I’ve run into this issue so many times, practically on a weekly basis, so once and for all, 1080i / 60i is NOT 60fps. That is 60 frames per second. Never was, never will be, it’s just not going to happen. Let’s take a look shall we?
First off, the “i” in the number. 1080i / 60. The “i” stands for Interlaced.
With 720p or 1080p the “p” stands for Progressive.
Those two letters make all the difference in the world and that letter “i” is why 1080i / 60 is not 60 frames per second.
Progressive Formats:
A progressive format is made up of a series of full frames. Each Frame is a single image that covers the entire screen. For example, the following is a single Frame of Progressive Video.
There’s the entire image in one frame, it’s not a split image as you will see shortly. So when you see a Progressive frame rate, the second number refers to the amount of FRAMES it takes to make up one second of video. 720p / 60 for example means there are 60 frames for every second of video. 60 images like the one above make up one full second of video.
Interlaced Formats:
An interlaced format is made up of two fields per frame of video. Each field is half of the single image. When combined, the two fields make up one frame of video that covers the entire screen. For example, here is the first field of of a single frame of video.
It’s exaggerated for illustration purposes. You can see how there’s half an image. And here’s field two:
All the black lines that were in Field 1 are filled in with Field 2. So there’s two fields of the image that combine to form one frame like so:
Again, this is exaggerated as you would NOT see the lines in a good interlaced picture.
So 1080i / 60 stands for 60 FIELDS per second, divide by 2 gives us 30 FRAMES per second, or as we all know in drop frame timecode, 29.97 FRAMES per second.
Therefore, 1080i / 60 has a framerate of 29.97.
Why is it called 1080i / 60? I have ABSOLUTELY NO IDEA. If anyone has the name and / or phone number of the engineer and / or marketing / sales person at the camera company who coined that format, please pass it along as I would LOVE to have a little chat with them. Calling the format 1080 / 60i has only created confusion where there shouldn’t be any.
I hope this clears everything up.
I think you’re actually wrong about this.
It’s called 1080/60i because there are 60 points in time that an image is captured. Yes, it’s only half a frame, but a single field is captured every 1/60 of a second. In an interlaced format the unit of measurement is a field, so it’s 60 fields per second, and 1080/60i is not only a perfectly appropriate name, it’s the only appropriate name.
If you *actually* put the upper and lower fields together, they wouldn’t equal one frame, as your illustration suggests, because they weren’t captured at the same time. Instead, the combined image would look just a tiny bit “slatted,” because the even lines were captured 1/60 of a second after the odd lines. This is usually only easy to notice on fast-moving footage.
So there really is movement on screen every 60th of a second, at the cost of half of the functional resolution of the image. Most people don’t notice the reduction of in line-information, but they do notice the presence of twice as much time-information. That’s why soap operas and other “video-y” footage has always had that “smooth” look, which newer 60hz and 120hz TVs have created artificially by interpolating extra frames.
And while each field does stay on screen for a 30th of a second, they continually overlap, so a new field in drawn on the screen every 60th of a second. There is no such thing as a single “proper” frame in this case. Any field can be combined equally well into a faux-frame with either the field before or after it (unless it’s a across a cut, of course).
This is an artifact of the early days of broadcast video, where there wasn’t enough bandwidth to transmit (or record) a full frame for every cycle of power (60hz), and it would be a bigger sacrifice to transmit a lower-resolution (but full) frame every 60th of a second (which they were locked to because of the way the power grid operated, in the UK it’s 50hz, thus the 50i/25p of British TV).
Now… these days, many cameras will capture a single full frame in memory, and then write it as two fields to the recording device, so that when an NLE or media player decodes it, the two fields *do* equal one frame, since both fields were captured at the same point in time. This is also how interlaced DV tapes (and true DVCPRO-HD) does 24p, by recording full frames into a buffer and then writing it to tape in fields in a reverse pull-down pattern, for the NLE to decode later.
