Originally Posted by van Gageldonk
I've been thinking about the interlace of our camerafootage: 50i for 2 eyes = 25i for 1 eye. So it's interlaced. Actually our footage is half HD.
For my personal use I used to export to halfSBS. (show with Playstation3 and PSmediaplayer).
Now I'm thinking: if I use half TopBottom it should be sharper than halfSBS. Creating halfSBS drops half the width and is lossy. Creating halfTopBottom drops the interlaced lines, thus keeping the right lines and dropping nothing.
What are your thoughts about this?
Unfortunately half height top-bottom does discard valuable video information for those parts of the scene that were static or moving only slightly. I explain this below.
People commonly assume that a video camera setting of 1080i60 (or for those in PAL DVD regions, 1080i50) gives inferior vertical resolution to a setting of 1080p24. However the analysis depends on how quickly the subject (or the camera) is moving. 24fps is a very slow frame rate, with a history stretching back to the late 1920s. During a pan, the image can become quite blurred. A typical exposure time of 1/48th second blurs the content but reduces the jittery stroboscopic effect of the pan, a good compromise for projection at the cinema. Hollywood movies require great care to avoid camerawork that would expose this limitation of frame rate.
60i is far better for pans or fast movement of the subject than 24p. A typical exposure time of 1/120th second keeps the successive half vertical resolution frames crisp and the frame rate is high enough to hide jitter between frames. It is a good frame rate for sport or for general unplanned camerawork. Despite the limited vertical resolution of 60i during movement, the video can for some scenes momentarily (and paradoxically) provide more
visible detail (after adaptive deinterlacing) than blurred Full HD frames. A lot depends on the speed of movement (of the subject, and panning/zoom of the camera). At sedate rates of change, 24p will provide a slight resolution advantage over 60i as the deinterlacing cannot quite reconstruct the missing information and introduces a degree of (deinterlacing) blur.
60i for a perfectly still scene, and using no panning or zooming, is equivalent to 24p in its resolution. That is because the deinterlacer in the display device (or editing device with adaptive deinterlacing) is able to combine each pair of interlaced captures into a set of progressive 1920x1080 frames. For example, one second's worth of 60i (containing 60 still captures that weave together to yield 30 full resolution still captures) can be used as source material for creating 24 progressive still captures in the editor.
1080i60 for a partially still, partially moving, scene provides Full HD resolution for the still parts, and a non-blurry high temporal resolution, but limited vertical resolution image, for the moving parts.
Some people prefer half-height top and bottom for use with passive Full HD LCD displays as such displays can only use 540 vertical lines for each eye anyway. That is an exception to video authoring practice.
The normal video authoring practice is to use half-width side by side when transferring 3D to a Full HD 2D video format. There is no reason not to follow that practice simply because the 3D capture is at 60i rather than 24p. I say that because, resolution-wise, 60i is
Edited by MLXXX - 12/15/13 at 6:06am
- for a static scene, as good as 24p [assuming sufficient light for adequate exposure]; and,
- for a slowly moving scene, almost as good as 24p, depending on the quality of the deinterlacing.