This article is a quick primer on how to work with Red’s proprietary file format, called Redcode RAW. We’ll look at:
- What it is,
- How it is structured, and
- How it can be used by your favorite editing applications.
What is Redcode RAW?
Redcode RAW is the proprietary format of the Red Digital Cinema Camera Company. Everything you record using the Red Scarlet, Red Epic and Red Epic Dragon is saved as Redcode RAW, with the file extension *.R3D. These cameras don’t save files in any other format.
Like any other file format, Redcode RAW has three main parts:
- The wrapper format, which contains RAW data from the camera
- The compression format, which compresses the RAW data
- The metadata, which carries forward camera, file and other information
If you are new to the terms mentioned above, please read through the links. The takeaways are as follows:
- Redcode is always compressed – there is no uncomopressed version.
- Redcode is always lossy, even at its best compression.
- Redcode is always RAW, so there is no chroma subsampling or color space attached to the data.
- Redcode will always carry relevant metadata, especially the settings used in camera while recording the files.
Red also states that the Redcode format is totally royalty-free, unlike H.264.
Understanding the properties of Redcode RAW
We’ll look at the properties in this fashion:
- File structure
Redcode has one limitation – it strictly follows the FAT32 file system. The reason they went for this is to create files that are equally supported on Macs and PCs (Windows) without any modifications required. The disadvantage of this format is that you can’t go over 4 GB (1024 MB) in size. Let’s say the Red Epic records in 150 MB/s. One minute will require about 9 GB of footage. If you keep the camera running for one minute, the camera will write two files, both with the extension *.R3D.
As you can guess, longer record times will break up the files and soon you’ll have a whole bunch of files that are hard to manage. For this reason, every time you start and stop the camera, it creates a folder which ends in *.RDC (stands for Red Digital Clip). The RDC folder contains all the broken up R3D files for each take. If you start and stop your camera for another take, it will create a separate RDC folder, which will contain the R3D clips.
If for some reason the shot doesn’t require more than 4 GB, you will only find one R3D file.
In addition to the R3D file, Redcode also has a sidecar file with the extension RMD (Red Metadata). This is the file that contains the metadata information (either created in camera or in Redcine-X Pro) and will always remain attached to its corresponding R3D file.
The names of the R3D files and RDC folders follow a pattern that looks something like this:
C RRR TTT dd mm HH NNN, where
- C – Camera letter, from A to Z
- RRR – three digit reel number
- TTT – three digit take number
- dd mm – date and month
- HH – unique hex code that ensures another file won’t have the exact same name
- NNN – spanned clip number (when R3D files are broken up)
RMD files will have the exact same names as their corresponding R3D files.
The compression used in Redcode can be broken down in this fashion:
A unique compression scheme that also stores full resolution, half resolution, quarter resolution and a thumbnail image information in the same file, without needing to duplicate the data (But still increases the data size by a negligible amount). Redcode is based on JPEG2000, though they have made their own modifications to ensure the best quality compression with their cameras.
The bit rate is variable (VBR), though Red gives approximate data rate guidelines for each compression level (next).
Redcode is compressed, and the lowest compression setting is 3:1, which effectively means the data is compressed to one-third its original size. E.g., a full raster 6K 6144 x 3160 24p file will have a data rate of 2 GB/s (!!). The same in RAW will have a data rate of about 670 MB/s (!), while in 3:1 the video might have a data rate of about 220 MB/s.
The highest compression level at the time of this writing is 18:1. The compression level is directly related with the write capability of the media used in the camera. At greater frame rates, there is more data being written per second, and Red cameras don’t allow you to go all the way to 3:1. Tables for compression-frame rate-resolution are available in the camera manuals.
The lowest compression setting will always yield the best quality, but each production will test various compression levels to find a suitable balance between image quality and data rate. E.g., even though major movies like The Hobbit and The Amazing Spiderman could afford 3:1 if they wanted to, they chose to shoot at a lower setting of 5:1, hence saving about 40% in storage needs.
Is it lossy or lossless?
It is a lossy codec, though Red claims it is visually lossless. At the lower compression levels, Redcode is good enough for the most demanding productions. The advantage is reduced file sizes.
Is it interframe or intraframe?
It is an intraframe codec. Each frame is compressed separately.
The color information can be broken down in this fashion:
There are four channels in any R3D file – one for red, one for blue and two for green. The data is RAW, and needs to be debayered in a RAW converter before it begins to look like an image.
Color space and gamut
Natively, Redcode is not assigned a color space, and you have the ability to map it to the color space of your choice in Redcine-X Pro or other software. Naturally, like every other camera, the data will be affected by the sensor’s gamut and is what gives its files the ‘Red look’.
Color bit depth
Red claims the files are 12-bit or 16-bit, but as far as I know the files are 12-bit. There is no documentation or explicit declaration of how 16-bit is obtained.
RAW files don’t have chroma subsampling.
There are two kinds of metadata:
In-camera refers to the camera settings and file information stored within the R3D file itself as part of the file format.
While you shoot, you can create a ‘recipe’ for how the final image will look. You can select and manipulate:
- Gamma – e.g. REDgamma versions
- Color space – e.g., REDcolor versions
- White balance
- Other color correction tools and features
- HDRx – which allows you to extend the dynamic range of the camera by shooting two frames with different exposures, which consequently doubles the data requirements
All this is stored as RMD files. The original R3D files are untouched, and this allows you to create new and separate looks later in post production, without resorting to destructive editing and file manipulation.
It also means, for a true Redcode workflow, the post production software of your choice must also know how to work with RMD files.
How to work with Redcode RAW footage in your favorite NLE
I have already covered Red workflows in detail elsewhere, but here’s a quick rundown (not up to date!!):
|Support for Redcode||Max Res.||All frame rates?||Color bit depth||RMD changes?|
|Adobe Premiere Pro||Native||6K||Yes||10-bit||Yes|
|Apple Final Cut Pro X||Native||5K||Yes||16-bit||Yes|
|Avid Media Composer 7||Native||HD||Yes||10-bit||Yes|
|Grass Valley Edius Pro||Native||4K||Yes||10-bit||No|
|Sony Vegas Pro||Native||4K||60p limit||32-bit||No|
|Editshare Lightworks||Native||HD||30p limit||10-bit||No|
*I’m not sure about this, though Smoke can natively take even higher resolutions.
The first four NLEs in the list all support RMD manipulation, which makes it way more easier to work with Redcode metadata and color. My favorite NLE for Redcode workflows is Adobe Premiere Pro without a doubt. It also has native support for Red Rocket (as do the other top NLEs) and the full 6K resolution.
Beware that some NLE’s that allow RMD manipulation will overwrite the existing RMD files. The best software to create your own recipe is Redcine-X Pro.
Many workflows recommend a Red Rocket card to work with Redcode RAW. The reason this card is required is not because Redcode needs to be debayered, but because it needs to be decompressed. Wavelet compression and decompression demands a lot of processing power, and this task cannot be relegated to the CPU which must also tend to other duties. This is why adding a custom Red Rocket card speeds up working with (includes playback and rendering) Redcode RAW.
The typical Redcode workflow starts with the camera, its settings and Redcine-X Pro. Once a ‘look’ has been achieved, all that remains to be done is record the files, just like you would in any other camera. You can always create new RMD files later in post production, or work with the same you have created. The exact same methodology applies in color grading as well. To understand how to grade in Redcine-X, check out this primer from Red.
I hope this simple primer has helped you understand how Redcode RAW works, and how to use it in your workflow.