I'm using mencoder/libav to capture an interlaced source to mjpeg.
Now, my hunch is that mjpeg is a codec that handles only full-frames, progressive video - because I assume it compresses single (full-)frames as jpeg. How would it handle the "half frames" of interlaced video??
Am I wrong? Did the mjpeg developers take that into account and figured out a way to properly keep the interlaced aspect of interlaced video when stored as mjpeg?
Or is my hunch correct and I am unknowingly "sort of" de-interlacing my material?
To encourage answers, my thoughts on this are:
Ok, Motion-JPEG compresses each frame as a jpg, more or less. Right? Then, how is interlaced material handled? Let's assume it vertically merges/compresses each half-frame from the "lined" state into what looks like a non-interlaced image and then compresses that as a jpeg. Well, then, the (possibly heavy) smearing/blocking/blurring of normal jpeg would kick in and the lines would bleed into each other. On reassembly of the interlaced picture-stream, separating the lines again, this blur then would end up in the final re-interlaced video. Resulting in a loss of quality a native, interlaced-capable codec wouldn't have introduced... At least that's my hunch, as I don't know anything about the internal workings of mjpeg (what probably became obvious above). Prove me wrong.
More encouragement: eye-candy:
This is a screenshot-section, from a mjpeg encoded interlaced video. And even though this image itself is again jpeg encoded (at q=85%), the "bleeding" of lines into each other is not that bad. Could be, jpeg resolution at average compression ratios is good enough to resolve the "lines", and I am worrying about nonsense. Still, technical details anyone?