Friday, April 22, 2011

How to Export for packing into a DCP

Please note, I will be updating this post (writing a new post, actually) with a lot more streamlined process in the near future.

This post is really for the 3D production class at IU, but other folks may like to read it as well, enjoy.  These examples are based on Adobe Premiere Pro CS5 and footage from our Sony XDCAM EX3 cameras.  I'll go over the setup and export settings for making intermediate files suitable for packing into a 3D DCP (digital cinema package) that we'll screen at the IU Cinema on May 1st.

First, a general overview.  To pack a DCP we need to get the picture and sound essences of your movie.  In the case of a 3D movie with stereo sound, there are four essences: Left Eye frame sequence, Right Eye frame sequence, Left audio, Right audio.  The frame sequences are high quality image files, usually in the DPX image format, that will fit the 2K image container and run at a rate of 24 true frames per second.  The audio files are monaural WAV files with a sample rate of 48 kHz, and a bit rate of 24.

First, you need to create two exact copies of your picture locked final edit.  Append an L at the end of one copy's name and an R at the end of the other copy's name.

  1. Go into the copied sequences and select all of your stereo container sequences
  2. Enable multi-camera on these sequences
  3. Use multi-camera 1 for the Left master edit copy (MC1)
  4. Use multi-camera 2 for the Right master edit copy (MC2)

Next, let's prepare a proper 2K container sequence for each eye in our project with our chosen cinema aspect ratio of 1.85:1 or 2.39:1 (a.k.a. Flat or Scope).  The example here is designed for Scope.  Keep the frame rate at 23.976 and set the pixel dimensions at 2048x858 (this is for Scope, if you're editing for Flat use 1998x1080) by enabling Desktop as the Editing Mode at the top of the new sequence dialogue.  You'll use this as a container sequence for framing in Scope and exporting a properly sized DPX sequence.

Nest your Left eye edit (MC1) into your Left 2K container sequence and view it to check for head room and proper vertical alignment (not 3D VIT disparity in this case).  This is pretty easy if you shot your project with a Common Top framing.  You'll also need to scale up your movie from an HD size to fit into the 2K size container.  For Scope we'll scale up 107%, for Flat you'll scale up about 104%.

Once you have your left and right 2K container sequences setup, you'll be ready to export your DPX sequence.

Hit Command + M to open up the export dialogue and select DPX as your format.  You'll need to make sure that the "Export as Sequence" checkbox is selected.  Verify that the pixel dimensions are 2048x858 and that the frame rate is 24 (not 23.976, but true 24).  Check the "use frame blending" option since you're asking Adobe Media Encoder to perform an inverse-telecine operation.  Make sure to create a destination folder for your Left Eye image sequence and another folder for the Right Eye image sequence.

Next you'll be ready to setup your audio export.  This is a little convoluted since the DCP packing software requires that each audio file be a monaural file for each channel at 48 kHz and 24 bit.  It simplifies things if you create two more container sequences for this part, label one of them as Left Audio and the other as Right Audio.  Drag your original picture locked edit into each of these container sequences.  You'll now use a stereo audio effect called channel volume on each of these sequences.  For the left sequence you need to mute the right audio channel.  Do the opposite for the right sequence.

Now you're ready to export your two audio files; one for the left and one for the right.  Hit Command + M again and this time select Waveform Audio File as your format.  Make sure to choose the sample rate as 48000 Hz, mono channel, and the sample type as 24 bit.

Once you have all of these files ready to go, we can pack a digital cinema package.


  1. wouldn´t the audio files have to be pulled up by 0.1% after rendering for correct syncro?

  2. Hi Aldo,

    I'm refining this method, but as posted it does work since the export setting of true 24 from 23.98 sequences means that the audio is correct and the frames are being blended to keep time. It's an inferior method, but it also works every time. I've had mixed results using Audition to speed up the audio to match the frame rate if the 23.98 frames are exported as is and the DCP packing software assumes they are true 24 fps.

    I'm currently playing around with having Premiere interpret all of the source clips as true 24 (rather than 23.98) and that process applies the 1/1.001 speedup of the audio automatically. The rest of the workflow is then at true 24 and the frames are keps intact with no inverse-telecine process being applied.

    I'm still testing but when I vet this new method I'll be sure to post an entry about it.


