Building a DCP (Digital Cinema Package) with Vegas

entilza72 wrote on 3/22/2013, 8:45 AM
Hi gang,

I've just come out of a project where I built a DCP of a project I'd edited on Vegas. For those that do not know, a DCP (Digital Cinema Package) is how cinemas play films these days. You no longer have to pay tens of thousands of dollars to strike a print - you can DIY.

The Vegas project looked fantastic on the big screen!

I found solid DCP information hard to come by, especially for Vegas, so I promised to return and paste up what I found out. Here goes!

WARNING - I'm not an expert. I simply lived through the project and can tell you what I did to get a reasonable result. Some of this was obtained by reading, other parts by asking, and others again were confirmed through trial and error with a very helpful cinema.

Vegas Config:

Video: Set your project to a 2K template (or 4K if your source material is 4K). Use either a "Widescreen" (16:9) template, or build a "Scope" 2.35:1 template if your project is 2.35:1.

Pixel format: 8 bit.
There's some contention around this choice. See discussion below. There's a bug in current versions of vegas with Sequenced Image output gamma level for 32 bit projects. So if you want more than 8 bit, you'll need to render a video that supports that, then slice it up into images somehow. There's also considerable debate in these forums on what (if anything) 32 bit yields you anyway.

Frame rate: 24 fps. You might be able to get away with 23.976. Try for 24. A lot of cinemas do not support frame rates outside of 24fps as they are running older servers.

Audio: 24 bit, with either 96,000 Hz sample rate or 48,000 Hz. They are the only two supported sample rates.

Of course, you can go HIGHER than these settings in the project, but please make sure your DCP project files (your Vegas renders) conform to these specs.


Video: Output the video as an Image Sequence, in DPX format. I found all other formats output as 32 bit images (even though its an 8 bit project) and the DCP packager rejected those. You will not need to set any render settings with an Image Sequence - click Render and wait.

N.B. DO NOT APPLY Computer RGB -> Studio RGB to your tracks or your project's master Video Output. Leave the video with Computer RGB levels. The DCP is able to use the extra headroom provided by Computer RGB.

For Stereo projects, render a 24 bit stereo WAV in either 48,000 HZ or 96,000 HZ.
For 5.1 Surround Sound (6 channels), take the WAV "44,000 Hz 24 Bit, Mono, PCM (multiple)" template and convert it into either a 48,000 Hz or 96,000 Hz template. Render.

That's it! Now time to make the DCP. I used DCP Builder, which you'll need to register beforehand. But, at least it is free.

In DCP Builder:

These instructions are for DCP Builder only. If you use something else, the same basic principles will probably apply.

Use New Project -> Simple Wizard to set up your config.

When it comes time to add the Image, select the first file of your DPX sequence. Note: I was using an external USB based HDD and was having trouble with DCP Builder thinking I wasn't registered at random points during a render. When I used my system drive, this went away.

When it comes time to load the audio file(s), select only the LEFT file. The wizard will continue on with other things.

When the wizard is finished, you'll need to go to the "INPUT" tab and populate the rest of your 5.1 files.

While still in INPUT:

Colorspace: sRGB.
(I wasn't sure about this, as it is based on what the colorspace is of the source image file, and it appeared DPX could be any colorspace the creating software set. Reading file metadata, I was able to confirm Vegas sets all OTHER image types to sRGB, but this info wasn't available in the DPX file. So I assumed sRGB. Further, I tested sRGB and BT.709 in the cinema and sRGB looked correct, while BT.709 looked a little washed out)

Confirm the other details.

Moving to the DCM tab:
Set the film's Global Duration in Frames (if it isn't already). Get this info from the Vegas timeline.

Set the Reel Duration. We want each reel to be LESS than 2GB. I recommend around 1.7GB to be sure. This is so we can use media that isn't able to support 2GB files (eg, FAT). A DCP server will play back your reels without a jump or stutter in between, so this is the perfect solution.

Set your total Bitrate to be less than 240MBPS. I recommend less than 200 for 2K, but you can probably do a LOT less than that.

Set your Aspect ratio: "HDTV" for 16:9 (known in cinemas as Widescreen), or "Scope" for 2.35:1 (known in cinemas as Scope).

That's it! You can use Peter Wimmer's Stereoscopic Player to play back individual DCP reels to test the vision and sound. And be aware that DCP Builder by default does not bother to convert the preview image back into sRGB for you, as it takes extra time - your preview image will look green (don't panic). You can change this setting if it worries you.

Transporting your DCP:

Using an empty USB key, copy your project onto the key's root directory, using the default formatting the key was made with. Beware of project size! My 10 minute project weighed in at 15Gb. If you can't use a USB key, you'll need to format a LINUX Ext2 filesystem on an internal HDD, and mount it in a caddy for the cinema. That is outside the scope of this article! :-)

At the cinema:

Most (all?) DCP servers will be able to read the USB key. A 15 Gb project will take around 50 mins to ingest.

Finally, you will need to tell the cinema what aspect the project is in, in their language!: Scope, Widescreen or Flat. This is important as both the projector's lens AND the mask curtain will need to change.

That's about it - I hope this is helpful to someone.


corug7 wrote on 3/22/2013, 12:43 PM
Thank you so much for posing this information. No doubt it will help some of the independent filmmakers among us. You are correct that there is very little information readily available about this subject and I certainly appreciate your contribution. Nice work!