But the format itself doesn’t know about any of this workaround tomfoolery, and there is still a lot of content that is recorded in true interlaced format with overlapping fields, so it only makes sense to call it 1080/60i, because there are always 60 interlaced fields in the format, regardless of how the content is encoded to it.
Computer playback often ignores interlaced formats and displays it as full frames by combining the upper and lower field and calling them, say, “Frame 5U” and “Frame 5L” instead of “Field 10” and “Field 11” (careful when interpreting field order!), so you edit and view it in a 30 frame timeline and use 30 frame terminology, but calling it, say, 1080/30i would be the more confusing name, because the format does not use frames to define itself, so the 30 doesn’t mean anything. It’s more workaround tomfoolery by your computer.
The presence of the “i” should be enough for editors, and other people who need to know, to understand what it means.
(But if you are feeling frisky, you actually can edit true 60i footage in a 60-frame timeline, you just need to artificially interpolate the remaining field so that each field’s missing half is filled in to become a full frame. Premiere Pro does this beautifully, by the way.
An old DV trick is to shoot slow-mo footage in 60i, when you don’t have a variable-frame camera, and edit it in a 24p timeline. It worked pretty well, actually.)
I hope I didn’t come off too “professorial” or something, I love this blog and find your posts very informative and insightful. Please correct any mistakes or fill in any blanks I’ve left here.
I do tend to go on and on about these things, hopefully I do it here as a service to your readers, since I find that too many professional editors (*cough* FCP users *cough*) don’t know some of this important minutiae because they never bother to “look behind the preset.”
Even those of us who mostly publish to our clients’ YouTube accounts need to know this stuff.
Mark, I welcome input as long as it helps further along the discussion. I’m glad this blog has been useful to you and I do and appreciate the kind words.
Mark, good points. But if this whole “60 points in time an image is captured” was true, then why is standard definition for NTSC not called 525i / 60? SD has the same 60 fields per second to make up the 29.97 frame rate.
The whole “60” was not added until we moved to HD and my feeling is someone wanted to make it sound like 720p / 60, so instead of calling it the true 1080i / 29.97 (which both AJA Video Systems and BlackMagic Design call the frame rate for capture / playback) they decided to add the “60” as a marketing gimmic.
Not to one of your points:
Yes, two fields of video are drawn together to form one frame. You are watching one full frame of video that has been drawn together as my illustration shows.
My discussion has zero to do with computer playback, this has everything to do with the true Frame Rate of 1080i / 60. The number means everything to your workflow. 1080i / 29.97 would be the correct term for this frame rate and much more accurate than 1080i / 60.
You mean 60p. We have done that many MANY times using Panasonic HD cameras. Shoot 720p / 60, drop that into a 24p timeline and voila, super smooth slo mo. But this doesn’t work with “60i” unless you want to render out all the extra frames and even there, you’re not going to get near as smooth a slo mo effect as you do with progressive footage.
So my point that 1080i / 60 is really a 1080i / 29.97 frame rate is completely accurate.
I’m a little surprised at your insistence here, and I’m concerned about the spreading of misinformation on this topic. Look up the Wikipedia article on interlaced video, it says it right there at the top: “Since the interlaced signal contains the two fields of a video frame shot at two different times, it enhances motion perception to the viewer and reduces flicker by taking advantage of the persistence of vision effect.”
Actually, it is! Again from Wikipedia, the article for “480i”: “There are several conventions for written shorthands for the combination of resolution and rate: 480i60, 480i/60 (EBU/SMPTE) and 480/60i.”
That naming convention was not invented for HD formats, it’s been around forever.
No, I mean that you actually can shoot footage in 60i and place a single field on each frame in a 24fps timeline. That’s why I called it a trick. The image is a little fuzzier because you are interpolating the other half of each frame (in a Premiere timeline, right-click>Deinterlace), but it works, and there really are 60 unique points of time captured in 60i video. Obviously you’re better off shooting 60p, but back in the wilderness of 2002 on an old DV camera, you didn’t usually have that option (Check out this old article from 2003 on the trick: http://rarevision.com/v1/articles/slow_motion.php).
