Little Frog in High Def

I’ve always found that confusing too. When people talk about 60fps, you don’t know if they’re talking about FRAMES per second or FIELDS per second, which makes a huge difference.

There needs to be some kind of notation like a capital F for frames and a small f for fields or something.

But why do people even talk about fields per second? It is always 2 per frame and never changes, so the more pertinent statistic is the frame rate.

yeah that confuses the hell out of me as well. I’d also appreciate it if someone knows why they went that way.

I think you answered it already in your post, because it’s “i” interlaced. And as Jon pointed out its Fields vs. Frames.

Here’s a good explanation:
http://en.wikipedia.org/wiki/NTSC#Framerate_conversion

and

http://en.wikipedia.org/wiki/ATSC_Standards#Resolution

But it is hard to explain….

But why is it called 1080i 59.94? EVERY OTHER format uses that number to indicate FRAME RATE. Why do they suddenly stop doing that for 1080i can switch it to FIELDS per second? Calling it 1080i 59.94 is highly misleading…it makes people think that it is 60fps. Because, as I said, all the other formats use that number to indicate frame rate.

This caused a lot of headache when I was trying to explain to an assistant editor and post super what settings they needed to use to capture their footage. And this really caused problems when a camera guy shot 1080i 59.94 thinking that it shot slow motion. This lead to him telling producers that yes, 1080i has a 60 frame rate with that camera so they can have slow motion just like they could with DVCPRO HD 720p60. Which, of course, is not true. And he should have known better.

It is stupid…misleading, and has to go.

Here’s what my engineer said:

It is (was) called 1080i 59.94 to somehow satisfy the networks who were battling between 720 and 1080 as a standard when HD was coming online. He wasn’t sure of the exact reasons but that was the jist of it. That made is “sound” like 1080i 29.97 had the same kind of motion as 720 59.94. The real reason it exists at all is that a single FIELD of 1080 has the same data rate / transmission rate / baud rate (whatever comes out of the antennas) as a single FRAME of 720. That’s another reason there came the two standards. Some networks wanted 1080 and some 720 so they needed to still be able to transmit both.

So if you notice you AJA presets in FCP they are all 1080i 29.97 which is actually correct since 1080i 59.94 really, actually, technically doesn’t exist. At least that’s how he explained it to me. I feel smarter now so that’s the story I am sticking to!!

but it is too damn confusing.

Thanks Scott…sounds just like the shit “they” would pull.

Yeah, the AJA settings saying, PROPERLY, 1080i 29.97, has caused much confusion in the post world…and the cause of one of my explainations to the post super. Because, well, DECKLINK has 1080i 59.94 listed “so why doesn’t AJA capture 1080 at 60fps? Decklink does!”

sigh.

I have to agree with Shane. This was the dumbest approach ever! No idea why they went this way – although I bet Mark Schubin would know – but it causes all sorts of confusion.

One of the things we can all do to help clarity is to be consistent in descriptions: 1080i59.95, 720p59.94 so at least we know whether we’re dealing with field rate or frame rate.

Philip

Just wait until the real and true 1080p60 becomes ubiquitous in the market. (Yes, it does exist) It is then that that confusion will come to a head.

“One of the things we can all do to help clarity is to be consistent in descriptions: 1080i59.95, 720p59.94 so at least we know whether we’re dealing with field rate or frame rate.”

I disagree, that’s the wrong thing to do since 1080i 59.94 is really 29.97. We know it is a field rate since it is 1080 (i) right?

Yah, but 720p broadcasts to your living room are 720p 29.97 (aka 720p30 – not 720p 59.94, aka 720p60)

Don’t forget about 29.97 psf (progressive segmented frame) like HDCAM

not to mention that in order to set an HDCAM-SR machine so that it will record and playback 1080/59.94i you must select 1080, Interlaced, 29.97! That (and the rest of the stuff mentioned already here) is why when I’m training assistants (and editors) in terminology I tell them always look at frame rate. We never called standard def 480/59.94i right? Why start this field-per-second nomenclature now? I try to make it a habit on any labels and correspondence with other studios and clients to use frame rates only. It’s not catching on like I wanted it to, but it’s a small peace of mind…

To Anonynous @ 4:49 PM-

All 720p is broadcast @ 720p59.94. That’s 60 frames per second. 720p30 only exists on computers and p2 cards/or similar media. When you watch that footage it’s actually 720p30 with 2:2 pulldown. To SMPTE, there’s no such thing as 720p30. Don’t add to the cornfusion!