[r]Evolution wrote on 3/22/2013, 3:28 PM
Wow. Crazy to think someone can now do this from their bedroom at a mere fraction of the cost from yesteryear. (and I complain about the kid with the iMac & iMovie stealing my potential clients)
Tisso Shark wrote on 3/22/2013, 10:50 PM
Thanks for sharing this information. DCPs were a mistery and this post brings some light into the topic.
K-Decisive wrote on 3/23/2013, 2:13 AM
Thanks this is great stuff,
I had one question. Your are rendering at 8 bit, I guess I was expecting something larger ( 12/16/32 float). Did you find something that spec'd it at 8 bit? Or did your project have 8 bit source material, so it wouldn't matter.
Forgive me if it sounds like a dumb question.
Marco. wrote on 3/23/2013, 6:55 AM
A DCP JPEG2000 should have 12 bit pixel depth. If you couldn't use more than 8 bit color signals for shooting, CGI and/or editing you're fine using an 8 bit Vegas Pro project and leave the 12 bit JPEG creation to the DCP software (of course this won't raise quality, it's just to fit the technical requirement then). Otherwise you should use a float-point project in Vegas Pro.
entilza72 wrote on 3/23/2013, 10:03 AM
OK - this comes with a warning - I'm not 100% sure I know what I'm doing in areas like colorspaces and bit depth, but ...

1. As I understand it, "8 bit" refers to 8 bits per channel, in other words 24 bit. 32 bit refers to 8 bits per channel PLUS an 8 bit alpha channel. If that holds true, then IMHO, SCS really should lose the 8 bit label and call it 24 bit, which would be consistent with the 32 bit label. It sure used to confuse the heck out of me, being an old 8 bit graphics guy. :-)

2. If the above is true, then I *guess* (note the use of the word guess!) that means 32 bit floating point projects do not offer extra colour information, just alpha channel info. But they do seem to offer a lot more pain.

3. My project was a 24 bit source material in Prores 4444 files, so 8 bit (per channel) was the correct config - nothing was being lost.

4. My project was initially in 32 bit floating point video levels, but there appears to be a bug in all recent versions of Vegas (very long thread, look for the posts from LoTN) that causes sequenced images in 32 bit floating point projects to be output with the gamma turned all the way up, leaving you with unusable images. You could try processing the output and running the gamma back down, but you'll probably be losing data.

And because we need to render a Sequenced Image, that is why I've said to use an 8 bit project.

Any of the above could be wrong or misinformed. :-) I reserve the right to strike out incorrect statements if proven later, as I hate contributing to misinformation. :-/

Marco. wrote on 3/23/2013, 10:44 AM
Float-point processing is quite a different thing than any kind of integer video processing. The terms "8 bit" and "32 bit" are confusing here.

If you use an 8 bit Vegas Pro project it means video processing will be based on 8 bit integer values: There are 256 steps for luminance information available.

If you use a 32 bit float-point Vegas Pro projekt it means there will be no integer processing, no limitation to 256 values.

These kind of 32 bit has nothing at all to do with using an alpha channel or not. It means the processing is based on float-point math and there will be 32 bit available to calculate any float-point value. 1 bit is used to determine whether the pixel value is positive or negative. 8 bits are used to determine an exponent. And 23 bits are left to determine the mantissa of the float-point value.
Take these math bases and see how many values it could generate, it will be some more than even an integer 32 bit signal would offer … ;)

For a regular video signal you would only use a small range of it: The range between 0 (which is black) and 1 (which is white) but any calculation which is done in the editing app can use and store values below 0 and above 1 which is very critical for many compositing processes and for exaggerated color correction.

If your project is based on 32 bit float-point processing your final output could be any bit depth available these days - 8 bit, 10 bit, 12 bit, 14 bit, 16 bit and more (each channel). It only depends on what your render format offers.
E. g. if you use a 32 float-point project to work with 16 bit CGI footage and finally you render to a TIFF image sequence, your output will be a true 16 bit (per channel) signal including an alpha channel.

Edit: I just realized Vegas Pro's DPX output is 10 bit (integer) only. So for a high-quality DCP I would not recommend to use DPX but EXR instead (which is 16 bit float-point then).
mraroid wrote on 3/17/2014, 6:10 PM
Thanks so much for your report. I would like to import my DCP files and burn a blu-ray. My film was scanned in at 2048 X 1536 (spirit scanner). Any tips before I dive into this? Do you think I can maintain this resolution and burn a 2K blu-ray?

The Sony manual has nothing on DCP. I do appreciate your post. All the best,

Entilza wrote on 6/12/2014, 7:52 AM
Sorry Jack - I didn't see your post, being so long after the original!

I have no idea how to bring a DCP into a timeline. Some DCPs are protected by a certificate - you might find it is a playback only format.

Blu-Ray is 16:9 native. Ignoring the minor res difference between 2K and Blu-Ray's 1080, if your film is 2.35:1, you'll lose almost 1/3 of your res if you go to Blu-Ray/1080, as the format will need a letter box to be applied and the image size reduced. Best to stick to full size for mastering.

If your project is 16:9, then there should be no harm.

Also - because I couldn't strike out my statements in error above (the bit about how an 8 bit project works) - I feel the need to reiterate to readers that Vegas has buggy sequenced image output, killing the gamma. You can only get proper output on an 8 bit project. PLUS, the DCP packager was rejecting the 32 bit images. It only wanted 8 bit. I do not know if this means DCP is an 8 bit format (surprising, but plausible - distribution has never been about quality), OR if the software is limited to 8 bit.
Rory Cooper wrote on 6/18/2018, 8:10 AM

good post thank you. i will give it a go, much appreciated.


wwjd wrote on 6/18/2018, 9:12 AM

is all this info still good in 2018 versions of DCP creation?