No, it would not be correct. Calling it 1080i/29.97 would be saying that there are 29.97 *fields* per second, and that the effective framerate is 14.985!
Again, with interlaced video, there is only “fieldrate.” “Framerate” is a necessary illusion of the editor. If your timeline/workflow is interlaced, your outputted video actually has 60 unique temporal points every second, even if your “effective framerate” is 29.97.
Misinformation on this topic is precisely why I wrote the blog post. Too many people are confusing frames with fields as you are doing in your response. The Frame Rate of 1080i / 60 is 29.97. You can easily verify this yourself by simply creating a new 1080i / 60 timeline in the NLE of your choice. Look at the Frame Rate in that timeline. It will show 29.97. Move your playhead frame by frame along the timeline. You will have 30 frames in each second. Not 60. This is true in Avid, Final Cut Pro and Adobe Premiere Pro
I have been editing in broadcast since 1990 and your response is the first time I’ve ever heard anyone refer to standard definition as 480i / 60. It’s 30 frames a second. It takes two fields to make up a single frame. Sounds like someone was trying to justify calling HD “60i” instead of 29.97. And quite honestly, Wikipedia is a good place to start your research, but not the end all.
That’s not even remotely the same as what I am discussing here. You are simply taking the footage and manipulating it in Post. You’re taking half of the screen information and making a full frame out of the field by adding interpoloated information to in effect create 60 frames out of the fields. You can do this with After Effects and Twixtor all day long and have something that looks halfway decent. But the frame rate of the original footage is still 29.97. Just like I can take 29.97 and make it “look like film” by rendering it at 23.98 in After Effects. So After Effects throws away a few frames each second and I have “film.”
You can manipulate the footage all day long in Post, but that doesn’t change the frame rate as shot originally in the camera. 1080i / 60 has a frame rate of 29.97
If you really want to do your effect correctly, you shoot 60p so you have 60 true frames of information giving you tremendous slo-mo. I can create incredibly realistic looking “film” from 29.97 video using After Effects, but that doesn’t change the fact that the frame rate of the original footage is 29.97.
Both AJA and BlackMagic refer to 1080i in NTSC as 1080i / 29.97 in their presets because this is the accurate representation of the interlaced frame rate. AJA always has since they started supported HD in FCP.
The number always refers to Frames when it comes to editing timebases. In editing we do NOT edit with fields, we edit with frames. We can manipulate fields to make them into fake frames, but we edit with frames. So calling it 1080i / 29.97 is the more accurate way to describe the format. 1080i / 60 only confuses the matter as we can see in this very discussion.
As our friend from Adobe notes, the distinction appears to have come to differentiate the true 30p shooting style (30 frames per seconds) vs Interlaced shooting style (60 fields to make 30 frames per second) but in reality, it just created more confusion since there is a true 60p shooting style.
Oh man, that’s a hair condescending 🙂 I’m sure you didn’t mean it that way. It looks like we’re each assuming the other knows less than he does. Sorry if I’m doing that, too.
But you seemed to miss my point. I was *very* clear that I wasn’t saying that 60i is 60 frames. But you didn’t (don’t?) seem to know that there are in fact 60 unique points of “time-data” in 60i footage. You still haven’t acknowledged it.
And I’m a little bummed that you insulted my use of Wikipedia without commenting on the facts. I pulled the Wikipedia quotes only after writing out my response, simply to show how blindingly available this information is.
In the end, it may just be about semantics, but my main concern is that your illustration is completely and totally misleading, because a combined frame simply would not look like that.
You seemed to suggest that interlaced footage is simply progressive footage sliced in half and put back together again, when it’s absolutely not (see: soap opera look, etc.). I mean, if that were true, then interlaced footage and progressive footage running at the same framerate would look absolutely identical, right?
