This post was 2016. Six years later, this is no longer true, at least here in my area (Montréal). At my work in a post-production company, we finish around 8-10 long form fiction movies a year and a bigger number of shorts, and I would say around 90% of thoses projects are now shot 24.0 fps.
All high-end cameras like the ones from Arri, RED and Sony can switch easily between 23.976 and 24.0 since many years. The sound recorders can now too, like the Sound Devices, which is the most common brand used (at least here in North America). The choice of shooting in 24.0 was already available in 2016, but not every gig would be used to shoot 24.0 so they chose 23.976 just to be safe, I guess.
Also, in 2016 television was still a major deliverable, but this has changed. Video on demand, which supports any framerates, is now the main distribution channel when a movie has finished playing on the big screen.
Documentaries are more of an exception because they are still aimed at television and also, they often use old footage which was telecine'd at 23.976 so it's easier to edit when everything's 23.976.
> Video on demand, which supports any framerates, is now the main distribution channel
But all the major VOD services, be it Disney+, Netflix, Hulu,... still mainly use 23.976 or 29.97 fps for their video distributions.
I don't know what your production company mainly works on, but these NTSC frame rates ares still the norm across the whole cinematic industry, and it's not going anywhere. It's not limited to documentaries at all.
The YouTubers etc. do use 30/60 fps more often OTOH.
Good question. Last time I've submitted to Netflix, it was 24, but it goes through a third party "packaging" company before going to Netflix.
So some of them might reconvert to 23.976 even when we submit 24 fps masters as per the contractual agreement. Maybe you are right. I will check.
Low budget web series shot at 23.976 even get delivered and played back at 29.97i on some local platforms here in Canada (Crave, Noovo, tout.tv), so anything's possible nowadays.
On the other end, Blurays can be encoded at both 23 and 24, and Vimeo, Youtube, etc all support 24fps. So 24 fps exists, not just on DCPs.
By the way, the loss of quality from going to/from 23.976 and 24 is not much. I've never heard any artifacts from that kind of conversion. But since cinema theatres are most likely to have a better sound system that a home system, I think it makes more sense to have the unconverted mix playing in the cinema and not vice versa.
Not well versed in video production but really interested layperson.
I understand that the two main reasons for 23.976/24 fps are that it’s:
1) been the standard for a really long time so you know everything will more or less support it (cinemas/vod/broadcast tv).
2) is now in people minds as what film should look like (in that 60 fps somehow looks “off” because we’re trained to expect 24 fps).
Given that analogue broadcast tv is dying off and that digital OTA tv is a similar case to vod in terms of codecs (maybe not receiver support?), wouldn’t a stopgap be some multiple of 24 fps (e.g. 48fps) that would allow better motion handling without the pace seeming off?
Production audio and video recorders generate or intake an SMPTE timecode signal, and stamp recordings with this timecode.
This timecode format is a timestamp with seconds resolution plus a frame count within each second. To properly sync, all the timecode generators must use the same framerate. In other words, the audio recorder’s timecode framerate needs to match the camera.
I'm guessing, because I'm bored, and hopefully an expert will confirm or correct, but I think it means that they place sync markers every 24th of a second, to make syncing with video efficient.
Because 60fps and 24fps don't look alike and video shot at a very high frame rate can look off to an audience.
Remember also that typically a 1/2 rule is followed where the exposure of each frame is half the time of its display. So for each of the 1/24s intervals, an exposure is made for 1/48s. A shorter exposure would have less motion blur but look jittery. A longer exposure would look too buttery. People are extremely used to 1/48s exposures displayed for 1/24s (or similar).
48fps video in the Hobbit movies was very negatively perceived. It looked too much like a video game.
As others have said, 60fps can actually look "off" to some audiences. It definitely looks "off" to me, and I can't quite put my finger on it/why. It could just be my age (late 30s) and my brain grew up on 24-30fps.
Many new TVs interpolate 24/30 fps to be 60fps. This is knows as "The Soap Opera Effect", and has been written about here before. [0].
To me, 24/30fps feels "smooth", almost "buttery". 60fps feels crisp. 60fps feels almost too real. It makes it more difficult for me to suspend disbelief, whereas 30 makes it easier to suspend disbelief and get lost in the imaginary world that's being presented to me as if it were real. It's really difficult for me to explain....
