This relates to Mark's question from a bit ago. Why is it that the 24P footage we are seeing looks so much more jumpy than 24 (P) from film. Considering slow shutter speeds such as 48 from Matt's examples and the smae frame rate / playback rate, the results are looking far different to me. Any help?
HD video shot at 24fps and 180 degree shutter with a 3:2 pull down applied should look exactly the same as film shot at the same rate viewed on video after telecine. Are you looking at the HD footage with pull down (from Cine Alta)?
I don't believe the pulldown has been applied in any of the posted examples. For instance, the ones in Matt's tests:
Like I said, if you're viewing on a monitor without 3:2 pulldown added, then the motion is not accurate and it might studder, jutter or whatever you describe as unpleasant. Only if filmed out and viewed on a motion picture projector at 24fps, where the dual bladed shutter and the effect of persistence of vision in the dark theater would effectively smooth out the motion could you view the proper effect. This is a particular flaw in the method Sony uses in the Cine Alta and the "native" Panasonic formats, such that "processing" must be applied in playback for proper motion display. Of course the alternative of saving storage space has many advantages too.
Very helpful explanation. Thanks John!
Since we're on the subject, I was just curious how or what is done to ensure smooth motion when showing a film on television that was originally destined for theaters.
[john sharaf] "This is a particular flaw in the method Sony uses in the Cine Alta and the "native" Panasonic formats, such that "processing" must be applied in playback for proper motion display."
Can you describe the flaw and the processing to fix it?
The problem is self evident when you view the output of the Cine Alta at 24p on a HD CRT; it flickers like crazy. This is even though the monitor doubles the frequency rate to try and smooth it out. On LCD's it's not apparent because it has an even higher refresh rate.
The solution of course is tho add the processing of a 3:2 pull down; this doubles and triples up successive frames and smooths out the motion. This processing is added in the playback deck and takes place when you make downconverts or dubs. In the Panasonic the logic of 24pN is the same as Sony's to use less storage (in this case precious P2 capacity) and then extrapolate the frames out later in post, or continue to save storage by editing in a 24p timeline. Remember that 24 is only 40% of 60 so that's the amount of storage you're saving - that's a significant amount. In the end of course, if you're not printing out to a 24p use (like film out or some DVD pressings) at that point you add the "processing" to make it compatible with 59.94 type CRTs.
I hope this helps. Perhaps "flaw" is the wrong word, but I hope you can now see how the processing takes place.
John, I understand that, I've been a video engineer. What I haven't figured out is why 23.98 video with 1/48 shutter does NOT look like 24fps film. I dislike 23.98fps video and can't fathom why some people say it looks "like film." (almost) Matching frame rate and shutter angle doesn't really seem to be a match in motion.
In my opinion video with pull down still doesn't look like film with pull down. I can't think of a technical reason why though and was hopping you had.
I do know that many people shoot 23.98p video because it's easy to transfer to 25fps or 29.97fps.
30p can have issues going to 25fps given the frame math an the lack of interlaced fields to "split."
I guess this is what makes horse racing!
I started in film, shooting quite a bit of 16mm that was telecined for broadcast on television newsmagazines, so I feel like I have some point of reference, and I feel that HD video shot at 24p does look quite a bit like film when displayed on video screens. The manufacturers have done a good job in imitating the frame rate (exactly) and some include the pulldown (and even variants like "advanced pulldown'), they imitate the gamma curve with various "film looks", "hyper gamma" and "cine gamma" settings and finally the video "noise" approximates the grain inherent in film, especially in the almost grainless look of the most modern stocks.
I don't mean to be a contrarian here, but I have to respectfully disagree with your assessment of the success of 24fps HD in looking like film. Technically the only subtle difference is the action of the shutter itself; in the film camera it moves either side to side or top to bottom, whereas in the video camera it captures the frame/field in one fell swoop. I don't think that this subtle difference is perceptible, especially enough to distinguish the look as not being filmlike.
I think short of a side by side comparison on identical video displays of the same shot on film and HD you could not really state with much authority that they are different, and in fact, were you to do such a test, I think you would find, once and for all, that you were mistaken.
