Originally Posted by John Mason
AIUI, sets with constant 540p bobbing (see Merson's three earlier articles
) use each 1/60-sec 540-line TV field from 1080i to create full frames. Newer test DVDs or a standard SMPTE test pattern detect this with flashing white and black screens as each TV field is displayed.
The two 540-line interlaced half-frames are supposed to be shown 1/60-sec apart on interlaced CRT displays, or digitally deinterlaced into full frames for fixed-pixel displays. With my 1080i CRT RPTV, my eyes merge the two 1/60-sec TV fields into 1/30-sec frames. With most fixed-pixel displays a buffer memory helps create 1/60-sec frames, preferably combining both fields/frame rather than creating frames from only 540-line half-frames.
The mid-90's ATSC-approval test was cited earlier above
. Test results (see simplified table
) showed that a static 1080i test pattern had 800 lines of effective vertical resolution (resolvable lines), while a dynamic (5-rpm) pattern only 400 lines. Vertical resolution is the number of resolvable B&W horizontal lines from screen top to bottom. This drop in vertical resolution with motion varies directly with the rate of motion, so movement less than ~5 rpm from a rotating test pattern shouldn't cause as much loss of vertical resolution. Also, the vertical resolution loss from movement should only take place where movement occurs. 540p bobbing, however, could diminish vertical resolution over the whole screen.
In the "2.1 Resolution" introduction of the ATSC report (linked above), the video experts wrote:
Again, that's where motion is taking place within the image, and the resolution drop varies with the rate of motion.
Also, as outlined originally above (yesterday), it's possible to use a non-sampled computer or pattern-generator test images and measure the full 1920X1080 resolution of many displays. Test patterns or actual programming sampled at ~74 MHz are filtered to reconstruct images from the samples, resulting in both vertical and horizontal resolution loss. Came across an interesting diagram, Fig. A3,
, from 2006 European HD studies, that compares the effective resolutions (resolvable details) of both 1080i and 720p after the Kell factor (~0.7 X line count) and anti-twitter filtering is factored in. Obviously such diagrams involving maximum resolutions expressed as lines correspond to equivalent details in scenes, and maximum resolvable details can be far less--if a film or video camera is heavily filtered, for example, or from many other limitations along the delivery chain. -- John