Now, imagine knowing very little about how film/video is shot to begin with, making a decision to become an editor and have someone explain that to you. And you thought YOUR head spins with this stuff…

my head is going to explode.

Seriously, I must be stupid. It doesn’t confuse me at all. Leave aside the resolution – 1080 or 720. It’s either 60p or 60i or 50i or 30p or 25p or 24p.

There is NO SUCH THING as 30i. NO 25i.

IOW, no matter what confused manufacturers may label it, the i or p tells you whether it’s fields or frames (no need for big F/little f to further confuse the issue). The number tells the refresh rate.

Finally, psf only matters as a tape format since it’s recorded as interlaced (fields), but each pair of fields is assembled into a progressive frame.

“Your Head A Splode!”

How in the world does a self respecting post SUPERVISOR not know this?

I agree that it’s counter intuitive notation, and that it would have been better for everyone involved if SMPTE or SONY/Panasonic/Canon/Jvc had made it fore consistent.

But i means interlaced, p means progressive, and notation out there calls NTSC 60i, and PAL 50i. I’ve never seen NTSC called 30i.

I totally understand an editor or post assistant, or post co-ordinator getting confused, as 23.98 gets reduced to 24p as short hand a lot, but a supervisor should know this.

There’s one place where SD NTSC is referred to as 30i NTSC – Avid Media Composer. Maybe looking at that for years, someone could get the idea if he saw 60i that meant 60 frames per second…

Shane, I didn’t realize that people talk about 1080i as 59.94… that’s crazy talk [smile].

When you get a chance, I’d love it if you’d delve into the 4.4.4 vs. 4.2.2 vs 4.2.0 world. I’m still trying to wrap my head about it, specifically how the CCDs use both the 4.2.2 and the YUV and somehow meld into an image with more resolution but less color… or something like that. I really don’t understand the CCD side of things.

I’m really interested in the way it actually works–on a camera level–but I’d also love your take on how the various compression options affect your editing.

I realize you’ve got a lot going on, so if you don’t get around to talking about that, that’s okay. I mostly just wanted to add another aspect of the HD world that is confusing–at least to me [smile].

~Luke

Sorry Luke…I don’t have a firm grasp on the ins and outs of 4:2:2 vs 4:4:4 color space. I know that one is better, that’s it. Mainly I am the creative cutter. The fact that I know as much as I do about other technical stuff is uncommon among creative offline editors.

Okay, here goes…

The way most people use it, 4:4:4 means RGB. And really they shouldn’t. They should just call it RGB, which has no chroma (colour) subsampling (it has red, green and blue values for each pixel), and therefore no need for a ratio. There is also 4:4:4 in YCbCr, but you need 8:4:4 for the equivalent to RGB.

It’s difficult to explain 4:2:2 and 4:2:0 without diagrams, so I’d direct you to Wikipedia to get a proper grasp of it. (And, incidentally, in these digital times it’s YCbCr not YUV. May sound pedantic, but there’s a difference.)

The concept of chroma subsampling is pretty digestible — it’s when you get to sensors it gets exponentially more complicated. That wasn’t so five years ago, when you could pretty confidently say all camcorder sensors were RGB and it not seem misleading. But nowadays there are things like pixel off-setting, Bayer sensors and a plethora of HD codecs to complicate matters.

First I’ll explain how a traditional DigiBeta camcorder works: the light comes in through the lens and hits a prism, which separates it into red, green and blue wavelengths and sends it off in different directions. There are three CCD chips surrounding the prism, one sensitive to the red part of the light spectrum, one to the green and one to the blue. The light that hits each is put together, so that you have a red, green and blue value for each pixel. At this point it’s true RGB, but in order to reduce bandwidth it’s chroma subsampled to 4:2:2 and from there laid to tape with mild compression.

