How 'jittery' should HDTV shot on film be?

IMO, the type of display and/or video processor you're using would play an important role in 24p (film or tape) judder. 24p-based material is delivered over 1080i sources by using 2-3 pulldown--adding TV fields (1/60-sec half-frames) and changing the speed slightly so 24 frames per second matches the 1080/60i (30i) that stations and other sources use. Displays/processors with good reverse 2-3 pulldown can weave each 24p frame back together from only the original TV fields (half-frames). Then, if they're displayed at even multiples of 24 fps, such as 48, 72, 96, 120 fps, judder (from displaying 'extra' TV fields)shouldn't be visible. Not many displays have such 2-3 reversal and even-24p-multiple display yet. Instead, most display 24p-based material at 1080/60p (or a scaled resolution other than 1080) which still has extra TV fields.


But 1080/60p is correct for non-24-based material from TV cameras. That includes live 1080/60i or typical travelogues/documentaries video taped at 1080/60i (30i). With these sources, which match the broadcast rate, 1080i's two 1/60-sec TV fields per 1/30-sec TV frame can just be deinterlaced (30 fps) and displayed twice (60p or 60 fps, which differs significantly from capturing images originally at 1080/60p). That's for fixed-pixel displays. For 1080i CRT-based displays operating in interlace mode, 1080i material is displayed as 1/60-sec TV fields, as it's delivered, and the eyes/brain merge two fields into 1/30-sec TV frames. The fast 1/60-sec refresh rate conveys rapid motion well, and that refresh rate minimizes flicker.


My 1080i CRT RPTV displays the extra 2-3 pulldown TV fields of 24p-based programming since it displays all HD at 1080/60i. Judder seems to vary, although suspect it would take direct A-B comparisons with a display using 2-3 pulldown removal (and 24p-even-multiple display) to confirm how much judder is present. One program series, Discovery's "Sunrise Earth" is video taped at 24p and seems moderately blurred (mostly static scenes) with my 1080/60i CRT display; grass blades or twigs, distant rocks, etc., aren't very crisp looking, although the color and scenes are nevertheless great looking.


That's in comparison with programming taped at 1080/60i, which appears very crisp since it's recorded, broadcast, and viewed at 1/60-sec TV fields per second on my screen. Motion also appears smoother with such programming captured each 1/60-sec, sometimes labeled 'wow' HD--providing inadequate bit-rate delivery and excessive filtering from multicasting or other HD processing doesn't rob resolution and cause blocking artifacts.


Movies captured at 24 fps (~1/48-sec exposure) have widely varying judder and blurriness displayed at 1080/60i with my CRT RPTV. Since "Sunrise" scenes are nearly static, it seems necessary to differentiate between judder caused by the extra TV fields of 2-3 pulldown, and the added judder from movement. (I've also noticed stuttering-type motion artifacts by a malfunctioning cable-STB MPEG-2 decoder, and some fixed-pixel displays might be decoding/deinterlacing 1080i poorly.) But lots of additional elements have to be factored in with films. The quality of prints optically telecined to tape/disks varies widely. Cinematographers deliberately filter their cameras for dramas and selectively focus (blurring backgrounds) to concentrate attention. Motion and camera movements may be restricted for the 24-fps capture, or 'overcranked' at higher fps when less blurring of rapid motion is desired or for slow-motion sequences. TV productions can risk using original negatives for telecines, not prints, providing crisper HD images, since rapidly-made TV production film segments aren't as 'precious' as major film feature segments.


So, to pinpoint judder-type artifacts, either a descriptive comparison in these forums with what others have seen might be needed, or home comparisons with other methods of display: fixed-pixel at 60p with/without 2-3 pulldown reversal, or, say, 72 fps with 2-3 reversal. -- John

CSI and other film-based series, or feature movies, don't have noticeable motion judder on my 1080i CRT RPTV. Side to side camera movements, as you mentioned, undergo blurring as expected from 24p capture but not really noticeable jerky movements beyond 24p capture (compared to 1080/60i capture). But can't say without A-B comparisons under the conditions outlined earlier, how much better images might be with 2-3 reverse telecine and 24p-even-multiple display. A further complication with 2-3 reverse telecine is what kind of deinterlacing is used. So-called 540p-bobbing could halve vertical resolution, and 'resolution pumping' might blur details during slight movements within images, contrasted with crisp static images formed with simple weaving of TV fields. One of the latest pixel-by-pixel deinterlacers could avoid these problems. -- John

Quote:
That would be news to me and perhaps others here. Maybe Glimmie knows. My understanding was that films were telecined to master tapes (or disk drives) at 1080/24p. For broadcast delivery and all the standardized gear in the broadcast chain, this 1080/24p undergoes 2-3 pulldown to achieve the 1080/60i (30i) broadcast rate; (or different pulldown for 720p). In homes this undergoes the varied processing/display I outlined initially.


It's intriguing to me whether the blurring I see with Sunrise Earth's somewhat fuzzy 24p-taped programming is caused by 1080/60i CRT display, and whether that blurriness might vanish with 2-3 pulldown reversal and 24p-even-multiple frame display as outlined earlier. That might imply filmed productions and other 24PsF-(segmented-frame)-taped productions would also be crisper, although comparisons really become complicated because of the factors I outlined that tend to soften filmed images.


canada_habs2004,

Since you're using a computer, perhaps different processing and display could achieve 2-3 pulldown reversal and display at other than 60 Hz. The folks in the home theater PC forum might suggest some approaches. If it's possible, it would be interesting to hear whether different computer processing eliminates the judder you're seeing. Video processors (see that forum) perform such variable conversions, too. -- John

Quote:
I wonder if anyone has gone back and compared Sunrise Earth from its first airings. The bitrates on DHDT airings in the first 60-90 days were MUCH HIGHER and DHDT did not insert the logo on the program during that time.


I'll be the first to admit I have not, but now I am slightly curious to know if that is per chance the reason for the blurring.