What Is RAW Format?
Ask any experienced photographer whether to shoot RAW or JPEG, and most will say RAW without hesitation. But ask them to explain what a RAW file actually contains and why it matters, and the answers often drift into vague assertions about “more data” and “better quality.” The reality is more specific and more consequential than those generalities suggest.
A RAW file is the unprocessed output of the camera’s image sensor. When light strikes each photosite (pixel) on the sensor, it generates an electrical charge proportional to the number of photons absorbed. The camera’s analog-to-digital converter measures this charge and records it as a numerical value — typically 12 or 14 bits per photosite, yielding 4,096 or 16,384 discrete tonal levels per channel. A RAW file stores these values with no color processing, no sharpening, no noise reduction, no contrast adjustment, and no compression-induced data loss. It is a faithful record of what the sensor measured.
A JPEG file, by contrast, is the result of extensive in-camera processing. The camera applies a color profile, white balance correction, contrast curve, noise reduction, sharpening, and then compresses the result into 8-bit-per-channel JPEG format (256 tonal levels per channel). This processing is permanent and destructive — the original sensor data is discarded. A JPEG is a finished photograph produced by the camera’s built-in software. A RAW file is the raw material from which a photograph can be produced with far more control.
How It Works
Digital camera sensors use a Bayer color filter array (CFA) in the vast majority of cameras. Each photosite records only one color — red, green, or blue — through a microscopic filter placed over it. The Bayer pattern arranges these in a 2x2 grid: one red, one blue, and two green (green is doubled because the human eye is most sensitive to green wavelengths). A 24-megapixel sensor has approximately 12 million green-filtered, 6 million red-filtered, and 6 million blue-filtered photosites.
A RAW file stores the single-color measurement from each photosite without interpolation. The process of generating a full-color image from this mosaic data — called demosaicing — happens later, in software. Different demosaicing algorithms (bilinear, VNG, AHD, DCB) produce slightly different results in edge sharpness, color accuracy, and artifact suppression. By deferring this step to post-processing, the photographer can choose the algorithm, adjust its parameters, and reprocess the file years later as better algorithms emerge.
Bit depth determines tonal resolution. A 14-bit RAW file records 16,384 tonal steps from pure black to pure white per color channel. An 8-bit JPEG records 256 steps. This difference matters most in gradients and shadow recovery. A smooth sky gradient that occupies 5 percent of the tonal range spans approximately 820 steps in a 14-bit file but only 13 steps in an 8-bit file. Those 13 steps can produce visible banding — abrupt tonal jumps — especially after contrast adjustments that stretch the tonal range further. The 14-bit file maintains smooth gradations even after aggressive editing.
White balance in a RAW file is metadata, not a baked-in correction. The RAW converter reads the camera’s recorded white balance setting and applies it as a starting point, but the photographer can change it freely to any color temperature from 2,000K to 50,000K with no quality loss. In JPEG, white balance is applied during processing and compressed into the 8-bit output — correcting a severe white balance error in a JPEG introduces color banding and posterization. Correcting the same error in the RAW file produces a result indistinguishable from having set the correct white balance in camera.
Dynamic range recovery is the single largest practical advantage of RAW files. A 14-bit RAW file from a modern full-frame sensor contains approximately 14 to 15 stops of recoverable dynamic range. Pulling an underexposed shadow area up by 3 stops in a RAW file reveals clean detail because the original 14-bit data contains sufficient tonal information. The same 3-stop push in a JPEG produces visible noise, color shifts, and banding because the 8-bit compression has already discarded the subtle tonal differences in shadow regions.
Practical Examples
Wedding photography illustrates RAW’s safety net. A ceremony in a church with stained glass windows may span 12 stops of dynamic range — from sunlight through colored glass to shadowed pews. Shooting RAW, the photographer exposes for the highlights (the windows) and recovers shadow detail in the pew area during editing, pulling up exposure by 2 to 3 stops with clean results. A JPEG exposed the same way would lose all shadow detail permanently, with no recovery possible.