We use frames as a convenience when editing interlaced footage because, as you say, we don’t edit in fields (that would be silly and impractical). This is also why your AJA hardware uses 29.97, because that *is* a practical way to describe how your *timeline* is set up (again, because we don’t edit in fields).
And I only pointed out the DV slo-mo trick to prove, using a well known example, that 60i footage does indeed have 60 unique points in time. I didn’t mean to argue that that old trick is still an effective way to create slow motion footage. But it doesn’t use frame blending, it uses field interpolation. When using the trick, in each second all 60 of the resulting (now-progressive) frames consists of some real and some interpolated lines, even though the footage was captured at only “30fps.”
If I see “30p,” I know it’s 30 frames, and if I see “60i,” I know that it’s 60 fields (which create an effective 30 frames)… as in “60i(nterlaced) fields,” vs “30p(rogressive) frames.”
Just assuming framerate as a constant unit of measurement doesn’t work, especially when dealing with content recorded using 3:2 pulldown, where you need to know both the fieldrate and the framerate. If you shoot “24p” in 1080 DVCPROHD, you can’t call it 1080/24p, because the format is actually 60 interlaced fields using reverse pulldown to spit out 24 progressive frames. You actually call it something like “1080/60i-24p.” If the second number always meant framerate, you could never accurately describe a format like that.
I bow to your greater experience, and maybe it sounds like that to you, but that is equivocally *not* what happened.
I’m not a veteran, but I’m no youngin, either. I started editing during the DV revolution around 1997, and I can tell you that we always used 60i when talking about SD and DV formats. Back then, none of the cameras we could afford could shoot 30p, so it was a holy grail of sorts… and we knew the difference. We understood that 60i meant editing at 30 frames per second, but calling it 30i just doesn’t make any sense, because “i” denotes fields as a temporal unit, not frames.
Right, 60i means 60 fields, while 60p means 60 frames. How is that confusing?
Okay… the confusion which you are describing does exist, I have to acknowledge, but it simply comes from people not understanding how fields work, or assuming that they’re merely a file format thing you can safely ignore, like your illustration implies. It would seem better to educate editors on what the notation means rather than change the meaning just to satisfy widespread ignorance.
How many times do I have to say there are 60 fields in a single second of NTSC interlaced footage to acknowledge there are 60 halves of an image in interlaced footage? If you want to call them “time-data” you’re welcome to do so. They are halves of an image and it takes 60 of them to create one frame of interlaced footage.
How in the world does my drawing suggest that is one frame of progressive video cut in half? That would be complete ignorance. It takes two fields to draw one frame. Your television draws Field one, and then field two. Combine them and you get a single frame of video.
The point of my blog is to educate editors and shooters the difference between 60p which is truly 60 progressive frames of video for every one second vs. 60i which is 60 fields of video that make up 30 frames of video. There is a clear difference between the two.
The use of “60i” in the format of 1080 has created so much confusion that the 60 equals frames. This is wholly incorrect. That is the point of my blog. There is no point in continuing this discussion as you’re just adding more confusion to the topic.
I’m not sure whether we can credit this to Sony or Panasonic, but it stems from the fact that when they started producing cameras that could shoot 1080i, they added the feature to allow you to shoot progressive – so to distinguish between the two modes, you get 1080 @ 60i or 1080 @ 30p. And that’s the way the cookie crumbles…
that sort of kind of makes sense I suppose, but all it does is create headaches for those of us in post when we receive 1080i / 60 “overcranked” footage. The only way to overcrank that is to run it through AE and most likely Twixtor. 🙂
Thanks for the input!
Unfortunately, the Noble Robot really muddied the waters here. For anyone lost in the discussion, Walter is correct.
The two major broadcast standards are 720p/60 and 1080i/30.
Yes, it’s true that the latter is sometimes designated as 1080i/60, but that’s really unfortunate and inconsistent, because in that case the 60 refers to the field rate, not the frame rate. This is of course the original point of Walter’s article.