That is not to say that 24p is the be-all and end-all frame rate. You are correct in saying that it is often used because of the ease in which it converts to 25 (for European release) and 30 (for NTSC use). In most cases, the 30fps frame rate is actually a better choice, in that it remains progressive, which all the inherent motion and sharpness benefits over interlaced, and it looks more "filmic" and less "live". If the bean counters had not had such an influence as the have always had, this might have been the standard frame rate from the beginning of sound film, rather than 24; but financial considerations have and will always have an influence over technical standards.
Finally, if for aesthetic or creative reasons you don't like the look of video at 24fps, you still have the alternative of shooting on film or shooting video at the 30 frame rate (if you prefer that).
[john sharaf] "Like I said, if you're viewing on a monitor without 3:2 pulldown added, then the motion is not accurate and it might studder, jutter or whatever you describe as unpleasant."
23.98p without pull down on a monitor that can play 23.98p . . . looks like 23.98p.
Computer monitor have no problem with 23.98p native. Many HDTVs can play 23.98p (and some really do a bad job of it).
Many editors don't understand the concept behind the cadence (both "standard" and advanced) when pull down is added and that can certainly impact the look if one is editing 23.98 with pull down added in 29.97 time line.
It's interesting that I'm not the only one who sees motion issues with 23.98 and video. I just viewed Doug Jensen's EX1 training video and while he doesn't verbalize it well, he does state clearly that he prefers 30p to 24p and shows examples of the same scene shot both ways. The scenes are displayed in standard def on a 29.97 TV set of course. I don't recall if he mentioned the shutter angle (or speed) but nonetheless I'm sure he and many other very experienced DPs (which I don't claim to be) see what I see.
In short, I can't nail why but 24fps film with proper pull down played 29.97 interlace looks different that video 23.98 with pull down played on 29.97 interlace. My eyes see it and I do know I'm not the only one who does.
There are also many means to add that pull down and maybe that's part of the difference. "Video" shot at 23.98 and played at 24fps in a movie theater is another story. MANY films are shooting with digital cameras at 23.98 and shown in theaters at 24fps and they look fine.
OOOOOOHHHHHHHH.....my head hurts now
When you telecine/transfer a 35mm film to video and the result is for NTSC frame rate, it gets transfered at 23.976 fps (23.98 fps).
It also is captured as RGB 4:4:4 and not Y, R-Y, B-Y (YUV) 4:2:2.
The difference between 4:4:4 and 4:2:2 makes a difference for motion representation and color. This difference in capture color space is what many point out as a strange difference between telecined 23.98fps and camcorder captured 23.98fps video.
What randy Strome saw, has been noticed since 2000. In fact, I read that the first person in SONY America to point it out publicly was fired. Not that it was a secret, by any means, but it was an issue that only engineers could point out and that most if not all viewers (non engineers) wouldn't notice or care about.
An interlaced HD video transfered to film and the result seen on a movie theater demonstrates the opposite. It looked like interlace HD video (movement and color space.) It did not look like film captured images.
On the other hand...
I have seen the last Star Wars episodes on film and on TV. They both look like film-captured images. The difference is that the HD data was in fact 4:4:4 captured and never went to 4:2:2 color space until broadcast masters where made.
You would probably not guess that color space could affect the representation of movement, but it does.
Now you can understand why the Sony F-23 4:4:4 camera head is so expensive (RGB 4:4:4 capture, F-11 at 2000 Lux, 98% reflectance.)
Please note that regular HD cameras and for that matter SD cameras that capture at 23.98 also portray movement correctly as film captured images would. The limitation here again is the difference between native all the way 4:4:4 color and native 4:2:2 color. You could say that interlace stuff definitely does not look at all times like film, when shot at 24fps/23.98fps, but the same is true of all 4:2:2 video (in the eye of a video engineer.)
Does this mean that 4:2:2 images captured with an HD camcorder should not be used for filmout or TV? The answer is: they should be used. Since the only people who will ever notice the difference between motion portrayal in 4:4:4 and 4:2:2 color space will probably only be... a video engineer.
