Here is a fairly typical enquiry we get from a production finishing a film and thinking about the nuts and bolts of getting to the Digital Cinema Package stage using our Post Factory DCP services.
DCP 24fps video conversion from 50p or 25fps video
“Hi, I’m currently doing post production for a feature film and I was considering using your DCP service once the film is ready for festivals.
I have a few questions about frame rates if I may:
We shot the film in 50p and would ideally like to keep that frame rate for exhibition but recognize this is highly unusual and could hinder distribution. Obviously exporting our ProRes HQ master for you at 25p is no trouble but you say you would have to convert to 24p and pitch-shift it.
Does this pitch-shifting maintain the original pitch of the 25p file and is there any loss of quality? The audio is all in 48khz and I was also wondering if a 25-24p conversion would affect sync at all. Is there any advantage to making the conversion at our end and giving you a 24p file to work with?”
And this is my response:
I’m very fond of high frame rate (HFR) productions myself and all those who hated the Hobbit at 48fps are probably mistaken in their focus of antipathy purely on the HFR aspect. The need for very clean lenses for 3D work should shoulder some of the blame for the “video” look than only the HFR, in my opinion. It would be interesting to see a HFR 2D film shot on more cinematic anamorphic lenses.
The reason for the 24 fps restriction is our suggestion based purely on broad compatibility. As a restriction, you can probably ignore it for a limited showing that you can control, say in the UK. The original spec that the first wave of DCI servers was made under only supported 24fps. Now in theory, every DCI server in the world should have been updated with the latest firmware and support HFR, 25fps and so on. Sadly if you have any experience of bureaucracy and institutional dysfunction you will understand why in some rare cases it may not have happened. Some cinemas may not have been upgraded since the day the server was installed.
So if you know exactly which cinemas will be showing your work and what frame rates they support, you can happily go ahead and make a DCP to the most latest DCI spec and know that will play on their systems.
But if you don’t know that, you may well be getting a call 30 minutes before your film is to be shown at a cinema in Aix-en-Provence complaining that your film will not play and it is too late to send and ingest another.
Of course, you could always have two DCP versions made to cover yourself. Large features often make many different DCP deliverables, even with different grades to combat specific display systems.
There are no sync issues converting to 24fps via our process. Picture and sound run at exactly the same speed, just slightly slower. This same process has been used for every single 25fps project made into film, and in reverse, for every 24fps acquired show shown on European 25fps television. The people who notice most are the composers. The theme to the series “Friends” by The Rembrandts runs just that bit faster in its UK television incarnation. At least pitch shifting can kep it in the same key.
QUALITY OF SOUND CONVERSION
In an ideal world we’d never ever resample digital sound from the moment is was converted at the initial A/D stage (mike to recorder) to the output D/A stage (theatre server to speakers) at the viewing end. In reality this ends up happening far more than we’d like. And there are many audiophiles who claim to have a more limited audtory experience in digital than they did in analogue since CDs are only 44.1 KHZ.
But carefully done I’d say this conversion would not be noticeable by 99% of the audience. I’d say the actual quality of the recorded content, and then the subsequent mixing etc is what will be noticed. Things like the quality and proximity of the dialogue microphone at the shoot, the presence or lack of input distortion and then its subsequent level relative to any other elements (music, fx).
So I’d be more worried about using a Schoeps mike over a Rode than the sample rate conversion.
WHO DOES THE CONVERSION?
If you supply a 25fps Prores HQ file we have to turn that into XYZ RGB TIFFS before authoring the DCP. We do this on a frame for frame basis. Frame 1 is Frame 1. Frame 2 is Frame 2. This is then simply interpreted as a 24fps image sequence so will run slower. There is no actual conversion, just a difference in playback speed.
If you do the conversion to 24fps we will have no idea how you did it. We so often see frame doubling, weird interpolation or frame skipping from people who attempted some sort of frame rate conversion themselves. If you are confident doing this yourself then great. But unless you are also confident making correctly colour space converted XYZ/P3 colour space TIFFS then we will still have to do that part anyway: in which case there is no point converting the video yourself.
As far as the audio is concerned you are welcome to conduct your own slow down and pitch shift. There are various ways to do this in Soundtrack Pro, Pro Tools etc. As long as you stick to the absolute values of conversion then nothing will go out of sync. And make sure our deliverable is 24bit 48Khz stems.
Good luck with finishing your project and we are here to help at any stage.