Intra-frame codec

Also known as: all-intra

On this page 7

An intra-frame codec is a video compression method that encodes every frame independently, using only the information inside that single picture and never referencing neighboring frames. Among video codecs, this approach is also called all-intra compression, and it trades larger files for fast decoding, frame-accurate editing and robust quality in every frame.

Intra-frame codecs dominate acquisition and post-production, where editors scrub, cut and grade footage constantly. Formats such as ProRes and DNxHR, and the all-intra profiles of H.264 and H.265, are typical examples, while delivery formats usually switch to inter-frame compression to save bandwidth.

How intra-frame compression works

An intra-frame encoder treats each frame like a standalone still image. It removes spatial redundancy, meaning similarities between nearby pixels within the same picture, but it ignores the temporal redundancy between consecutive frames that inter-frame codecs exploit.

The typical pipeline divides the frame into blocks, applies a frequency transform such as the discrete cosine transform or a wavelet transform, quantizes the resulting coefficients and then packs them with entropy coding. Quantization is where most information is discarded, so its strength largely determines both quality and bitrate.

The following table summarizes the main stages and their effect on the result.

StageWhat it doesEffect on output
Block partitioningSplits the frame into regionsSets the granularity of compression
Intra predictionPredicts a block from already coded neighbors in the same frameReduces residual data
TransformConverts pixel values to frequency coefficientsConcentrates energy into few values
QuantizationRounds coefficients to coarser stepsMain source of loss and size reduction
Entropy codingEncodes symbols with variable-length codesLossless size reduction
Stages of an intra-frame encoder

Because no frame depends on another, every frame is effectively a keyframe, and the GOP length equals one frame. Any frame can be decoded directly, without first reconstructing a chain of predicted frames.

Intra-frame versus inter-frame codecs

The essential difference lies in what a frame may reference. An inter-frame codec, also called a long-GOP codec, stores occasional full I-frames and encodes the frames in between as P-frames and B-frames that describe only changes and motion relative to other frames.

This comparison shows how the two families behave in practice.

PropertyIntra-frameInter-frame (long GOP)
Frame referencesNone, each frame stands aloneFrames depend on other frames
Bitrate for equal qualityRoughly 3 to 10 times higherMuch lower
Decoding load per frameLow and constantVariable, can be high
Scrubbing and cuttingInstant and frame accurateSlower, needs reconstruction
Motion handlingQuality stable in fast motionCan degrade in complex motion
Error resilienceAn error affects one frameAn error can spread through a GOP
Typical roleCapture, editing, masteringStreaming, broadcast, distribution
Intra-frame and inter-frame compression compared

The bitrate ratio is not fixed: it depends on content, resolution and encoder settings. Static scenes favor inter-frame coding strongly, while grain, noise and fast motion narrow the gap because there is little temporal redundancy to exploit.

Common intra-frame formats and profiles

Intra-frame compression appears both in dedicated intermediate codecs and as all-intra profiles of general-purpose standards. The table lists widely used examples with their typical use.

FormatBasisTypical use
ProRes familyDCT-based intermediate codecEditing, camera recording, mastering
DNxHD and DNxHRDCT-based intermediate codecBroadcast and editorial workflows
H.264 All-IntraH.264 with intra-only profileCamera recording at high quality
H.265 All-IntraH.265 with intra-only profileEfficient high-resolution capture
JPEG 2000Wavelet transformDigital cinema packages, archiving
Motion JPEGJPEG per frameLegacy and simple capture devices
Widely used intra-frame formats

Many cameras offer the choice between an All-I and an IPB mode within the same codec. For example, a camera may record H.264 All-I at around 400 Mbit/s and H.264 IPB at around 100 Mbit/s at the same 4K resolution.

Choosing an intra-frame codec

Intra-frame coding is the right choice when editing performance and image integrity matter more than storage. The decision table below matches common situations with a recommended approach.

SituationRecommended choiceReason
Heavy grading and effectsIntra-frame at high bitratePreserves detail and avoids generational loss
Multicamera editingIntra-frame proxies or mastersSmooth playback of many streams at once
Long event recording on limited mediaInter-frameMuch smaller files
Final online deliveryInter-frame such as H.264, H.265 or AV1Bandwidth efficiency for viewers
Fast action with grain or noiseIntra-frameStable quality without motion artifacts
When to use intra-frame compression

Storage planning is critical. At 400 Mbit/s, one hour of footage occupies about 180 GB, so media, transfer speed and backup capacity must scale accordingly. Many workflows therefore record intra-frame in camera and transcode to a long-GOP format only at the delivery stage.

Limitations and common mistakes

The main limitation is storage and bandwidth cost. Intra-frame files are unsuitable for streaming to viewers and can overwhelm slow cards, which must sustain the full write speed without dropping frames.

A second limitation is efficiency at low bitrates. When an intra-frame codec is forced to a small bitrate, every frame must be quantized heavily, and visible blocking and smearing appear sooner than in a long-GOP encode of the same size. Intra-frame compression is therefore worthwhile only when the bitrate budget is generous enough to keep each frame close to visually transparent.

  • Assuming all-intra means lossless: most intra-frame codecs are lossy, only at a gentler level.
  • Using cards below the required write speed: high-bitrate All-I modes need media rated for sustained throughput.
  • Confusing intra prediction with inter prediction: intra prediction works only inside one frame.
  • Delivering intra-frame masters to audiences: delivery should use an efficient inter-frame codec.

The idea in brief

Video is a rapid sequence of pictures. An intra-frame codec shrinks each picture on its own, the way a phone saves every photo as a separate file. No picture depends on the one before it.

That makes the files bigger, but it also means any frame can be opened instantly. Editors like this because jumping, cutting and adjusting color feel smooth, much like flipping through a stack of printed photos instead of rewinding a tape.

A practical scenario

A documentary team films a fast-moving street festival in 4K at 25 frames per second. They set their cameras to an intra-frame mode at about 470 Mbit/s instead of the 150 Mbit/s long-GOP option, because confetti, crowds and handheld motion would challenge an inter-frame encoder.

The choice costs storage: each camera produces roughly 210 GB per hour, so the team carries several high-capacity cards and backs them up each evening. In post-production, the editor cuts a four-camera sequence on a laptop without creating proxies, and the colorist pushes the footage hard without blocking artifacts.

For delivery, the finished film is exported to H.265 at a fraction of the size. The result is clean, detailed footage during editing and an efficient file for viewers.