I've completed a video shot at 24fps, some footage was recorded at higher frame rate for slow motion, but you can tell the diference. The 24 fps is the best, as it looks and feels better in the eyes for most people and also my audience :)
To be fair, the first line of the response heavily qualifies the answer: "Truth is that most digital projects shot for either broadcast or cinema use 23.976 instead of 24 in the U.S."
I think they mean the VOD service doesn't restrict the creator on what frame rates they can author their content in, the box just makes it work for whatever the display requires.
I prefer to write them as 24000/1001, 30000/1001, 60000/1001 fields-per-second to avoid the ugly decimals and because I think it makes it a little more intuitively clear what’s going on to achieve those numbers.
Reminds me of Facebook's Flicks, which are a super short tiny common denominator time unit:
> A flick (frame-tick) is a very small unit of time. It is 1/705600000 of a second, exactly.
> 1 flick = 1/705600000 second
This unit of time is the smallest time unit which is LARGER than a nanosecond, and can in integer quantities exactly represent a single frame duration for 24 Hz, 25 Hz, 30 Hz, 48 Hz, 50 Hz, 60 Hz, 90 Hz, 100 Hz, 120 Hz, and also 1/1000 divisions of each, as well as a single sample duration for 8 kHz, 16 kHz, 22.05 kHz, 24 kHz, 32 kHz, 44.1 kHz, 48 kHz, 88.2 kHz, 96 kHz, and 192kHz, as well as the NTSC frame durations for 24 * (1000/1001) Hz, 30 * (1000/1001) Hz, 60 * (1000/1001) Hz, and 120 * (1000/1001) Hz.
Guess the project got archived. Seemed like a semi sensible high res time format that was very media friendly.
The code amounts to a definition of the unit in a way that’s usable with the C++ standard library, there are no exchange formats or anything. That the repo got archived might mean that they are not planning to promote it, but neither the code nor the unit itself really need any further development—if you have a place where you want to use them, you can just do that.
Unfortunately most digital video formats don’t actually store them this way, either because the editing software rounds it off incorrectly or because the format itself stores timestamps in decimals for some reason (Matroska).
<rant>Ah...Matroska was nearly perfect, and then they chose nanosecond-precision timestamps. Countless attempts at frame-perfect seeking thwarted and videos imbued with jitter over the years, subordinate to the tyranny of 33 34 33 33 34 33...
Even worse, not only is the recommended precision only 1 ms, but almost all muxers support nothing else, ffmpeg included*. A shame that despite such versatility, mkv is scuffed as an intermediate/archival format by lossy muxing, yet almost no one is aware.
* https://lists.ffmpeg.org/pipermail/ffmpeg-devel/2021-May/280... matroska precision parameter coming soon? It's been over a year, though - looks like the patch is dead. Meanwhile, the bikeshedding of rational timestamps in the matroska lists can now be expressed with a timebase of decades....
Many multimedia formats, frameworks and libraries actually store them that way.
In Mpeg 4, each track header has a “time scale” integer field. For a video track of 23.976 FPS the value can be 23976, or half of that, 11988. Audio tracks typically have different time scale field, often 48000 for 48kHz audio.
In Media Foundation, MF_MT_FRAME_RATE media type attribute is a rational number with uint32_t numerator and denominator. In DXGI, the RefreshRate field of DXGI_MODE_DESC structure which contains a refresh rate of a monitor is a rational number, for instance my display actually uses 59997/1000 Hz frame rate, not exactly 60 Hz.
But don’t you then have to then explain this every time you try to communicate with these numbers to someone else. Are you just writing this down for yourself?
“Teachable moment” I guess? I usually use that notation in my nfo blurb when encoding DVDs for a modern flatscreen target. Anecdotally, nobody has ever asked me to clarify. Small sample size though due to niche nature.
The amount of hidden engineering in Analog systems is mind blowing. Imagine how many hours were spent in a lab figuring out the exact right timings, chemical mixtures, and circuit design needed to make modern Cinema exist as we know it.
Yeah, I read stuff like this and think the engineers back in the 50's were some of the smartest. I certainly would not have been able to cut the muster back then.
