures against.
A shot count that grows after the freeze. Late additions are the single most common cause of a slipped drop, and they are nearly invisible if nobody is comparing the launch-morning image count against the day-20 signed-off number.