Simple. The only thing in this generation of camcorders that came along to complicate things was widescreen, which involved luma subsampling to keep the bandwidth the same.

Then came HD… I’ll use the Panasonic HVX-200 as an example. It uses the same three-chip prism array as described above — it even uses the same SD chips, but in order to generate an HD image it off-sets them. That is to say it intentionally misaligns them so that the corresponding pixels overlap one another, achieving higher apparent resolution. Again, this is difficult to describe without visual aids, but imagine it as like a Ven Diagram. Without going into mind-bending detail, you could simply say this works to an extent. But never as well as genuine full-res chips.

The DVCProHD codec then chroma subsamples to 4:2:2 and luma subsamples to 1280 (29.97fps) or 1440 (25fps). 1080i HDV camcorders chroma subsample to 4:2:0 and luma subsample to 1440. At this point the ‘4’ seems misleading. It works as a relative terms, but seems wrong when there is luma subsampling. (It has no intrinsic value anymore, it’s just comparative.) HDCam advertises 3:1:1 subsampling, and that seems more transparent in seen as they luma subsample to 1440.

If all that seemed a bit messy, it’s because it is. There is no standard any more.

Bayer sensors are found in prosumer camcorders, DSLRs, and digital cinema cameras like the Red ONE and Arri D20. They are single-chip and don’t have a prism. Where the prism split the light and enabled each pixel to have a red, green and blue value, Bayer sensors have only one colour per pixel. (Do a Google image search for a diagram.) They then have to use an algorithm to give each pixel an RGB value, in a process called de-bayering or de-mosaicing. This is why they are often high resolution, and why a lot of people take issue with Red claiming their camera is 4K. Much like the off-setting wizardry Panasonic use in the HVX, the result somewhere short of a genuine full-res RGB sensor. In the smaller of their advertised numbers, Red claim they get 3.2K of detail from their 4K sensors.

Wow that was long (oh, and I haven’t been bothered to read it through and check for errors so there will likely be some). Hope it did more good than harm.

Pip, thanks so much for typing all that out. I really appreciate the complexity and craziness of that part of the technological world now. Unfortunately, I’m still lost [laughing]. Thanks for making a little clearer in my mind.

~Luke

the solution to all the frame rate BS is to insist to your producers that “our decks and edit systems only work in 23.98.” if we all band together and insist on this, all over the world, we can have a standard for once.

Great comment on why even note the number of fields…aren’t there always 2?? I agree that it is confusing…as a DP shooting for 10 years mostly analog…I find all these codecs and format options head spinning!

Woa!! great post! I have just spent my time in about four forums asking people why I cant work with 59.94 in after effects if my Media Composer said it was 60i. Now I have it clear.

My footage in Media Composer is 59.94i wich is the same as 29.97fps right?? So there is no problem when working it as 29.97 and then returning it to MC??

So many thanks really!!!

Yeah, but we have always call NTSC 29.97fps and in Media Composer it says 30i for SD projects and it is not 15fps. That was my confussion, thanx a lot.

Panasonic AJ-PX270 P2 HD Camcorder (Firmware: 11.31)
Manual (PDF)
ftp://ftp.panasonic.com/pub/panasonic/business/provideo/op_manuals/AJ-PX270P_Operating_Instructions.pdf

The PX270 can record in formats
[AVC-G25/60P] & [AVC-G25/60i]

Does this mean it is P (Progressive) as defined below?
Or are there 60 unique (different) frames?

P (Progressive) movies are recorded with duplicate frames.

PA (Progressive-Advanced) movies are similar to the above, but the duplicate frames are flagged for automated removal on ingest.

PN (Progressive-Native) movies are recorded at 24 frames, period. Not compatible with firewire out to a disk recorder, but saves a ton of space when recording and does not require pulldown removal.

Thanks in advance.

So, can anyone shed light as to why they call 1080i 59.94 that, and not 1080i 29.97? – For a 1080 source, a 59.94 frame rate expresses the FIELDS of an interlaced signal, 29.97 expresses the FRAMES of a progressive signal. 1080i 2997 is often used but is actually 1080i 59.94, these are often used interchangeabily by folks that don’t need to know the difference.