CR3 vs JPG: What Is the Difference?

A sensor does not record colour. Each photosite reports a single brightness value through a coloured filter, and turning that mosaic into a full-colour photograph requires reconstruction, plus decisions about white balance, contrast, saturation, sharpening and noise handling.

Shoot JPG and your camera performs the whole pipeline in the moment, writes the result, and discards the measurements. The picture style you had selected is not a label on the file — it is the file.

Shoot CR3 and the measurements are kept. The camera still records which settings were active, and any decent raw editor will open the file showing roughly what the camera showed you, but those settings are now instructions rather than facts. Change the white balance from daylight to shade three weeks later and the result is exactly as if you had set it that way at capture, because the underlying data never changed.

That is the whole difference. Everything else — the file sizes, the compatibility problems, the workflow friction — follows from it. The container itself is worth understanding too, because a CR3 also carries a full-size JPEG the camera made, which is why converting one is so much faster than people expect.

Expect a raw file to be five to ten times the JPG of the same frame. On a 45-megapixel body that is 20 to 60 MB against two to eight.

Three things account for it. Bit depth: 14 bits gives 16,384 levels per channel against 256, and most of the surplus sits in shadows and highlights where editing latitude lives. Compression: JPG uses lossy compression tuned to human vision and discards what you are unlikely to notice, while raw compression is far more conservative. And the embedded previews: every raw file carries the camera's own full-size JPEG rendering inside it.

Canon does offer a middle option. Selecting C-RAW rather than the standard raw setting produces files roughly 30 to 40 percent smaller at the same pixel dimensions, using a wavelet transform with a quantisation step — so some information is genuinely given up, unlike the standard setting. DPReview's testing found the difference invisible in normal use, with faint artefacts appearing only when shadows were pushed around four stops. For most people C-RAW is close to a free saving.

The real cost of raw is rarely the memory card, which is cheap. It is the archive, the backup, and the fact that every frame now needs a processing step before anyone else can see it.

In flattering light, a well-exposed JPG and a developed raw file of the same scene can be very hard to tell apart, and anyone claiming an enormous visible gap on a straightforward frame is overselling it.

The gap appears at the edges of the exposure, which in practice means:

Recovering highlights. A JPG that has clipped a bright sky has stored pure white and there is nothing behind it. The raw file frequently still holds detail a stop or so above what the rendering showed, which is the difference between a usable frame and a deleted one.

Lifting shadows. Push a JPG's shadows and you amplify eight-bit values plus whatever noise reduction the camera already applied, which shows up as banding and mush. Fourteen bits takes that push far more gracefully.

Fixing white balance. Mixed lighting — tungsten indoors against daylight through a window — is genuinely difficult to judge at capture. Raw lets you decide afterwards; a JPG needs colour correction that damages the file.

Heavy local adjustment. Dodging, burning and strong curves all benefit from having more levels to work with.

None of that matters if the frame was well exposed under even light and you liked the camera's rendering. That is the case for shooting JPG deliberately, and it is a real one.

This is where JPG wins outright, and it is not close. A JPG opens in every browser, every email client, every phone, every upload form and every printer's system, and has done for thirty years.

A CR3 opens in software that has a profile for your specific camera model — which is why a file can open on one machine and not another, and why a brand-new body sometimes opens in nothing but the manufacturer's own software. Uploading a raw file to a website or attaching one to an email for a client to view is not a workflow; it is a support ticket.

So raw is a working format and JPG is a delivery format. The photographers who look organised are simply the ones who accepted that early: shoot raw, develop, export JPGs sized for the destination. If your frames need no development, extracting the camera's own rendering takes seconds and skips the editor entirely.

Shoot raw for anything that matters or anything difficult: paid work, unrepeatable events, high-contrast scenes, mixed lighting, anything you might print large or edit heavily. The insurance is worth the disk space.

Shoot JPG when volume and turnaround beat latitude — a school sports afternoon delivered the same evening, real-estate frames going straight into a listing template, event photographs handed over on a card at the end of the night. Also when you shoot to the camera's rendering intentionally and have no intention of editing.

Shoot both if your camera supports it and your card can take it. Raw plus JPG gives you immediate deliverables and a negative to go back to, and it is the setting most working photographers land on for exactly that reason.

The one option to avoid is JPG-only for work you cannot re-shoot. Everything else here is a preference; that one is a risk.