Numerous examples in early digital era as well, see e.g. fast inverse square root, cga color hacks (also entire demoscene) and endless tricks in software that has been outrunning hardware for a while.
Later these were declared wrong, considered harmful, having bad smell, not passing code review and it all became a boring task of combining lego blocks. Sure these analog guys had insane tricks in their sleeves, but our “collective industry” would probably stigmatize them immediately if met today.
Certainly I think we should ignore the trope that we've left that analog crap behind up and realise that those are just some of the giants whose shoulders we stand upon.
Every generation has its challenges, but we can be greatful that the generation before us gave us new ones. It has not always been the case.
Only because you are sitting on top of a million prebuilt tools and products. Essentially the same as someone who just purchased the hardware and plugged a video feed in from the camera.
Great answer for where 29.97 came from, but I seem to be missing the explanation of 23.976 fps which is suddenly introduced as just an effect of audio post, but not how or why exactly...
Edit: more specifically, if audio was being converted to NTSC for dailies, why would it be 23.976 and not 29.97?
Because audio for 23.976fps film is playing at the same speed as 29.97fps video. Which is really interlaced video at 59.94 fields per second. 3:2 pulldown is used to synthesize extra video frames to transfer to NTSC video so for a given instance of time, for every 4 frames of film you get 5 frames of video. (23.976/29.97 = 4/5) The instance of time remains unchanged so the audio is playing back at the same speed. This is all inherited from the CRT days before progressive video playback on flatscreens, and the video fps was tied to the local electrical frequency (50 Hz in Europe vs 60 Hz in USA / Japan) which is how the CRT timings were derived.
Coincidentally PAL video plays back at 25fps, to convert film to PAL they just speed it up from 24 to 25fps, audio and all, which is why if you watch a Euro DVD release of the same film all the actors sound like they just ingested helium before the scene because the audio is playing back ~4% faster. This was largely fixed with Blu-Ray as films play back at their native rate[1] on progressive displays but our European friends who have only seen films on VHS/DVD or TV and not in the theatre or Blu-Ray have been done a disservice, because the video and audio is sped up on playback and not as the director intended it to be shown.
[1] A notable exception is some unscrupulous distributors abroad release shitty PAL 25fps Blu-Rays of films which are simply upscaled PAL DVD versions that they then market as Blu-Ray but are worse than DVD quality but at marked up prices. This is more common than you think. Official releases from the studios are encoded at 23.976fps on the disc. TV shows shot on PAL video (native 25fps) will always play back at 25fps.
Around 2009, my street flooded and I decided to record it and sell it to a local news station. I had one bite and I took my camera over to the TV station (CTV).
When I got there, I was beyond disappointed that (1) they didn’t have an HDV player and that (2) I forgot my FireWire cable. I rushed home, imported/exported it in FCP, burnt it to a DVD and rushed back.
In the rush, I forgot to ensure the right frame rate was set. When I got there, the techs were able to ingest it no problem, but my heart dropped when I noticed the import rate on a little VFD was 1080/50i. Fortunately, it worked out and when my video aired, it looked like it was from an international source because of the pulldown.
I ended up getting a cheque for $135 for 8 broadcasted seconds of video and a day’s work of rushing back and forth. I’ve extrapolated it to be my most valuable piece of work at about $60,000 per hour.
> This is all inherited from the CRT days before progressive video playback on flatscreens, and the video fps was tied to the local electrical frequency (50 Hz in Europe vs 60 Hz in USA / Japan) which is how the CRT timings were derived.
I'll have you know, there were CRTs that did progressive scan too. Early HDTVs were CRTs, before flat screen TVs got big.
Plenty of PAL CRTs can accept NTSC at 60Hz on a 50Hz supply too, of course. It's nice to tie the supply and video rates together to reduce interactions between the two rates.
On the topic of playback rate/pitch fuckery, Janet Jackson's "Rhythm Nation" was sped up ever so slightly after recording (as was sometimes done at the time to make the music sound more "exciting" or "intense"), and that's one of the hypotheses for why playing it interferes with laptop hard drives: the dominant E frequency/pitch got shifted up to somewhere slightly sharp of E (but not quite E#), which just so happens to be closer to some drives' resonant frequency.
