Compression and metadata: what is different?
Separate a container edit from a new video encode.
3 min read · Updated November 3, 2026For anyone comparing an editor export, a prepared file and a platform playback stream.
Locate the change before judging it
An export is not one indivisible operation. An editor can crop or resize the image, encode its frames, encode audio, then package those tracks with metadata. A later tool might repeat all of those operations or only rewrite the container.
That distinction matters when a product says it “optimizes” video. Ask which operation it performs and what evidence confirms it. A new filename or a reduced file size does not answer either question.
- 01Pixels and sound
- 02Encoded tracks
- 03Container and metadata
- 04Platform renditions
Four operations that often get confused
Use these names when comparing workflows. A single export can contain several operations, so a table of final dimensions cannot identify the complete processing history.
| Operation | What changes | Useful evidence |
|---|---|---|
| Re-encode | Picture or sound is decoded and encoded again. | Encoding settings and a media comparison; dimensions alone are insufficient. |
| Resize or crop | Pixel dimensions or framing change. | Source/output dimensions and a visual framing check. |
| Remux / stream copy | Encoded tracks are repackaged. | Matching encoded media plus compatible timing and stream configuration. |
| Metadata edit | Selected descriptions or container fields change. | A before/after field inspection and checks that media was retained. |
Why a changed file hash is expected
A whole-file hash covers every byte, including the container index and metadata. Editing even one of those bytes changes that hash. Two different whole-file hashes therefore do not, by themselves, prove a second video encode.
The reverse claim needs care too: visually similar frames cannot prove that compressed samples are identical. To test preservation, compare the relevant encoded media and timing while keeping source and output identities fixed. The preservation guide explains that narrower test.
Bitrate is a rate, not a quality score
At a given duration, more bytes generally mean a larger average data rate. That is arithmetic, not proof that the extra bytes contain useful detail. Motion, noise, codec, encoder settings and previous compression all affect the result.
A bitrate reported in metadata may describe a particular track. A number calculated from whole-file size and duration includes other tracks and container overhead. Keep those labels separate when comparing a local export with a published playback stream.
Metadata removal cannot restore missing detail
Once an earlier export has blurred fine texture, removing an editor tag cannot reconstruct the original texture. Similarly, adding a larger resolution label cannot create pixels that were never captured.
Fulres changes supported container structure and selected metadata, then verifies its output. The current default is not a blanket promise to remove every editor-related field. It does not apply a sharpening filter, upscale the image or tone-map color.
Make a controlled comparison
Keep one source and change one variable. If you export twice with different resolution, codec and bitrate at the same time, it becomes difficult to explain the difference. Include a moving section and small text rather than comparing only an easy still frame.
- Keep the original source and one clearly named export for each test.
- Write down dimensions, frame rate, codec and the basis of any bitrate figure.
- Compare matching moments at the same display size.
- Distinguish the prepared file from each published rendition.
- State exactly what your comparison established and what remains unknown.