[Paul Thurston] "Does this mean that 4:2:2 images captured with an HD camcorder should not be used for filmout or TV? The answer is: they should be used. Since the only people who will ever notice the difference between motion portrayal in 4:4:4 and 4:2:2 color space will probably only be... a video engineer."
This has been a superb thread and your explanation puts the capper on it. Thanks!!
Because of the entire thread discussion, I think that I now have an understanding of why I, like Craig and others have expressed, tend to feel that 30p looks in some ways more filmlike than 24p when run through a complete YUV, 4:2:2 production/post-production chain. Or maybe it doesn't look more filmlike, because there are other things going on, of course, but it looks better to me, if it is destined to be video all the way. As John points out 24p in the film projection theater, whether coming from film or video, benefits from multi-blade projection which tames the judder/flicker of 24p.
And it tames the judder/flicker without resorting to artificial cadence pull-down. For video that won't ever be optically projected, 30p tames the judder/flicker pretty good, again without pull-down.
Of course what I'm talkin' about applies to NTSC. How Mark deals with the flicker of 25p PAL, I dunno? I suspect that a considerably higher percentage of folks can see flicker at 25p than at 30p.
Thanks to all,
More excellent info!
"The difference between 4:4:4 and 4:2:2 makes a difference for motion representation and color. This difference in capture color space is what many point out as a strange difference between telecined 23.98fps and camcorder captured 23.98fps video."
I can certainly see how capture color space could impact the "film look", but am still unclear how it would impact motion representation. Could you clarify?
The capture of the telecine is dependent on what recording format you use. It is inaccurate and misleading to say that "It is also captured at RGB 4:4:4 and not..."
I fail to be convinced that color space affects the temporal quality of the 24p film look. What if it was in B&W? The motion would be the same!
So to sum it up.... If shooting for video, shoot 30. If shooting for a filmout, shoot 24. If you shot 24 for film but need to show it on television or DVD, apply 3:2 pulldown.
OTOH.... If you really think 24 looks better and are shooting for video, go ahead and shoot 24 and add pulldown in post.
Did I get it??
Higher Ground Media
[Mick Haensler] "So to sum it up.... If shooting for video, shoot 30. If shooting for a filmout, shoot 24."
That's how I grok it. Of course, if the temporal resolution is more important than the spatial resolution, such as with dance and sports, shoot 1080-60i or 720-60p. The EX does it all.
[Mick Haensler] "OTOH.... If you really think 24 looks better and are shooting for video, go ahead and shoot 24 and add pulldown in post."
Yeah, but you don't see pulldown in the optical projection theatre. It's an artificial adaptation to NTSC electronic display and I believe does effect spatial resolution to some degree, which is why I prefer 30p production for electronic display. The other side of the coin is that we've gotten used to seeing movies on TV with pulldown, so we accept some degradation due to the pulldown that we wouldn't see in the optical projection theatre.
So to each his own.
[Mick Haensler] "Did I get it??"
Ya gots it for me, but perhaps not for you or the next person. (g)
Shoulder-High Eye Productions
CreativeCOW Forum Host for Discreet edit*
Not so fast!
If you want it to look like film (on TV like displays) shoot 24p. The 3:2 pulldown will be added so it doesn't flicker. Same speed if you intend to film out (unless your in a 50Hz country where projectors run at 25 fps, then use that rate).
For video, you can still shoot 24p (again if you want it to look like film with pulldown), or shoot 30p for smooth motion and some-what film look (progressive) or shoot 60 for live look.
It's really great to have all these choices, you just must consider what your final product will be, how it will be seen and educate your clients about the various advantages and pitfalls.
I agree with the OP, and I'm not convinced the 4:4:4 colorspace is to blame. I don't accept the projector explanation either. When comparing a DVD movie on a progressive screen to video shot 24p (either 720 or 1080) with the EX1, to me there is a huge difference in motion stutter and flicker. This has nothing to do with pulldown or projectors, as both sources are being displayed in native 24p. For some reason, the film source appears to have smooth motion, even in 24p, while the EX1 footage simply looks jittery. I'm glad I'm not the only one to notice this. I would like to figure this out because there are many advantages to 24p in regards to making a DVD, but so far the results I have seen simply don't cut it.