That PAL speedup drives me nuts. I thought I was losing my mind the first time I noticed it. Nobody else did. Eventually I reasoned out why and found the appropriate name for the phenomenon.
You can get pitch-shifters, but on DVD players, the only ones with this feature I can find happen to function as karaoke machines.
>Because audio for 23.976fps film is playing at the same speed as 29.97fps video. Which is really interlaced video at 59.94 fields per second. 3:2 pulldown is used to synthesize extra video frames to transfer to NTSC video so for a given instance of time, for every 4 frames of film you get 5 frames of video.
The traditional frame rate for film is 24fps, which is converted using 3:2 pulldown to 30 fps. 23.976 fps converts using 3:2 pulldown to 29.97 fps (23.976*30/24=29.97).
The answer is a bit more complex when it comes to the transition to modern high definition television. It's a bit of a long story with very interesting twists and turns that involve politics, national security, the Pentagon, US Congress and none other than Donald Rumsfeld (CIA and Defense Secretary).
The story is well chronicled in a book I read about twenty years ago:
"Defining Vision: How Broadcasters Lured the Government into Inciting a Revolution in Television"
I can't possibly do it justice here. I'll just mention that one would not be wrong to call Donald Rumsfeld the father of high definition television. His approach to wrangling the ATSC and FCC into adopting a cornucopia of standards was, from a business perspective, nothing less than genius while, from a technical perspective, a complete mess. The fractional frame rates would have evaporated from this planet had it not been for this part of the story.
Well worth reading for anyone interested in the technology or working in associated industries. Your jaw will most definitely drop as you get deeper into the story.
One bewildering thing that I think isn't covered there is that old Quicktime video was specified as 23.98 simply because they were using fixed precision. An apparently harmless issue until you tried reading metadata and converting frame rates assuming that just doing the maths would work.
“… by introducing a color subcarrier of precisely 3.579545 MHz (nominally 3.58 MHz). The precise frequency …”
Does anyone have a circuit diagram for any of this? What were they using as a timing reference in the 50s that allowed a frequency lock accurate to 7 significant figures? There must be a fairly large error allowed.
It's 315/88. Chosen to be a half-integer multiple of the line rate. The line rate was also adjusted so that the audio subcarrier is a multiple of the line rate. The audio subcarrier is 4.5 MHz.
So it's really 4.5 MHz, multiplied by 35/44.
In practice, you might buy a 3.579545 MHz crystal oscillator and then use dividers and PLLs to make all the other frequencies you need. The actual error is going to be based on the oscillators you use, but for the past few decades, we've just been using ordinary crystal oscillators and that means something like 50ppm, maybe (for cheap ones).
Crystal oscillators were available prior to the invention of NTSC. Quartz oscillators achieved 0.1ppm accuracy at some point in the 1920s.
Readit News
All high-end cameras like the ones from Arri, RED and Sony can switch easily between 23.976 and 24.0 since many years. The sound recorders can now too, like the Sound Devices, which is the most common brand used (at least here in North America). The choice of shooting in 24.0 was already available in 2016, but not every gig would be used to shoot 24.0 so they chose 23.976 just to be safe, I guess.
Also, in 2016 television was still a major deliverable, but this has changed. Video on demand, which supports any framerates, is now the main distribution channel when a movie has finished playing on the big screen.
Documentaries are more of an exception because they are still aimed at television and also, they often use old footage which was telecine'd at 23.976 so it's easier to edit when everything's 23.976.
But all the major VOD services, be it Disney+, Netflix, Hulu,... still mainly use 23.976 or 29.97 fps for their video distributions.
I don't know what your production company mainly works on, but these NTSC frame rates ares still the norm across the whole cinematic industry, and it's not going anywhere. It's not limited to documentaries at all.
The YouTubers etc. do use 30/60 fps more often OTOH.
So some of them might reconvert to 23.976 even when we submit 24 fps masters as per the contractual agreement. Maybe you are right. I will check.
Low budget web series shot at 23.976 even get delivered and played back at 29.97i on some local platforms here in Canada (Crave, Noovo, tout.tv), so anything's possible nowadays.
On the other end, Blurays can be encoded at both 23 and 24, and Vimeo, Youtube, etc all support 24fps. So 24 fps exists, not just on DCPs.
