Tips on converting Film Festival programmes to the DCP (Digital Cinema Package) format.

This year I converted all films that was not 35 mm or DCPs at the Films from the North 2011 section at Tromsø International Film Festival 2011 to DCPs. In total I made 33 DCPs (about 10 hours) in two weeks. I was also responsible for screening the DCPs at the film festival. I checked the sound level on each DCP in the cinema theatre and wrote down what level they should be played back at. I also wrote down what audio preset should be used (5.1 or 2.0 with Pro Logic decoding). If you are converting to the DCP format for screenings where you will not be present I recommend using standard cinema level 5.1 sound mixes.

Here are some tips if you are thinking about converting a film festival programme to DCPs:

    I recommend using batch processesing when making DCPs of short films, I used Fraunhofer Easydcp + to make 1-5 DCPs each night.
    Have a tested worfklow before the festival films arrives. It must cover both interlaced and progressive material. Standard definition and high definition. Stereo and 5,1 sound. Material that has subtitles outside the safe zone and material that does not use a standard resolution. . .
    Use different quality control stages, I recommend two computer screens so you can look at the material fullscreen 1:1 . I recommend using the RGB parade scope in Apple Color/Resolve to check how the source is color corrected/graded. I recommend checking the DCPs in a movie theatre, preferably the screen you will screen the film at or one that is suited.
    Find out how much the picture is cropped at the theatre where the films are screened. Also check for a 1920×1080 1.77 preset.
    If using 25 fps or 30 fps DCPs, check if they work in the theatre that will be screening the films.
    If using DCPs with stereo sound make sure the theatre has Dolby Pro Logic 1 or 2 decoding or something similar to get the speech in the center channel. If a stereo soundtrack is played in stereo everyone on the right side of the screen will hear speech coming from the right and those one the left will hear it coming from the left.
    In total I made 33 DCPs (about 10 hours) in two weeks. I always had 1-5 DCPs being made each night that I tested the next day. I could convert 1,5 hours from one day to the next with easydcp + on a quad core computer. Some people uploaded their films as Prores quicktime to a FTP server or used Filemail Corporate instead of delivering on Digibeta or HDCam. This helped speed things up.

Some ideas for best common practices when mastering DCPs

Before sending a hard drive with a DCP at least check the DCP with the free or full version of Fraunhofer Easydcp Player. That way you will see if the DCP fails the basic tests done by Easydcp player like “Check file hashes” and that the disk works and have the correct DCP on it. The Doremi and Sony (with recent firmware) accepts NTFS disks, otherwise a ext2/ext3 linux disk could be used.

If a movie uses a workflow like the movie Inception where there is no Digital Intermediate (DI) and the film is color timed photochemically in a film lab, the color timed print has to be scanned and fine tuned for 2.6 gamma 12 bit linear DCI P3/XYZ color space to make the DCP. Otherwise the 35 mm print has to be screened. The offline edit or a SD video tape should not be used to make a DCP for theatrical distribution.

A DCP should be made from the DCDM with correct XYZ colorspace and 2.6 gamma made from the Digital Intermediate (DI) that has been graded on a DCI P3/XYZ color space projector with 14 footlambert of light. The DCP is normally made from the emulation of how the 35 mm print will look on a digital projector. This emulation (3D LUT) is baked into the DCDM. The resolution should be 2048×858 or 1998×1080.

A DCP could also be made from a file based master in 10 or 12 bit, Rec. 709 HDTV colorspace graded on a HDTV monitor. The Director of Photography should approve of how this looks in the movie theatre because a HDTV monitor is not ideal because you don’t typically have 14 footlambert and a dark room when grading on a HDTV monitor. But the results you can get from a 10 bit HD master is good enough for films like The Social Network so it could be great when done correctly. The resolution should be 2048×858 or 1998×1080. If you grade the the film in rec. 709, the movie could have less brighter colors in the 35 mm version than in the DCP version.

The sound must be 5.1 (85 db). If the theatre you will screen the DCP at accepts a stereo mixed DCP you could deliver that, but otherwise it has to be 5.1.

The Digital Cinema Naming Convention must be used to name the DCP.

