How to prepare your project for a DCP Digital Cinema Package

How to prepare your project for a DCP Digital Cinema Package

Producers come to us with their projects in all different sorts of states when they want us to make a DCP for them at the Post Factory in London..
Major films will have a post-production supervisor who will get everything to exactly what we need.

But at a more indie film budget, producers sometimes just want to give us their files and we will sort out the rest, which we are only too happy to do. We use the best possible tools to do resizing, scaling, audio conversions and assess what needs to be done for your film.

But some producers also want to have the option to save some money and do as much as possible themselves. If you are fairly technically inclined, or can find someone who is, here is what you could do to get your film better prepared.

Work in Reels

Ok, so this is not completely necessary these days but can still be helpful for anything longer than twenty minutes. In the film print days every single director was faced with making points in the films every twenty minutes or less where the picture and sound had to cut. It is highly unlikely you will make a film print these days but there still good reasons to work in chunks.

  • It helps keeps timelines working faster and mores responsively in your editing, grading and sound mixing packages.
  • Finding errors and fixing them is quicker
  • Disk folders of TIFF files are more manageable

Let’s say we’ve made the whole DCP as one picture file and you realise you’ve got a small error in the credits. Any change now means we have to re-encode the whole DCP. If it was in reels, we’d only have to do reel 5, say, again.

Preparation for Cinema

A DCP, to be truly universal like film and play without problems in all DCI servers should be at 24fps progressive, at least 5.1 audio measured by the Leq-M standard and sized to 1998×1080 (2k 1.85 Flat), 2048×858 (2K 2.39 Scope) or double those numbers for 4K versions (3996×2160 or 4096×1716). It will also be in the XYZ colour-space.

Quite often indie producers in Europe will have managed to make a 25fps HD 1920×1080 stereo mixed film to the EBU standard in Rec709 (HD) colour-space. So it needs to be slowed down to 24fps, scaled to 1998×1080, converted to 5.1 at an acceptable Leq-M and converted to XYZ colour.

Resizing/Scaling

If you have finished in HD at 1920×1080, you need to scale up slightly to 1998 width but still with 1080 height. This is around a 4% upscale, and you will lose a little of the top and bottom of the image. If you have framed tightly, you may prefer to do this yourself on a shot by shot basis so there are no unexpected results.

Of course if you actually shot with a camera that shoots larger than HD, you should be really be going back into your finishing system and scaling from the camera originals there. It seems an almighty shame to shoot on a 6K camera like the Epic Dragon, then scale down to HD only to then scale up again slightly to 2K Flat. If you have to scale, try and only ever scale once. If you have shot on a 4K capable camera then pay serious mind to making a 4K DCP.

The only exceptions to those sizes would be if you had a historical piece that was say 4:3 and did not wish to crop it. Then we’d scale it to fit the height but with black borders left and right. In these cases special notes for the projectionists mean if they can they may wish to bring in the masking curtains manually on the sides.

Progressive and Interlaced

A few years back most video cameras shot interlaced (two fields of alternating lines of half the image, taken at half the duration each). But film is progressive: a frame is a frame, all shot/sampled at the same time. So to show interlaced source material on a progressive display, we’d have to de-interlace (combine the lines in some way) so you wouldn’t see strange ragged edges on movement.

Luckily now nearly all cameras shoot progressive. So that raw material is perfectly ready for a DCP. However, editors very often make the mistake of putting progressive material into a sequence that their editing software treats as interlaced. So when captions, rollers and speed change effects are done, these are done with interlacing, which will create artefacts when made progressive. You may not see this if you are watching on an LCD or LED that de-interlaces the video for ou on the fly.

So double and triple check that your post workflow has sorted out any of these problems, or allow us to de-interlace the footage. We generally use a “smart interpolation” algorithm which loses the least amount of detail with the smoothest action.

Anamorphic or Cinemascope

Of course, if you have shot anamorphic, or if you are going for the cinemascope look by cropping your images top and bottom, you need to be scaling to 2048×858. So no actual letter-boxing will be seen, those are the pixels you need.

A lot of the time we see people shoot HD then put a letterbox effect on their output. This will often not match the exact 2.86 aspect that is the DCI spec of 2048×858. So yet again we end up having to make a compromise in scaling up and cropping. You are best off finishing your film in a sequence that is 2048×858 wide and then if you need to make an HD output of that, take that composition and nest it letter-boxed into an HD sequence. That was you have both output options covered.

XYZ Colour-space

Most colour grading systems available to indie producers are at best set up and calibrated to the HD video standard called Rec709. This is also quite near the computer SRGB standard. (I discuss the issues of colour-space in a lot of detail in this post). Rec709 was designed before more modern display technologies came along, and particularly DCI cinema projectors are capable of showing more colour via a standard called XYZ. But the happy thing is it is perfectly easy to represent your film within the XYZ container. You just are not going to be making use of some of the more extreme colours available. But neither did film.

But if you do have the option to grade in a an XYZ capable suite, and you think you would like to take advantage of the additional colour depth, you could make two output versions: one XYZ for the DCP, and one Rec709 for TV and computer display.