By the way, the loss of quality from going to/from 23.976 and 24 is not much. I've never heard any artifacts from that kind of conversion. But since cinema theatres are most likely to have a better sound system that a home system, I think it makes more sense to have the unconverted mix playing in the cinema and not vice versa.
I understand that the two main reasons for 23.976/24 fps are that it’s:
1) been the standard for a really long time so you know everything will more or less support it (cinemas/vod/broadcast tv).
2) is now in people minds as what film should look like (in that 60 fps somehow looks “off” because we’re trained to expect 24 fps).
Given that analogue broadcast tv is dying off and that digital OTA tv is a similar case to vod in terms of codecs (maybe not receiver support?), wouldn’t a stopgap be some multiple of 24 fps (e.g. 48fps) that would allow better motion handling without the pace seeming off?
https://partnerhelp.netflixstudios.com/hc/en-us/articles/360...
What does that mean? I thought 44khz sampling meant 44.000 integers are recorded per second.
This timecode format is a timestamp with seconds resolution plus a frame count within each second. To properly sync, all the timecode generators must use the same framerate. In other words, the audio recorder’s timecode framerate needs to match the camera.
Remember also that typically a 1/2 rule is followed where the exposure of each frame is half the time of its display. So for each of the 1/24s intervals, an exposure is made for 1/48s. A shorter exposure would have less motion blur but look jittery. A longer exposure would look too buttery. People are extremely used to 1/48s exposures displayed for 1/24s (or similar).
48fps video in the Hobbit movies was very negatively perceived. It looked too much like a video game.
Many new TVs interpolate 24/30 fps to be 60fps. This is knows as "The Soap Opera Effect", and has been written about here before. [0].
To me, 24/30fps feels "smooth", almost "buttery". 60fps feels crisp. 60fps feels almost too real. It makes it more difficult for me to suspend disbelief, whereas 30 makes it easier to suspend disbelief and get lost in the imaginary world that's being presented to me as if it were real. It's really difficult for me to explain....
[0] https://news.ycombinator.com/item?id=10613575
I don't think so, majority of displays are still 60Hz without VRR.
> A flick (frame-tick) is a very small unit of time. It is 1/705600000 of a second, exactly.
> 1 flick = 1/705600000 second This unit of time is the smallest time unit which is LARGER than a nanosecond, and can in integer quantities exactly represent a single frame duration for 24 Hz, 25 Hz, 30 Hz, 48 Hz, 50 Hz, 60 Hz, 90 Hz, 100 Hz, 120 Hz, and also 1/1000 divisions of each, as well as a single sample duration for 8 kHz, 16 kHz, 22.05 kHz, 24 kHz, 32 kHz, 44.1 kHz, 48 kHz, 88.2 kHz, 96 kHz, and 192kHz, as well as the NTSC frame durations for 24 * (1000/1001) Hz, 30 * (1000/1001) Hz, 60 * (1000/1001) Hz, and 120 * (1000/1001) Hz.
Guess the project got archived. Seemed like a semi sensible high res time format that was very media friendly.
https://github.com/facebookarchive/Flicks
Even worse, not only is the recommended precision only 1 ms, but almost all muxers support nothing else, ffmpeg included*. A shame that despite such versatility, mkv is scuffed as an intermediate/archival format by lossy muxing, yet almost no one is aware.
* https://lists.ffmpeg.org/pipermail/ffmpeg-devel/2021-May/280... matroska precision parameter coming soon? It's been over a year, though - looks like the patch is dead. Meanwhile, the bikeshedding of rational timestamps in the matroska lists can now be expressed with a timebase of decades....
In Mpeg 4, each track header has a “time scale” integer field. For a video track of 23.976 FPS the value can be 23976, or half of that, 11988. Audio tracks typically have different time scale field, often 48000 for 48kHz audio.
In Media Foundation, MF_MT_FRAME_RATE media type attribute is a rational number with uint32_t numerator and denominator. In DXGI, the RefreshRate field of DXGI_MODE_DESC structure which contains a refresh rate of a monitor is a rational number, for instance my display actually uses 59997/1000 Hz frame rate, not exactly 60 Hz.
