RAW Files in Client Review: Why Clients Get Proxies

Why a client cannot review a RAW file at all

The first reason is practical. RAW is not one format. Canon, Nikon and Sony each write their own proprietary containers (CR3, NEF and ARW are the common ones), and Adobe's DNG is a fourth. A browser has no decoder for any of them, and a phone's gallery app is no better. Even on a laptop with the right software, the client would see the camera's embedded preview or a default rendering, not your edit.

The second reason is that a RAW file is not a picture yet. It is sensor data plus instructions: white balance, exposure, tone curve, lens corrections. Two people opening the same RAW in different software can see two different images. A review needs everyone looking at one rendering, and only a rendered file such as a JPEG can promise that.

The third is size. The illustrative numbers below are assumptions, not measurements, chosen so you can swap in your own camera's figures:

Dividing 40 GB by 2.4 GB gives about 17 times less data for the full-size proxy, and the grid view the client first lands on is smaller still. That ratio, not any clever feature, is why a review link opens on a phone on hotel wifi while a RAW delivery does not.

The three sizes a review needs

A review gallery is really three images per frame, each doing a different job:

• Thumbnail. Small, fast, used for the grid. Its only job is recognition: which frame is this? It should load in bulk without the client noticing.
• Preview. Sized to fill a screen, not to be printed. This is the image the client actually judges, marks up and comments on. Sharpness and colour matter here, so it deserves a good JPEG quality setting rather than an aggressive one.
• Master. The full-resolution deliverable, which stays out of the review path until the client has chosen.

Appruva follows the same shape internally: every image keeps a storage path for the file plus a thumbnail and a preview derived from it, and the newest round of a frame is the one every surface shows. The point is not the product detail but the pattern, which you can reproduce with any tool: the client browses small, judges medium, and receives large only after the decision.

What a client can and cannot judge from a proxy

Clients decide on composition, expression, moment, mood and rough colour. A 2,000-pixel-wide JPEG shows all of those perfectly well. Where a proxy falls short is anything that depends on pixels at 100%: critical focus on an eye, dust spots, fine skin texture, moiré, banding in a gradient.

That gap is a reason to treat the review as the selection step, not the quality-control step. The client chooses which frames and what direction the edit takes; you or your retoucher inspect the full-resolution file before delivery. It also tells you what not to put in the client's hands: do not ask them to confirm sharpness from a proxy, and do not accept "it looks soft" as a verdict on the RAW without checking it yourself.

If you are deciding how much of the shoot the client should see in the first place, show clients every frame or only your selects covers that choice and why it changes how many proxies you need to make.

Where the decision gets lost: mapping back to the RAW

Here is the failure that costs real time. You cull 800 RAWs, export 120 JPEG proxies, and the client picks 31 of them. Now you need the 31 matching RAW files. If the proxies were renamed on export, or the folder was reorganised, or two cameras produced the same frame number, you are matching by eye.

Three habits prevent it:

1. Keep the original filename in the proxy. Export A7R_0412.ARW as A7R_0412.jpg, not wedding_final_17.jpg. The stem is the join key.
2. Make stems unique across cameras. If you shoot with a second body or a second shooter, prefix with a camera or shooter code before export. Our post on file naming conventions for photo teams walks through a scheme that survives a handoff.
3. Take the selects as a list, not as a folder. Ask the review tool for the chosen filenames and filter your catalog by them. Copying files by hand into a "client picks" folder is where the stragglers appear.

The arithmetic of getting this wrong is simple. If hunting down mismatched frames takes just two minutes each for 31 frames, that is 62 minutes (31 × 2) of unbilled time on one shoot. Do that across 20 shoots and it is more than 20 hours a year (62 × 20 = 1,240 minutes), all caused by a filename decision you made at export.

Edits, versions and the proxy that goes stale

A proxy is a snapshot. The moment you re-edit a RAW, every proxy of it is out of date, and the dangerous case is the one nobody notices: a client approving round one's rendering while you believe they approved round two's.

Treat each re-export as a new version of the same frame, not a new file beside the old one. In a review tool that keeps rounds, the newest version becomes current and the earlier ones stay behind it for comparison; in a folder-based workflow you need a convention that does the same job. Our piece on finding the current edit goes through the options.

Keep the RAW and its edit instructions together, export the proxy from that pair, and never edit a proxy in place. The negative stays the source of truth; the proxy is only ever derived from it.

When a client genuinely asks for the RAW files

It happens: an in-house retoucher wants to grade the shoot themselves, or a magazine's production team insists on source files. Treat that as a separate deliverable with its own terms, not as an extension of the review.

Settle three things before any RAW leaves your hands. First, licence: who may edit them and under whose credit. Second, scope: all frames, or only the approved selects. Third, timing: after final payment, not before. Sending only the approved selects keeps the delivery small and keeps unflattering outtakes out of circulation, which matters more than most photographers expect.

Even then, the review itself should stay on proxies. The client approves the look on the proxy, and the RAW handover is a logistics step that follows the approval. Mixing the two is how a quick sign-off turns into a week of back-and-forth about files nobody can open on the first try.

A practical RAW-to-review checklist

Pulling the previous sections into a sequence you can run after any shoot:

1. Cull in the RAW catalog. Reject, then flag the candidates for the client.
2. Apply the baseline edit and soft-proof in the colour space you will deliver.
3. Export proxies at screen size with the original filename stem intact.
4. Upload the proxies to the review link and let the thumbnails and previews be generated from them.
5. Collect the client's selects and per-frame notes in one place.
6. Filter the catalog by the selected filenames and finish those RAWs at full resolution.
7. Deliver the masters as a new round, so what the client approved and what they received can be told apart.

None of this needs special software. It does need the discipline of treating the review as a conversation about proxies and the delivery as a separate step about masters, and a tool that records the client's decisions against the frame rather than in an email.