Note that in another common format, 1080p/24, the 24 certainly means 24 frames per second, not fields.
So…
720p/60 is 1280×720 at 60 progressive frames per second.
1080i/30 is 1920×1080 at 30 interlaced frames per second. Those 30 frames are composed of 2 fields each. So there are 60 fields per second. That number has no bearing on the frame rate, however. (Yes, I understand that each field is temporally distinct, but it takes 2 frames to form a full frame.) Therefore, the number 60 should not have any place in the notation.
It’s distinctly possible that Sony or Panasonic used an alternate notation to market certain lines of cameras too, but from an editor’s viewpoint, there is really no such thing as 1080i60. That would be 60 frames per second and 120 fields per second.
There IS such a thing as 1080p60 (60 progressive frames-per-second), but it’s less common, particularly in the world of DSLRs. It would certainly be nice. To shoot for slow-mo, you generally have to use the lower resolution 720p60 to get the super smooth motion you’re looking for, just as Walter described.
As far as I can tell, Noble Robot is the only one who seems to get it.
While it’s fine and dandy to say that 1080i/30 is 30 FRAMES per second, the reality remains that it is not truly 30 frames per second. It is, explicitly, 60 fields per second, and those 60 fields represent 60 moments in time, not 30 frames divided in half.
Proof of this comes in the form of tearing or stair stepping flaws in an interlaced image. It is the entire reason that motion adaptive deinterlacing has come about as a technology.
In the simplest form, if you have the numbers 1-60 and you flip through them in one second, each number will appear for 1/60th of a second.
Now you make a video recording of them at 1080i.
At no point during playback will you see the complete detail of any single number. You will capture half the number 1 (odd lines), then half the number 2 (even lines), half the number 3 (odd lines), half the number 4 (even lines), etc.
During playback, it would appear as a muddied version of the numbers in question. Half of 1 would be overlapped with half of 2, and then half of 3 would replace the 1, etc. There would never be one instance of a ‘frame’ by the traditional term.
Here’s an excellent example of exactly what I am talking about with a video ‘frame’ which shows that two fields don’t ever equate to one actual ‘frame’.
http://www.surfacedstudio.com/tutorials/video-modes-and-pixel-aspect-ratio
Now, I suppose it depends on what your definition of a frame is. For most people, when they think of a ‘frame’, they are talking about one moment in time. They think of a static full image like the airplane pictured above. They don’t think of half an image or two separate moments in time which aren’t properly overlayed with each other. This distinction is why 1080i60 is not 1080i30. 1080i60 is 60 unique points in time. It is 60 distinct images which represent half a frame, and this half a frame is called a field. 1080i can’t be brought down to 30 frames per second with video captured at 1080i. It can only be 60 fields, exclusively.
Now, I certainly agree that fields and frames aren’t the same thing, that is very clear and should be clear, but the concept that two fields can be combined in a video with motion into a single frame with the pristine quality of what is shown in the example above with the airplane is ludicrous. It’s not just a poor representation, it is a flat out lie which misrepresents how 1080i video is captured.
As is sometimes the case, every rule has an exception…
Television shows are often captured at 1080p/30 and then the 30p frames are split into 60i fields so you end up with 60 fields which can be recombined into 30 frames. But, it doesn’t change the fact that there are 60 interlaced fields being broadcast and displayed. With a proper digital display which can recognize that the two fields are part of a original progressive image, then those fields can be recombined into a true 1080p/30 playback image with each frame doubled up to achieve 60hz. But, this doesn’t change that at the root, the delivery method was using 60 fields originating from a progressive source.
So, no… 1080i60 is not 1080i30 if video is captured using a true interlaced method. Perhaps the argument could be made if video is being captured at 1080p/30 and then split into an interlaced delivery method, but to my understanding, 1080i capture remains at 60 fields per second – 60 unique moments in time which don’t overlap and do not, will not, and never have created ‘frames’.