RAW video

RAW video is a recording method in which a video camera stores the sensor’s photosite data before demosaicing and before most image processing, frame by frame, instead of a fully developed image. White balance, color rendering and much of the tone mapping become adjustable metadata that is applied later in post-production.

In video cameras, RAW sits at the top of the recording hierarchy, above log recording in a high-quality codec. It offers the greatest latitude for color grading, at the cost of larger files, heavier processing and a more complex workflow.

How RAW video captures sensor data

Most cinema sensors use a color filter array, usually a Bayer pattern, in which each photosite records only red, green or blue light. A conventional recording demosaics this mosaic into full RGB pixels inside the camera, applies white balance, a color matrix and a gamma curve, and then compresses the result with a codec such as H.264 or ProRes.

RAW recording stops much earlier in that chain. The camera writes the linear or log-encoded sensor values with their original bit depth, typically 12 to 16 bits per photosite, together with metadata describing the intended white balance, exposure index and color space. The demosaic and color transform then run in the editing or grading software, where they can be changed without losing quality.

Some formats take a middle path. They perform a partial demosaic or sensor-specific preprocessing in the camera to reduce the computing load in post, while still keeping white balance and exposure index as non-destructive settings.

RAW formats and compression

RAW video is not one format. Each manufacturer defines its own container, compression method and decoding software, and the table below compares the main families by their general approach.

Compression in RAW is applied to the mosaic itself. Lower compression ratios preserve fine detail and noise texture, while higher ratios can introduce artifacts in fine patterns and dark gradients. Many productions treat moderate ratios as visually lossless for delivery, but the safe setting depends on resolution, scene content and the intended grade.

RAW video compared with log and processed recording

The key distinction is what is baked into the file. Log recording, a neighboring concept, stores a debayered image with a flat curve to preserve dynamic range, but white balance, demosaic quality and noise reduction are already fixed. RAW leaves more of these decisions open.

A common misconception is that RAW increases dynamic range. It does not: the sensor sets the dynamic range. RAW preserves that range with more precision and flexibility, while a well-exposed log recording at 10 or 12 bits can already retain most of it.

Practical use and choosing settings

Choosing RAW is a workflow decision, not only an image-quality decision. Storage, offload time, computer performance and the experience of the post-production team all matter, so productions weigh the benefits against the cost per hour of footage.

On set, exposure still matters. RAW cannot recover highlights that clipped at the sensor, and underexposure still raises noise. Teams monitor with a viewing LUT that previews the intended look, while the recorded data keeps the full range for the grade.

Some cameras cannot record RAW internally and instead send the sensor signal over SDI or HDMI to an external recorder, which writes ProRes RAW or a similar format. This adds weight and cabling but can unlock RAW on otherwise compact cameras.

Limitations and common mistakes

RAW video brings real operational costs that are easy to underestimate when only the image benefits are considered.

• Storage and backup volume can reach several terabytes per shooting day at high resolutions and low compression.
• Decoding demands strong hardware, so editors often create proxies or rely on GPU acceleration.
• Software support varies by format, and an older application may not read a newer camera’s RAW.
• Metadata can be lost if files are transcoded too early, which removes the adjustable settings.
• Assuming RAW fixes exposure errors leads to clipped highlights and noisy shadows that no grade can repair.

A disciplined workflow avoids most of these issues: verified offloads with checksums, a consistent color management pipeline and early tests of the full chain from camera to final delivery.

In everyday language

RAW video is like keeping the undeveloped negative instead of a finished print. A normal camera decides on colors, brightness and sharpness the moment it records. A RAW camera saves the raw ingredients and lets the editor make those decisions later.

That flexibility helps when lighting was tricky or when the film needs a very particular look. The trade-off is that the files are much bigger and need more powerful computers. RAW gives more freedom later, not a free pass for mistakes now: a scene that was far too bright or too dark when filmed still cannot be fully rescued.

Seen in the real world

A small crew is filming a short drama in a house where daylight from the windows mixes with warm tungsten lamps. There is no time to gel every window, and the director wants to decide the final color mood later with the colorist.

The cinematographer records RAW at a moderate compression ratio at 4K and 24 fps, exposing carefully to protect the window highlights while monitoring through a viewing LUT. Each day produces roughly one to two terabytes, which the data wrangler copies to two drives with checksum verification.

In the grade, the colorist adjusts the white balance per shot without banding, warms the interior scenes and cools the night exteriors. The result is a consistent, deliberate look that would have been difficult to achieve from footage with the camera’s white balance baked in.