Compression
What is Compression?
Compression makes image, video, and audio files smaller so they're easier to store and share. Depending on the type, it either keeps all the original data intact or throws away some details the viewer is unlikely to notice.
At a glance
- Also known as
- EncodingVideo encodingCodec compressionData compression
- Used for
- Reducing file size for storage and deliveryStreaming optimisationArchival encodingWeb and mobile delivery
- Common tools
- HandBrakeFFmpegAdobe media encoderDaVinci resolveCompressor (macOS)
- Related terms
- CodecBitrateArtifacts (visual)ResolutionProRes
Ready to create?
Direct scenes, design characters, and ship full films
All-in-one AI creative platform with simple, transparent pricing, no speed throttles, and an infinite Canvas for max creativity.
How it compares
Lossless compression reduces file size without any quality loss, allowing perfect reconstruction of the original data: suitable for editing and archival but achieving smaller size reductions. Lossy compression achieves much smaller file sizes by permanently discarding data, introducing some quality degradation: appropriate for final delivery and streaming where bandwidth efficiency is prioritised.
Think of it like…
Lossless compression is like folding a map: all the information is still there, just packed more neatly, and you can unfold it to get back exactly what you started with. Lossy compression is like summarising that map: you keep the major roads and landmarks but leave out the minor details, resulting in something much smaller that's still useful, but not quite a perfect reproduction of the original.
Pro tip
When preparing footage for AI upscaling or restoration, always use the highest bitrate source available: heavily compressed input material limits the quality ceiling of any AI processing applied to it, regardless of how powerful the model is.
Types and variations
- Lossless compression preserves all original data and is used for archival, editing, and master files.
- Lossy compression discards perceptually less important data and is used for delivery and streaming.
- Intra-frame compression (such as ProRes and MJPEG) compresses each frame independently, making it edit-friendly but less space-efficient.
- Inter-frame compression (such as H.
- 264 and H.
- 265) encodes differences between frames, achieving higher efficiency but making the footage harder to edit without re-encoding.
- Perceptual compression (used in audio formats such as MP3 and AAC) is specifically designed to remove frequencies human hearing is less sensitive to.
Ready to make your first scene in Morphic?
Try MorphicCommon use cases
- Compression is applied at virtually every stage of a media production workflow.
- Camera footage is captured in compressed or uncompressed formats depending on the camera and production requirements.
- During editing, high-quality intermediate codecs such as ProRes or DNxHD are used to maintain quality through multiple processing generations.
- At delivery, masters are compressed to streaming formats (H.
- 264 or H.
- 265) optimised for the target platform.
- In AI workflows, source footage quality ( and therefore its compression level ) directly affects the quality of outputs from upscaling, restoration, and generative tools.
Ready to create?
Direct scenes, design characters, and ship full films
All-in-one AI creative platform with simple, transparent pricing, no speed throttles, and an infinite Canvas for max creativity.
FAQs
A codec is the algorithm that compresses and decompresses the data. A container (such as.mp4 or.mov) is the file format that holds the compressed video, audio, and metadata together. The same container can hold different codecs: for example, an.mp4 file can contain H.264 or H.265 video.
Yes. AI models trained on or processing heavily compressed footage may produce lower-quality outputs because compression artifacts in the source material introduce noise and ambiguity that the model must work around. Higher-bitrate source material generally yields better results.
ProRes (on macOS) and DNxHD/DNxHR (on Windows) are widely used intermediate codecs for editing because they are high-quality, intra-frame formats that hold up well to multiple rounds of processing without significant quality loss.
Each time lossy-compressed footage is decoded, edited, and re-encoded, new compression decisions are made and additional data is discarded. This generational quality loss accumulates across multiple re-encodes, which is why editors work with high-quality intermediate codecs and only compress to delivery formats at the final stage.
Bitrate is the amount of data used per second of video, measured in megabits per second (Mbps). Higher bitrate means less aggressive compression and better quality; lower bitrate means more aggressive compression with greater potential for artifacts.
H.265 (HEVC) achieves roughly double the compression efficiency of H.264, meaning it can produce equivalent visual quality at approximately half the file size. However, it requires more processing power to encode and decode, and is not universally supported across older devices and platforms.
To a significant degree, yes. AI-based restoration and enhancement tools such as Topaz Video AI can detect and reduce compression artifacts, effectively recovering apparent quality from degraded footage. However, truly lost data cannot be fully recovered: these tools estimate and reconstruct missing detail rather than restoring it literally.
