Post-Production : Post-Production, DCP delivery by Georgia Hilton

I was asked today about creating a DCP delivery for a small indie project... You want to make a DCP delivery from your finished project? EASY ANSWER: You don’t’ have to worry about it. You just deliver to the DCP mastering facility at 24/48 or 23.976/48 and they will take care of it. Done. Easy. HARD ANSWER: First I preface this with: DO NOT TRY THIS AT HOME FOLKS… I AM A TRAINED PROFESSIONAL, AND I WON’T DO THIS AT HOME, I DO ALL THE PREP FOR MY CLIENTS AND THEN GO TO THE LAB AND SUPERVISE THE CREATION… IF YOU ARE CREATING A DCP DELIVERY HIRE A PROFESSIONAL !!!!!!!! Here the the primary steps to create DCP. Content normalization – Importing/Converting content into JPEG2000 frames and PCM WAV files. Color space Conversion – Converting content into the XYZ color space. Audio Conversion – Converting audio to match video frame rate and wrapping audio in MXF Container MXF Container – Putting MXF wrappers around your content. XML Descriptors – Generate XML files to describe content for ingest and playback. Video is encoded in DCP to JPEG2000 image 2K 4K or stereoscopic sequences and audio is uncompressed PCM WAV container at 24bits 48Khz/96 Khz… All of this is wrapped in MFX containers as described in the DCP docs. So the first step is to take the video file ( whatever it is ) and convert to JPEG2000 tiff images, but you also need to convert color space… Then you can import the JPEG2000 images into DCP software like OPEN DCP and set it to play at 24 or 30 for safety or you can leave it at the other frames rates as available. The best way to do this at home is to convert the video to the closest you can get to DCP spec prior to creating the DCP container using a tool like FFmpeg. (the real issue is where the Composition play list will be shown and does that location support the appropriate frame rate) Setting the frame rate to 24 is safe. If you do this then, you have to covert the audio by pulling up or down by the appropriate amount (just like a 35 film project). DCP plays back at true 24fps. 24/25=0.96 Meaning 24fps is 96% the speed of 25fps, therefore a slowdown of 4% is required for the audio to remain in sync at 24fps. 24 –> 23.98 (which is equivalent to 29.97)=-0.1% slower. The same process for any other conversion, just different numbers. Delivery for DCP is like any other delivery… No difference for the post audio professional, or even for the Visual editorial. The difference comes in the DCP “PRINTMASTERING” I’ll use this term if though its incorrect because it is something we all understand. The problem does not lie in frame rate conversion, or any other specific item… It lies in the details. To deliver Quality DCP you need to be very careful and you need to do QC. You need to do tests during the DCP creation to verify color space conversion, frame rate conversion, audio speed conversion and the like, not to mention all the materials for the metadata. To create a DCP you start with HD files, then rescale to 2K or 4K, then color convert from RGB/YUV to XYZ, the encode to JPEG2000, then create the video steam and MXF wrap and then create the DCP elements… Yuck…. DCP is built on a hierarchal file structure. Elements are between 10 and 20 min in length. Each asset is separate and can be replaced or added at will without affecting any other element within the confines being delivered correctly. DCP then simply becomes a system that runs off a ‘Play List”. The play list can be the film reels, ads, trailers, whatever. The reels can be the film and the English audio, or German or whatever you choose and whatever has been delivered. These play lists are called Compositions Play list. The Composition Play list is sent to the Theater and the TMS /SMS ( management system ) assembles the SHOW play list at the theater. During this process METADATA is added to the elements. The metadata includes physical FORMAT information and other specific info required by the DCP systems. This Meta data come from the DCP creator as does the files. There is just too many things that you can screw up by trying to create DCP elements and the DCP. Including, BTW, even the drive delivery spec… I recommend a Linux EXT drive versus a MAC or PC formatted drive. So bottom line… It’s the same work flow as any other postproduction project. When you are ready to print master you are ready to deliver DCP ( as long as you have the proper reel lengths, and BTW the reel length limit is very real in DCP) Side note for the Geeks in the audience: JPEG 2000 uses Wavelet compression with a quality that is much better than standard JPEG. Wavelet compression works on a similar principle to DCT in that it assumes that parts of the image that are close to each other are likely to be similar and can therefore be grouped to reduce the amount of information with a minimal effect on the image quality. This is why JPEG2000 was selected. The reason MFX wrappers are used is simple that it is not specific to a company and it’s easier to move, and keep track of a single container file versus a gazillion frame files. XYZ color space was selected because its more “friendly” to human eyesight than RGB or CMYK. RGB being easier for computers and CMYK for print. Obviously the post world is headed to WAV file formats and wrapping multiple WAV files in an MFX wrapper works quite nicely. Bottom line… I recommend you hire a trained professional to create your DCP materials… I’d hate to see your project get screwed up at the very last step. cheers geo www.hiltonmm.com