Concretely, a horizontal or vertical flip is applied as a filter, which needs a re-encode because the coded frames themselves change; the audio stream is copied through untouched. You choose the axis — horizontal produces a mirror image, vertical produces an upside-down reflection — and the whole file is transformed on it.
What is the difference between flipping and rotating?
+
A flip reverses the content across an axis and produces a mirror image; a rotation turns it without reversing anything. Text is the giveaway: it reads backwards after a flip and stays readable after a rotation.
Does the source format matter when flipping?
+
It does. 3GP is the cut-down MP4 profile built for early mobile networks, usually very low resolution with AMR audio. That is what decides whether the mirror can be done by rewriting the file or requires decoding it first.
Anything specific to 3GP to be aware of?
+
Yes — 3GP files are usually very low resolution with AMR speech audio, so there is little detail left to protect and a lot of size to reclaim. Worth a moment before you flip something you cannot easily recreate.
When would I actually want to flip rather than rotate?
+
Correcting a scan or a webcam capture that came out mirrored, making a symmetrical pair from a single asset, or matching the direction a subject faces to a layout. If the content contains text, logos or numbers, check the result — those mirror too.
Can I flip several files at once?
+
Yes. The same axis is applied to every file in the batch, which is how you fix a whole folder of mirrored captures in one pass instead of one at a time.