Landscape photography at golden hour often involves a bright sky and dark foreground spanning 10 to 14 stops of brightness. A single RAW exposure processed with graduated adjustments can recover 4 to 5 stops of highlight detail in the sky and 3 to 4 stops of shadow detail in the foreground, often eliminating the need for graduated ND filters or HDR bracketing. The same latitude does not exist in JPEG files, where blown highlights contain no recoverable data.
Sports photography produces thousands of frames per event, and RAW files ensure that auto white balance inconsistencies across different lighting zones (daylight from windows, tungsten arena lights, LED scoreboards) can be corrected to a consistent color temperature in batch processing. Applying a unified white balance to 500 JPEGs with inconsistent in-camera processing produces uneven color. Applying it to 500 RAW files produces uniform results because the color temperature was never baked in.
Portrait retouching benefits from RAW’s tonal flexibility. Skin tone adjustments — reducing redness, evening out exposure across the face, recovering detail in highlights on the forehead or nose — are cleaner in 14-bit data than in 8-bit. A skin smoothing workflow that involves luminosity masking and frequency separation produces fewer artifacts when starting from a 14-bit source because the additional tonal resolution prevents banding in the subtle gradients of skin.
Advanced Topics
RAW file formats are proprietary and camera-specific. Canon uses .CR2 and .CR3, Nikon uses .NEF, Sony uses .ARW, Fujifilm uses .RAF, and Panasonic uses .RW2. Each encodes sensor data differently, with different metadata structures and sometimes different compression schemes. Adobe’s DNG (Digital Negative) format is an open standard intended to address this fragmentation — it wraps RAW data in a standardized container with documented metadata fields. Some cameras (Leica, some Pentax models) write DNG natively; others require conversion.
Lossy RAW compression is offered by some cameras as a space-saving option. Nikon’s compressed NEF files use a curve that reduces 14-bit data to approximately 12 bits of effective precision, saving 30 to 40 percent file space with minimal visible quality loss. Sony’s “compressed RAW” applies a similar lossy reduction. The differences between uncompressed and lossy-compressed RAW become visible only in extreme editing — 4+ stop shadow recovery — making lossy RAW a practical choice for most workflows and a poor choice for high-dynamic-range landscape work.
Lossless compression, distinct from lossy compression, uses algorithms similar to ZIP that reduce file size by 20 to 40 percent with zero data loss. The decompressed file is bit-identical to the original. Most cameras default to lossless compressed RAW as the best balance of file size and quality. There is no reason to shoot uncompressed RAW unless your workflow requires maximum write speed (uncompressed files write faster because the camera skips the compression step).
RAW files contain a full-resolution JPEG preview embedded within them. This is what the camera displays on the rear LCD and what produces the histogram on the camera screen. The histogram shown on a camera reviewing a RAW file represents the embedded JPEG preview, not the RAW data. This means a RAW file may contain 2 to 3 stops of additional recoverable highlight data beyond what the camera histogram shows as clipped. Photographers who shoot RAW can safely “expose to the right” — deliberately overexposing by 1 to 1.5 stops beyond the JPEG histogram’s clipping point — to maximize signal-to-noise ratio in the shadows while recovering highlights in post-processing.
Computational RAW formats are emerging. Apple’s ProRAW combines traditional RAW data with computational photography processing (Deep Fusion multi-frame noise reduction, Smart HDR tone mapping) into a DNG file that retains 12-bit editing flexibility. Google’s UltraHDR format embeds a gain map in JPEG files that extends dynamic range beyond 8-bit limits. These hybrid formats blur the line between RAW and processed output, offering some of RAW’s flexibility with the convenience of a finished image.
ShutterCoach Connection
ShutterCoach reads the file format from your image metadata and adjusts its feedback accordingly. When analyzing a JPEG, it notes which editing options are limited by the 8-bit compression and flags situations where shooting RAW would have preserved highlight or shadow detail that was permanently lost. When analyzing a RAW file, it takes full advantage of the available dynamic range to assess exposure accuracy across the entire tonal range.