Every time a digital camera captures an image, the sensor records raw electrical signals from millions of photosites. What happens next — how those signals are converted into a viewable picture — depends entirely on whether you record in JPEG or RAW. The difference is not just technical: it affects how much flexibility you have later in post-processing, how much information you lose at capture, and what mistakes can and cannot be corrected afterward.

What a RAW file actually is

A RAW file is not an image in the traditional sense. It stores the unprocessed sensor data alongside metadata about the camera settings at the time of capture. The camera's image processor has not applied sharpening, noise reduction, contrast curves, or colour science — all of that happens when you open the file in a RAW converter.

The sensor uses a Bayer filter array — a mosaic of red, green, and blue filters placed over individual photosites in a 2:1:2 ratio. Because each photosite only measures one colour channel, the RAW converter must estimate the other two using surrounding pixels through a process called demosaicing. Different converters use different demosaicing algorithms, which is one reason the same RAW file can look visually distinct in Lightroom versus Darktable or RawTherapee.

Bayer colour filter array pattern used in digital camera sensors
The Bayer colour filter array. Each photosite captures only red, green, or blue — the missing colours are interpolated during RAW conversion. Source: Wikimedia Commons (CC BY-SA).

RAW files from different manufacturers use proprietary formats: Canon uses CR2 and CR3, Nikon uses NEF, Sony uses ARW. Adobe's DNG is an open alternative that many photographers use for long-term archiving, since it does not depend on any single manufacturer's software remaining available in the future.

Bit depth and dynamic range

Modern digital cameras record RAW data at 12 or 14 bits per channel. A 14-bit file can represent 16,384 distinct tonal values per channel, compared to 256 per channel in an 8-bit JPEG. That extra information is what allows you to recover detail in highlights and shadows that would be permanently lost in a JPEG.

Dynamic range — the camera's ability to capture detail across bright and dark areas simultaneously — is closely tied to bit depth, but also to sensor design and ISO setting. At base ISO (typically 100 or 200 depending on the camera), most modern full-frame sensors can capture between 13 and 15 stops of dynamic range. At higher ISO settings, that range compresses rapidly, and the noise floor rises, reducing how much shadow recovery is practical in post-processing.

Shooting RAW does not automatically produce better images than JPEG — it preserves more data from the sensor and gives you more choices later. If your exposure is accurate and your colour balance is set correctly in-camera, a JPEG and a RAW will look nearly identical after basic processing.

The exposure triangle and RAW files

Aperture, shutter speed, and ISO interact to determine how much light reaches the sensor and what the resulting image looks like. Even when shooting RAW, getting the exposure as accurate as possible at capture reduces the amount of correction needed in post — and always produces a technically better file than one that was significantly over- or underexposed and then corrected.

Aperture

Aperture is expressed as an f-number (f/1.4, f/2.8, f/8, etc.). Wider apertures (lower f-numbers) let in more light and produce shallower depth of field. At f/1.8 on a 50mm lens, a subject 2 metres away will be in focus while the background blurs into a smooth gradation — useful for isolating a subject in portrait work. At f/11, depth of field extends significantly further, which is typically desirable for landscape photography.

Comparison of depth of field at different aperture settings
The same scene photographed at a wide aperture (left) and narrow aperture (right). The plane of sharp focus is much shallower at wider apertures. Source: Wikimedia Commons (Public Domain).

Shutter speed

Shutter speed controls how long the sensor is exposed to light. Fast shutter speeds (1/1000s or faster) freeze motion — useful for sports, birds in flight, or water droplets. Slow shutter speeds (1/30s or slower) introduce motion blur in moving subjects and also amplify camera shake if the camera is not stabilised. Long exposures of several seconds or minutes are used for astrophotography and light trails.

ISO

ISO amplifies the signal from the sensor. At ISO 100, the amplification is minimal and noise is low. At ISO 3200 or 6400, the signal is amplified significantly, which also amplifies electronic noise — visible as a grain-like texture, particularly in shadow areas. Modern cameras handle high ISO reasonably well, and noise reduction in RAW converters has improved substantially. Still, using the lowest ISO that gives a correct exposure at a suitable shutter speed remains the standard approach.

In-camera settings that still matter when shooting RAW

Many photographers assume that because RAW files preserve raw sensor data, all camera settings are irrelevant. That is not entirely accurate. Certain settings affect the RAW file directly:

  • ISO: Directly affects signal amplification and noise in the RAW data.
  • Exposure compensation: Changes the actual exposure, not just an in-camera preview setting.
  • Long exposure noise reduction: Takes a dark frame and subtracts it from the RAW data — it affects what is recorded.
  • Lens corrections: On some cameras, vignetting and distortion corrections are baked into the RAW metadata and applied automatically by some converters.

Settings that affect only the embedded JPEG preview (and are non-destructive when processing RAW) include: picture style/profile, sharpening level, saturation, and white balance. These influence how the image looks on the camera screen and in software that uses the embedded preview, but a RAW converter will use the actual sensor data.

White balance in RAW files

White balance tells the camera and converter what colour temperature to treat as neutral white. In RAW files, white balance is stored as metadata — the actual sensor data does not change. You can set white balance to any value in post-processing without any quality loss, as long as you did not clip the highlights in one channel.

Setting white balance correctly in-camera is still useful because it makes the preview on the screen and the converter's starting point more accurate, reducing the time spent adjusting. Using a grey card or X-Rite ColorChecker at the start of a shoot gives a reference for the exact colour temperature and tint of the light, which can then be applied to the entire set of images in batch.

Histograms and exposure

The histogram displays the tonal distribution of an image: dark tones on the left, bright tones on the right. Clipping on the right side of the histogram indicates blown highlights — areas where one or more colour channels have reached maximum value and detail cannot be recovered. Shadow clipping on the left is generally more recoverable in RAW files, but only to a degree determined by the noise floor of the sensor at the ISO setting used.

A technique called ETTR (Expose to the Right) involves deliberately placing the exposure as bright as possible without clipping the most important highlights. This maximises the signal-to-noise ratio in the RAW file. The image will look overexposed in the preview, but the tones are brought back down in the RAW converter. At low ISO, ETTR can meaningfully improve shadow quality. At high ISO, the benefit diminishes because read noise is already low relative to the signal.

RAW converters and their role

The RAW converter is where sensor data becomes an image. The major options are Adobe Lightroom and Camera Raw, Capture One, Darktable, and RawTherapee. Each applies its own demosaicing algorithm, colour profiles, and default processing. Because of this, the same RAW file opened in different converters will look different by default, even with identical settings.

Colour profiles are particularly significant. Adobe includes camera profiles built from profiling individual camera models — DCP profiles — that map sensor colours to known reference colours. Capture One has a reputation for its colour rendering, particularly skin tones. Darktable uses a combination of colour matrices and tone curves that can be configured in detail. None of these is inherently superior — they represent different choices about how to translate raw data into a finished image.

For further reading on RAW file structures, the Adobe DNG specification is publicly available and documents how metadata, colour information, and image data are stored. The DPReview sensor comparisons and studio test scenes are a useful reference for comparing dynamic range across camera models.

Understanding RAW files as raw sensor data — not as finished images — is the foundation of effective digital photography. The choices you make at capture affect what is recoverable in post, and the converter you use shapes how that data translates into colour, contrast, and detail.