  3. What do you think is the best way to pull up the 23.976 video from cameras like the Sony to 24.000? If you just comform a quicktime movie from source material in 23.976 to 24.000 in cinema tools, would it be the same as this Premiere process? I guess not, since the file stays the same, it's just the speed at which the system plays it that changes. Don´t know if the DCP creation software / hardware takes that as true 24.000.

    1. In reverse order:

      DCP packing software assumes that the frame sequence is at one of several frame rates if SMPTE packing is used: 24, 25, 30, 48, 60 (all true numbers, so no 23.976, or 29.97, or 59.94). The finished intermediate files, generally a DPX sequence and WAV files, must all conform to a valid cinema frame rate as defined by the DCI specification.

      Getting camera footage to these official frame rates can happen three ways that I'm aware of:

      1. (best) shoot at that frame rate, this usually requires a top shelf camera like a Sony F35 or F65, Arri Alexa, RED, Phantom, SI2K, scanning film

      2. (good) shoot at 23.976 (like the Sony EX3) and have your NLE interpret the footage as true 24 fps, it'll play 0.1% faster (the conforming option you mention)

      2. (average) shoot at 23.976 (like the Sony EX3) and export your frame sequence through an inverse-telecine to blend the frames up to 24 fps, it'll play true speed at the cost of blending frames through a 3:2 pulldown cadence

  4. Hi Chris,

    My name is Emmanuel and I have a short which I shot on a DSLR - it is a H264 1980x1080 footage which I edited on Premiere Pro 5.5

    I am trying to submit it to festivals and some of them require a DCP - NTSC

    Will you be able to outline what does the DCP package consist of?

    From reading your post I understood that it consists of 1 film file exported as a DPX and 2 mono channels in WAV format.

    I'd appreciate your help.

  5. Hey Chris, Emmanuel again, will you have the time to drop a couple of lines regarding formatting as a DCP please

    I'd really appreciate your help

    1. Hi Emmanuel,

      You have the gist of the intermediate process correct: you need to export your video as a frame sequence (use DPX if you shot 10-bit color or higher, use JPEG or PNG if you shot 8-bit color) and your audio as a separate file (old DCP software needs a single file per channel, the newer version can take a multichannel file).

      The trick is getting the frame rate to conform to DCI compliant standards, which mean one of the following rates: 24.000, 25.000, 30.000, 48.000, 50.000, 60.000 for 2D material at 2K resolution. Once you have the intermediate files you can then pack them into a DCP. There is commercial software out there that costs money and there are a few open source projects like:

      Some festivals will create a DCP for you if you're accepted, check with them for the type of file or files that they want. You can also hire a post-house to create a DCP for you for a fee:


  6. DCP requires a minimum of 3 audio channels; left, centre and right. 2 channel stereo is NOT supported on DCP!


  7. Hi Greg,

    Can you cite your source for this information regarding number of channels? We've packed over two dozen DCPs and they support from 1 to 16 audio channels as per the current DCI specification version 1.2 here:

    While up to 16 channels can be supported, there are a few "standard" configurations that are guaranteed to work in a DCI compliant cinema: 2.0 and 5.1 will always be decoded correctly if the cinema is fully DCI compliant. 7.1 is also supported in the DCP packing software, though decoding at the cinema could be SDDS 7.1 (5 across the front) or consumer 7.1 with the two additional rear channels.

    I've just tested a monaural 1.0 DCP and while valid, there is a warning flag that some screen servers may misinterpret the 1.0 audio essence. So, packing a 1.0 audio essence into a 5.1 matrix with silent tracks on the non-center channels ensures proper decoding on all DCI compliant equipment while preserving the original monaural program.

    Please reply with your citation as to the number of audio channels supported. Whomever wrote that should go and read the DCI specification before posting again:


  8. This plug-in makes life easier:

    1. Thank you for the link, this plugin looks to be a great resource. It looks to be 149 Euros for the non-watermark version, not too bad.

      I'll be updating this post soon since I've learned a LOT about DCP preparation since this post was originally written. It's a far smoother process now.

      I should also note that I've been testing the j2k plugin from fnordware ( as a free plugin that allows for DCI compliant 2jc codestream exports directly from PP and AE. Once the j2c codestream is exported it is then possible to easily and quickly pack a DCP with OpenDCP (