Tromsø International Film Festival/ Films from the North 2011 Deliver your HD master and we will convert it to a 24, 25 or 30 fps SMPTE Digital Cinema Package (DCP).

We have posted what deliverables/file formats are accepted for the Tromsø International Film Festival/ Films from the North 2011 festival programme

Translated to english:
Deliver your HD master and we will convert it to a 24, 25 or 30 fps SMPTE Digital Cinema Package (DCP) with Fraunhofer Easydcp.

Fileformat: Prores 422 HQ 10 bit Quicktime.
Image sequence: DPX 10 bit, TIFF 16 BIT (Adobe After Effects).

Frame Rates: 23.97, 24, 25, 29.98 fps.

Resolution:
1:1.77 = 1920×1080
1:1.85 = 1998×1080
1:2.39 = 2048×858

Color space Rec.709 HD or DCI XYZ
Sound 5.1 (or stereo)

DCPC, a frontend to open source DCP tools.

An free alternative to Easydcp and Cineasset that makes DCPs from image sequence and wav files is the Windows DCP tool called: DCPC. It is made by Lars Reichel.

Updated: DCPC was buggy and could crash on longer clips, but the maker Lars Reichel reports in the comment section that the new 1.4 version is more stable. It supports 25 fps, and can rewrap both 24 and 25 MXF files with missing or wrong PKL or CPL metadata files. I have not tested DCPC 1.4 except the rewrapping feature. The Easydcp software that I use has a faster jpeg2000 encoder.

Link to the german homepage for DCPC

Link to the german projectionist web forum thread where DCPC was first posted

To use it you download :
dcp-tools

and extract the files to some folder with winrar or 7-zip
and install at least ImageMagick and Goldwave.

Then you download

DCPC

and extract the files to some folder with winrar or 7-zip

Double Click Digital Cinema Package Creator
and in File-settings
choose the tools folder in the folder where you extracted DCPC.

I tested it on some small clip in Windows 7 and it worked.
On Windows XP you need to install Microsoft .NET Framework 3.5

Avoid blank spaces in the filenames.

(updated: this is now fixed)
I had to use a small fix to get DCPC To work,
I had to open the wav files
in the goldwave sound editor that is bundled with DCPC and save as “(obsolete) PCM signed 24bit” .
This fix was described here:
Have heard reports about DCPC crashing on longer clips,
but it has a feature called “Recovery start” that will let it start again where it crashed.
You may need to delete temp files.

Notes on DCPC

DCPC can make a DCP from sound and picture MXFs. So you could use this software to fix a DCP made from one audio MXF and one video MXF that has broken metadata (eg. someone has edited the CPL file and the DCP will not ingest because there is a mismatch between the hash values of the original files and the edited files.)

You can also use asdcp-test to wrap the jpeg2000 and wav files in MXF containers like I described in this post: and then use DCPC to make either a 24 fps or 25 fps SMPTE or interop DCP in 2D or 3D.

DCPC can make a 2D and 3D DCP from DPX or TIFF image sequences and wav files.
Like the sequences I made here:

I have not tested the Gamma/ color space conversion.

DCPC can also make DCPs from one still picture or multiple still pictures.

DCPC uses either 2, 4, 6, 8 or 16 openjpeg 1.3 jpeg2000 encoders at the same time, but it is still much slower than the Fraunhofer Easydcp jpeg2000 encoder.

Tips on DCP resolutions and some other things

edit 4
Summary:
DCPs uses these standard resolutions:
2K Flat 1.85: 1 1998×1080
2K Scope 2.39:1 2048×858
4K Flat 1.85: 3996 x 2160
4K Scope 2.39: 4096 x 1716
-Getting the resolution correct is important.
-Flat and Scope are the standard presets in all cinemas.
-1.78:1 (16/9) and 2.35:1 films can get image artifacts (trapezoid shape) in the cinema because the standard side masking presets are made for 1.85 and 2.39.
-Scope films are projected wider than Flat films in a common height cinema.
-A 4K DCP is compatible with 2K and 4K projectors.
-Full container (C) 2048×1080/4096×2160 is not used by Hollywood feature films. Films made with these resolutions will probably be projected in Flat with cropping on the sides at most cinemas.

edit 3: This post is a bit technical. Basically what I am saying is that you will lose pixels in both constant height and constant width cinemas if you don’t use either a Flat 1998×1080 or a Scope 2048×858 resolution when making a DCP. In some cinemas 1920×1080 can’t be masked on the sides and you will loose some perceived contrast but you will not lose any pixels.