“Teachable moment” I guess? I usually use that notation in my nfo blurb when encoding DVDs for a modern flatscreen target. Anecdotally, nobody has ever asked me to clarify. Small sample size though due to niche nature.
Later these were declared wrong, considered harmful, having bad smell, not passing code review and it all became a boring task of combining lego blocks. Sure these analog guys had insane tricks in their sleeves, but our “collective industry” would probably stigmatize them immediately if met today.
Every generation has its challenges, but we can be greatful that the generation before us gave us new ones. It has not always been the case.
We’ve just moved it.
Hell, just look at ffmpeg commandline if you miss some complexity in your life
Edit: more specifically, if audio was being converted to NTSC for dailies, why would it be 23.976 and not 29.97?
Coincidentally PAL video plays back at 25fps, to convert film to PAL they just speed it up from 24 to 25fps, audio and all, which is why if you watch a Euro DVD release of the same film all the actors sound like they just ingested helium before the scene because the audio is playing back ~4% faster. This was largely fixed with Blu-Ray as films play back at their native rate[1] on progressive displays but our European friends who have only seen films on VHS/DVD or TV and not in the theatre or Blu-Ray have been done a disservice, because the video and audio is sped up on playback and not as the director intended it to be shown.
[1] A notable exception is some unscrupulous distributors abroad release shitty PAL 25fps Blu-Rays of films which are simply upscaled PAL DVD versions that they then market as Blu-Ray but are worse than DVD quality but at marked up prices. This is more common than you think. Official releases from the studios are encoded at 23.976fps on the disc. TV shows shot on PAL video (native 25fps) will always play back at 25fps.
Around 2009, my street flooded and I decided to record it and sell it to a local news station. I had one bite and I took my camera over to the TV station (CTV).
When I got there, I was beyond disappointed that (1) they didn’t have an HDV player and that (2) I forgot my FireWire cable. I rushed home, imported/exported it in FCP, burnt it to a DVD and rushed back.
In the rush, I forgot to ensure the right frame rate was set. When I got there, the techs were able to ingest it no problem, but my heart dropped when I noticed the import rate on a little VFD was 1080/50i. Fortunately, it worked out and when my video aired, it looked like it was from an international source because of the pulldown.
I ended up getting a cheque for $135 for 8 broadcasted seconds of video and a day’s work of rushing back and forth. I’ve extrapolated it to be my most valuable piece of work at about $60,000 per hour.
I'll have you know, there were CRTs that did progressive scan too. Early HDTVs were CRTs, before flat screen TVs got big.
Plenty of PAL CRTs can accept NTSC at 60Hz on a 50Hz supply too, of course. It's nice to tie the supply and video rates together to reduce interactions between the two rates.
You can get pitch-shifters, but on DVD players, the only ones with this feature I can find happen to function as karaoke machines.
would you explain this to me like I'm 5?
Deleted Comment
The story is well chronicled in a book I read about twenty years ago:
https://www.amazon.com/Defining-Vision-Broadcasters-Governme...
The title is intriguing enough:
"Defining Vision: How Broadcasters Lured the Government into Inciting a Revolution in Television"
I can't possibly do it justice here. I'll just mention that one would not be wrong to call Donald Rumsfeld the father of high definition television. His approach to wrangling the ATSC and FCC into adopting a cornucopia of standards was, from a business perspective, nothing less than genius while, from a technical perspective, a complete mess. The fractional frame rates would have evaporated from this planet had it not been for this part of the story.
Well worth reading for anyone interested in the technology or working in associated industries. Your jaw will most definitely drop as you get deeper into the story.
Does anyone have a circuit diagram for any of this? What were they using as a timing reference in the 50s that allowed a frequency lock accurate to 7 significant figures? There must be a fairly large error allowed.
So it's really 4.5 MHz, multiplied by 35/44.
In practice, you might buy a 3.579545 MHz crystal oscillator and then use dividers and PLLs to make all the other frequencies you need. The actual error is going to be based on the oscillators you use, but for the past few decades, we've just been using ordinary crystal oscillators and that means something like 50ppm, maybe (for cheap ones).
Crystal oscillators were available prior to the invention of NTSC. Quartz oscillators achieved 0.1ppm accuracy at some point in the 1920s.