Time stretching

I’ve covered this previously in this blog here

The usual way to go from 25fps to 24fps is simply to play the 25fps at 24fps to slow down the film. This means no slurring or blending in the images as each frame is exactly as it was in the original.  But it slows down the pace ever
so slightly which is most notable in the sound. We normally adjust the pitch of the audio up slightly so the sound is not slightly lower.

So you know, this is the way every professional production shot at 25fps and then converted to 24fps is done. Only people very close to the production such as you, the editor and the composer will really to notice this. You can use various plugins in Protools and Logic to do this time-stretch and pitch shift. Of course, you should then check the result against the 24fps version of your images all the way through to make sure you’ve got the formula right. If you are going to do this, make sure you finally output 48Khz 24 bit wavs for our source.

Surround Sound

If you have a stereo mix, we will use an algorithm to convert to a “pseudo” 3.1 in a 5.1 container. If we don’t, the center speaker of the cinema is not used, and that is the one the dialogue is meant to come out of. If it comes out of the stereo speakers, they are much wider apart in a cinema than a television and the film sounds hollow. The algorithm uses math to figure out what is mono, ie the same in both channels, and moves that to the center speaker. It then subtracts that from the left and right to maintain the balance and so they still have the wider stereo imaging.

It will also take some of the lowest frequencies and pass them to the LFE channel which has a special bass speaker of its own in the cinema. Little is passed to the rear surround speakers, hence the 3.1.  All of this means your film will sound a lot more in keeping with the other material played in the cinema. You don’t want to have a great trailer of commercial before your film coming out of all the speakers and then your film starts and only comes from two of them.

Again, you can choose to do a proper 5.1 or 7.1 mix yourself (or even 22.1 if you a really up to it!), or do the faux 3.1 conversion yourself using plugins for ProTools or Logic from companies such as TC Electronics.

Loudness and Audio Levels

Finally we’ll assess the loudness levels of the final soundtrack.

When Dolby completely controlled the quality and level of sound going to film prints, the cinemas had a reliable standard of loudness. But it takes a lot of know-how and investment to have a Dolby certified mixing studio and so using them can be pricey. For an indie producer the chances are unless the studio you have mixed at makes a lots of cinema mixes, they will not have felt it cost-effective to spend the money getting that Dolby certification. That does not in any way mean they can’t do a good mix to Dolby standards.

Now you do not need a Dolby certificate to make a DCP (unless you are submitting to a distributor or ad chain that specifies that). But for the sake of all our ears and the annoyance factor in cinemas, we will always run your audio mix through a Dolby meter to get a Leq-M measurement of the overall loudness in cinema terms. Quite often, dubbing mixers will mix slightly hotter and more compressed for a more TV orientated mix. In these cases, we will drop the level of the audio to reach the maximum accepted overall loudness for cinema. If you mix in a studio that does have a Dolby meter, you can pre-empt this by mixing to a Leq-M of between 74 and 79. Over 82 (where trailers can go to) will be too high and seem fatiguing over anything longer than a few minutes.

But under 74 can run the risk of the cinema playback being too soft. You do need to “venue-proof” your film. We have found that some multiplexes may not have the best sound insulation between their closely packed together screens. So they have turned their amplifiers down from the standard they should be at because noisy action films disturb the quiet foreign film next door. So if you have a quiet film already, you need to make sure it is not TOO quiet.

On to the DCP

Now that your assets are all in the right shape we can get on with making a DCDM (the interim master before the DCP) and then the DCP itself.

Share This Post On

4 Comments

  1. Upmix 2 to fake 5.1
    ________________________
    Hi, thanks for your article. Do you know any free vst plugin similar to TC Electronics? Specially I am interest for function “algorithm uses math to figure out what is mono, ie the same in both channels, and moves that to the center speaker”
    Thanks for your answer

    • You can do this manually in Audacity using the various functions. There is a tutorial here
      https://www.youtube.com/watch?v=zu37UaVlLJE

      Else if you are windows, the BeHappy suite has upmix functions as well. There are various formula options but you may find you need to tweak the resulting tracks in audacity to make sure the sound is still representative of your mix.

  2. Hi! I made a DCP of 9 minutes duration, and the size was of 2.8GB, and it seems to me that is a very low size, I ask to CuteDCP, and they say to me that is for the quality of picture: More noise/grain=bigger size (in GB)

    I make test with the same project (one with noise and other with low noise) the noisy is more large (9GB), than the clean one (2.8 GB), I’m looking for opinions about this, Can you help me?

    Thanks a lot!

    • Hi and sorry for the late reply. I haven’t used CuteDCP but my rule of thumb for a 2K DCP is that a minute is a gig on average. Of course the amount of detail in the imagery can make a difference and if you have very soft imagery, then being JPEG 2000 compression that can take less space. I’d suggest trying to encode your JPEG 2000 using the free version of DaVinci Resolve from BlackmagicDesign and seeing how big that folder is. Don’t forget you want to choose either 2K DCI Flat (1.85) or Scope as your project setting.

Submit a Comment