 

Posted this at a webforum:

HD workflows where the 1920×1080 container of HDCAM-SR tape is standard should be changed to 2K workflows for digital cinema work.

The 1.78 aspect 1920×1080 fits ok, but 1.85 or 2.39 inside a 1920×1080 container does not fit.

The RED 2K 2:1 2048×1024 resolution does fit the 2.39 digital cinema presets with cropping to 2048×858 (if this was intended when shooting), but to get to the 1.85 digital cinema preset you have to crop and scale to get to 1998×1080.

1920×1080 1.77 material can have black padding on the sides to achieve 1998×1080 or be scaled and cropped. Digital cinema servers can play 1920×1080 but the standard is 1998×1080 so it is preferable.

The reason that 1080 material fits better than 1024 material is that the side masking can be adjusted in a constant height cinema. I would recommend padding to 1998×1080.

If all the 2048×1024 information is vital, scale to a 1998×1080 container, but remember that having this much black padding above and below the picture kills the contrast.

The main problem when making 24 fps DCPs from 25 fps material, even with software like time factory 2 which keeps the channels in phase and keeps the pitch, is that they could alter the sound mix in the conversion process and introduce artifacts. If you slow down the sound you get a pitch change. So being able to play 25 fps DCPs on the doremi cinema server is great.

I did use the openjpeg encoder first to make DCPs but found it too slow for time critical use.

The biggest problem with mastering DCPs is the Quality Control stage. The best option is to check it where it will be played, if you get a DCP made or checked at a post house make sure they know what they are doing and tell them exactly what you want and what equipment the DCP will be played on.

edit: Answered a comment about DCP resolution:

All digital cinema projectors have one preset for Flat (1998×1080) and Scope (2048×858).
In a constant height theatre the resolution, the side masking and zoom changes when projecting the two presets.

The flat preset has 1080 pixels in height, the Scope preset has 858 pixels in height.
When going from Flat to Scope the lens zooms and the masking widens.
The pixels beyond 858 would have hit above the screen if they were not masked in the projector.

The flat preset has 1998 pixels in length, the Scope preset has 2048 pixels in length.
When screening 2048×858 in Flat the pixels beyond 1998 is masked in the projector and the image is letterboxed.

That is why you have to scale 2048×1024 down to fit the Flat preset or crop it to fit the Scope preset even though 2K is 2048×1080.

You should always avoid black padding/letterboxing, it kills the contrast.
That is why 1998×080 or 2048×858 should be your target.

It seems you have been to a constant width theatre where Scope is much smaller than Flat. This is a terrible solution. But the common width theatre could also just have presets for Scope and Flat. Those should always be your targets.

edit 2: Answered another comment about why you are going to loose pixels when using 2048 x1080 resolution.

At this lens searcher website from Sony you can see that projectors and masking is programmed with a minimum of Scope and Flat presets and that theatres are either constant height or constant width:


http://assets.sonybiz.net/Web/DCinema/10-02-17%20DCinema%20Lens%20Finder/TheatreLensSearcher/information.html

You can also see that some cinemas have both side and top masking. A constant height theatre could also have top masking so theoretically a 2:1 ratio film in a Flat container could be masked. Presets for masking 1.33:1 and 1.78:1 is common. 2048 x 1080 will not fit any of the presets. A custom preset could be made in a constant height theatre if you program the side masking to go a little further than flat and remove the option to mask away pixels beyond 1998 in the projector. In a constant width theatre the custom preset could be programmed to use the same zoom as the scope preset, remove the masking of pixels beyond 858 in the projector and make a custom top masking.

How to use Open Source DCP tools to change the sound in a DCP (or make a DCP)

Outdated post: You can now open DCPs and change the picture and sound in Easydcp Creator.

