Flikipedia A branching dictionary of the moving image

raw image

noun

Minimally processed data read directly from an image sensor, recorded before demosaicing and rendering so that white balance, exposure, and color can be reinterpreted later with greater latitude.

Origin From the ordinary adjective raw, meaning uncooked or unprocessed; applied to sensor output that has not yet been developed into a finished image. Often written in capitals (RAW) by analogy with file-format acronyms, though it is not itself an acronym.

  1. Imaging The unrendered numerical output of an image sensor, in which each photosite is stored as a single measured value together with the metadata needed to interpret it, before demosaicing, white balance, and tone mapping are applied. A camera set to record raw retains the sensor data itself, so exposure and color-temperature decisions can be revised in post-production without the loss inherent in re-editing a rendered file.

  2. File format A file or data stream holding such output, as distinct from a rendered and encoded image (for example JPEG) that has already been processed into display-ready pixels.

What raw data is

A raw image is the data an image sensor produces before it is turned into a viewable picture. Each photosite on the sensor measures light intensity as a single number, and in a raw recording those numbers are stored directly, accompanied by metadata such as the intended white balance, exposure index, and the color-filter-array layout. Because most sensors carry a Bayer or similar color-filter array, a raw frame is typically a single-channel mosaic rather than a full-color image: the red, green, and blue values at each pixel have not yet been reconstructed. Interpreting the data into a color image requires demosaicing, white-balance scaling, and tone mapping, steps that a raw file defers rather than bakes in.

Difference from a processed image

A processed or encoded image, such as a JPEG still or a delivery-ready video stream, has already had those rendering decisions applied and fixed. White balance, contrast, gamma or a display transfer function, sharpening, and compression are committed to the pixels, and the sensor’s original values are discarded. A raw image keeps the sensor measurements intact, so the same rendering choices remain adjustable afterward. The practical distinction is one of stage: raw sits upstream of rendering, while a processed image is the output of it.

Role in grading and dynamic range

Retaining unrendered data widens the latitude available in color grading. Raw values are usually stored at a higher bit depth, commonly 12 to 16 bits per photosite, than the 8 bits typical of many rendered formats, and they are recorded before the highlights and shadows are mapped to a display range. This preserves more of the sensor’s captured dynamic range, allowing exposure and color temperature to be reset and tonal detail in bright or dark regions to be recovered in post-production with less banding or clipping than editing a rendered file would allow.

Trade-offs in file size and workflow

The added latitude carries costs. Raw files are substantially larger than encoded equivalents because they store more bits per sample and use lighter or lossless compression, which increases storage and data-rate requirements, especially for motion-picture footage recorded at high resolution and frame rates. Raw material is also not display-ready: it must be processed through a decoding and rendering pipeline, governed by a codec or manufacturer-specific software, before it can be viewed normally, adding a step to editing and review workflows. These demands are weighed against the flexibility raw provides.

See also

  • image sensor— The electronic component that captures a digital image, converting light into electrical charge at a grid of photosites — the digital counterpart to photographic emulsion.
  • color grading— The post-production process of adjusting the color and tonal values of a film's images for consistency across shots and for expressive or narrative effect.
  • dynamic range— The span of light intensities a recording medium can register between its noise floor in the shadows and its point of saturation in the highlights, conventionally measured in photographic stops.
  • codec— A method, implemented in software or hardware, for encoding and decoding digital video — compressing image data for storage or transmission and reconstructing it for playback.

Sources

  1. SMPTE and camera-manufacturer documentation on raw acquisition formats
  2. Blain Brown, "The Filmmaker's Guide to Digital Imaging" (color and workflow chapters)

Last revised: September 18, 2026