I sometimes use the command line tool asdcp-test from Cinecert to change the sound in a DCP instead of making a new DCP from scratch in Easydcp Creator. This comes in handy if the DCP takes 8 hours to make and you are screening the movie in 2 hours and the sound must be changed. Another option is to encode the jpeg2000 sequence seperately in easydcp creator + or extract the jpeg2000 sequence with Easydcp Player or asdcp-test, then you can make change sound or subtitles and make a new DCP in Easydcp fast. You could also make a complete DCP with free tools like asdcp-test and opendcp.

To compile asdcp-test yourself:
Source for asdcp-test/ASDCPLIB:
http://www.cinecert.com

There are also some old binaries for Windows of acdcp-test, Openjpeg and opencinematools here:
http://code.google.com/p/opencinematools/

You can use http://code.google.com/p/opendcp/ instead of opencinematools. It includes Color space conversion.

To change the sound in a DCP:

Copy the compiled asdcp-test files to /usr/local/

Generate 4 GUID
with e.g. http://www.guidgenerator.com/online-guid-generator.aspx

Keep the movie mxf file with the jpeg2000 sequence from the first DCP and name it with the first GUID: 5f243728-ef5b-42ca-bad7-47b59b2aea01.jp2k.mxf

Then make a new sound MXF and name it with the second GUID: 47365f11-3a45-4f77-8389-f4a145aa232b.pcm.MXF with the 24 bit 48 khz wav sound files in this order: L,R,C,LFE,Ls,Rs.

/usr/local/bin/asdcp-test -p 24 -L -l 5.1 -c 47365f11-3a45-4f77-8389-f4a145aa232b.pcm.mxf sound.L.wav sound.R.wav sound.C.wav sound.LFE.wav sound.Ls.wav sound.Rs.wav

Then make the new metadata with opendcp
Use http://digitalcinemanamingconvention.com/

Update:
You may also use the windows software DCPC to make the CPL and PKL files metadata files.

Copy and paste these with the correct names to a bash script
or execute each command manually in terminal in the folder with the mxf files

Start the Bash script with
#!/bin/sh
and use .sh in the filename, e.g. makedcp.sh

Execute it the folder with mxf files with the terminal command sh makedcp.sh
or execute each command manually in terminal in the folder with the mxf files.

This is the updated commands I used to make a 25 fps DCP:

Use terminal or make shell scripts that run these commands:
/usr/local/bin/image_to_j2k -cinema2K 24 -ImgDir foldername with Gamma and colorspace converted TIFF sequence/ -OutFor j2c

/usr/local/bin/asdcp-test -p 25 -L -c movie.jp2k.mxf foldername with jpeg2000 sequence

/usr/local/bin/asdcp-test -p 25 -L -l 5.1 -c movie.pcm.mxf folderwith5.1sound/sound.L.wav folderwith5.1sound/sound.R.wav folderwith5.1sound/sound.C.wav folderwith5.1sound/sound.LFE.wav folderwith5.1sound/sound.Ls.wav folderwith5.1sound/sound.Rs.wav

use opendcp or dcpc to package the dcp.

This how it was done in Opencinematools
/usr/local/bin/mkcpl --kind feature --title test25 --annotation test25 movie.jp2k.mxf movie.pcm.mxf > movie.cpl.xml

/usr/local/bin/mkpkl --issuer KESE --annotation test25 movie.jp2k.mxf movie.pcm.mxf movie.cpl.xml > mymovie.pkl.xml

/usr/local/bin/mkmap --issuer KESE movie.jp2k.mxf movie.pcm.mxf movie.cpl.xml mymovie.pkl.xml

How to make a SMPTE 25 fps (or 30,50,60 fps) DCP (Digital Cinema Package) with Fraunhofer Easydcp 1.4

UPDATE: In Easydcp 2.0 you can make DCPs from (Prores 422) Quicktime files. Prores 422 exported from Color do not get darker.

Tested a 25 SMPTE DCP made with Fraunhofer Easydcp Creator 1.3 beta on a Doremi DCP-2000 cinema server. It worked, I am not sure if all cinema servers supports 25 fps SMPTE DCPs yet. I know the Doremi and Sony does. Update: Have also tested a 30 fps DCP on a Doremi DCP-2000. It also worked. But it may not work on all projectors. To convert the sound in a 29.97 (no inverse telecine) video file to 30 fps you could use the cinema tools tips from this post:
How to make dpx and wav files that can be converted to a DCP (Digital Cinema Package) with Fraunhofer EasyDCP Creator using Final Cut Studio 2 or 3

Links:

Fraunhofer Easydcp Creator

Final Cut Studio 3 with Apple Color and Soundtrack Pro (Discontinued in 2011)

Note: Check that the DPX sequence from Color matches your monitor. You may have to compensate that the picture gets darker.

Update: To export a DPX sequence from After Effects follow this step by step guide.
In Apple Color:

To export a image sequence from Apple Color you have to work with a supported file format like Prores 422 HQ.
MPEG Streamclip is good tool to convert file formats like VOB, WMV, MPEG to Prores 422 HQ.
MPEG streamclip

MPEG streamclip can also do a quick deinterlacing if the clip is interlaced.
Alternative deinterlacers are:
JES deinterlacer, Final Cut/Compressor, Color, Adobe After Effects.

To convert a WMV file you have to have Flip4mac WMV player Pro:

Step 1.

Export a TIFF or DPX image sequence Hopefully from an 10 or 16 bit application like Apple Color or Adobe After Effects. Quicktime Pro 7 is as far I know not capable of more than 8 bit.
Note: Check that the DPX sequence from Color matches your monitor. You may have to compensate that the picture gets darker.

In Apple Color:
Import shot. I let Color match the timeline to the clip. This way the framerate will be the same as the clip.
import shot

Change the Project Settings
Render File type: DPX
Width and Height : All DCI digital cinemas have two presets:
Scope (2.39:1) = 2048 x 858 and
Flat (1.85:1) = 1998 x 1080. HD
(1.77:1) 1920 x 1080 will fit inside the Flat preset with perhaps some empty space on the sides.
I let Color scale my clip to the preset.

Broadcast Safe: If you turn this off you will see in the RGB parade scope if material is graded with values above 100 and below 0. The material should be graded so black is not below 0 and white is not above 100. Everything below and above will be clipped. If you turn broadcast safe on everything will be forced inside the legal 0-100 values.

Choose a render folder with a lot of free space. Each frame is about 8 mb.

Sometimes you want to fix the geometry. Scope film inside a HD container (1920×800) will have to be zoomed and RED (2:1) material in 2048 x 1024 does not fit either of the presets.

Example: zooming in on a scope film inside a HD container. Choose 2048×858 in project setting and zoom in on the Geometry tab.
Remember to check that the project has the correct framerate.

Finally you want to export the image sequence:
In the Render Queue. click on the the clip, click add selected and start render. Then the start render button will turn into rendering,

Step 2.

Make the sound files

Drag the video file to a multitrack project in Apple Soundtrack Pro
drag video file

Click on the sound to highlight it and export it (cmd + e)
Use these settings:
Exported Items: Selected
File Type: Wave File
Create multiple mono files
Bit depth: 24 bit
Sample Rate: 48 kHz

Watch out that you have not just selected a part of the sound.

Step 3

Make the DCP

Start Fraunhofer Easydcp 1.4

Choose a Project name
2K
uncheck “Use interop instead of SMPTE packing”
Choose framerate 25 fps
and Create
new project

Add picture track

Choose the first DPX image from the Color render to select the image sequence

Add sound track

Use the sound files from Soundtrack Pro export

trackfiles

Drop the track files in the new reel window.

Give the project a name and double click the perforation to the right of the name input field to edit the composition settings. Choose content type, eg. Feature or Trailer.
Choose a annotation text
Choose a Content title,

new reel

uncheck Options – Automatic pillarboxing -/letterboxing.. If you have a valid resolution.

Check that the frame numbers match and click the Generate DCP button.

Sometimes the frame count is off, but this can be fixed with the offset option.

How to make dpx and wav files that can be converted to a DCP (Digital Cinema Package) with Fraunhofer EasyDCP Creator using Final Cut Studio 3

UPDATE: In Easydcp 2.0 you can make DCPs from (Prores 422) Quicktime files. .

Example:

Convert a short film in 1920×1080 16/9 25fps 10 bit Rec. 709 ProRes 422 (or 4444) with 5.1 sound to a standard DCP (DCI, 24 fps XYZ colorspace, 2.6 gamma, 5.1 sound) .

Note: Check that the DPX sequence from Color matches your monitor. You may have to compensate that the picture gets darker.

Update: To export a DPX sequence from After Effects follow this step by step guide.

Use Cinema Tools to convert to 24 fps

  • Make a copy of the ProRes file (or just rename it ….24 fps.mov) and open it in Cinema Tools
  • Meny: Clip – Open
  • Meny: Clip – Conform: 24 fps

Video

Open the 24 fps conformed Cinema Tools Prores file in Apple Color to export a DPX sequence

  • Import the 24 fps conformed prores file
  • Go to setup tab- Project settings
  • Change Render Dir to where you want to render
  • Resolution Presets: 1920 x 1080 (or 1998 x 1080)
  • Render File Type: DPX
  • Printing Density: Linear (0 Black – 1023 White)
  • Go to “Render Queue” Tab
  • Click “Add selected” button
  • Click “Render” button

Sound

Open the 24 fps conformed Cinema Tools prores file in Soundtrack Pro to export wav files (without pitch shift)
Update:Another option is to use Wave Agent and Audacity to do this

  • In Soundtrack Pro: Drag the Prores file from browser to a Video track in a Multitrack Project.
  • Select Audio Track
  • Export
  • All Tracks, Wave, multiple mono, 24 bits, 48 khz

These files can now be used to make a DCP with Fraunhofer EasyDCP Creator. Choose the first image in the DPX sequence to make the picture part. Compare the frame count with the sound part, they should be the same.

Link to Fraunhofer EasyDCP Creator (mac/pc)


To ingest a DCP on Doremi and Sony Cinema Server you can use a NTFS disk.

Update: Or you could copy the DCP to a linux ext3 harddrive

Fraunhofer EasyDCP assumes the DPX files are in the REC.709 2.2 gamma. You should check how it looks in a cinema and adjust your settings or make a 3D LUT.

Other stuff



Updated : This step is no longer necessary. Easydcp 1.2.0 support dpx sequences from Color and 16 bit tiff sequences.Convert DPX (changes endian type and dpx version)

Install Macports http://www.macports.org/

  • Open Terminal
  • Install ImageMagic with this command
    sudo port install ImageMagick
  • Go to directory with DPX sequence eg.
    cd /Volumes/RAID/dpx/short1/1
  • convert dpx with this command


find . -iname "*dpx" | while read f;do mogrify -verbose -depth 10 -format dpx "$f";done

Or try this: (thanks Eirik Schwenke. see comments for more info).


find . -iname "*.dpx" -exec mogrify -verbose -depth 10 -format dpx "{}" \;





Using Soundtrack Pro with pitch shift

Stretch the sound from 25 fps to 24 fps.
24 fps at 48 khz is 2000 samples per frame.
25 fps at 48 khz is 1920 samples per frame.
23,97 fps at 48 khz is 2002 samples per frame.
Example:
50000 samples 25 fps / 1920 = 26,041666 frames
26,041666 frames x 2000 = 52083,3333 samples 24 fps

  • Use Process – timestretch to new samplerate.
  • Save as Wave, 24 bit 48 khz. This works best if the 5.1 mix uses the center channel only, otherwise phase sync could be lost. Time stretch works best on the source audio files.


Notes on time stretch/pitch shift:

To keep phase sync on a 5.1 mix or music mixed in stereo try Prosoniq TimeFactory 2. Quote from their website: “Version 2 supports all popular file formats and offers 5.1, 6.1 and even N.1 surround processing in perfect phase sync – converting between different frame rates for film and TV has never been easier!”

The pitch change is: 0.706724268642822855 semitones

In Timefactory you can use the original sound from the 25 fps file and convert between 25 fps to 24 fps to get the correct pitch shift. You can also keep the formant with the natural setting. To get a good result on more difficult mixes I recommend using a professional.

It’s easier to do a pitch change when going from 24 fps to 25 fps.

If the 24 fps mix is the most important, most of the sound mix could be done in 24 fps natively. Then change the pitch to do the 25 fps mix.