Transcoding, down converting and compression.
Transcoding, down converting and compression.
Here are some of the questions, oh!, and answers.
“I absolutely agree on the painful lack of knowledge that exists about this stuff in the professional sector. Eight out of ten professional editors I know have little more than a thin veneer of knowledge. But they get through because in most broadcast, feature and TV production they are almost always dealing with known quantities, singular formats and consistent day in day out deliveries. There are very few variables, formats are never mixed and outputs are set in a post wide workflow template”.
These are the words of Mike Jones. Currently Mike is Lecturer in Screen Studies at the Australian Film TV and Radio School (AFTRS). He holds undergraduate and master’s degrees in creative arts and new media and is busy completing a PhD in cinema studies. Mike's knowledge and experience covers a breadth of forms and roles; camera, sound and direction through all areas of editing and post-production. Over 15 years his work has taken in film, video, motion graphics, 3D and animation, broadcast, live-events and online media along with photography, radio and music.
I’m John Westwood, journalist, filmmaker and photographer. Over the last thirty years I’ve made more than three thousand broadcast and corporate documentaries, pioneered satellite and laser technology and generally been involved with the creative industries both here and overseas. After all that time I still have burning questions in regard to transcoding, down converting and compression and to be honest I believe many of you have the same questions. So, let’s talk to Mike Jones and find his expert answer to these difficult topics and the easiest way to do that is simply ask.
Mike, the video enthusiast and professional alike at some time need to transcode, down convert or compress the footage from their camera to another medium, particularly a computer. Will you explain to me just what these terms mean and when they’re used?
Jones:
Transcoding and Down converting are entirely different. But sometimes they happen at the same time. Down converting is only about a change in resolution and doesn’t change the codec or the structure of the data compression. Transcoding is about changing the codec and actually re-writing the data stream of the video itself in an entirely new way using a different algorithm.
Okay Mike, I get it so far but how does the auto-convert on say a HDV camera work? You know, lots of people capture their HDV footage to a computer in standard definition digital video (DV).
Jones:
The auto down convert on a HDV camera does both at the same time, down convert and transcode, it changes the resolution on your tape from high definition HDV to standard definition DV. This is the down convert part but it also, at the same time, re-writes the video data from the Mpeg2 codec to the DV codec. In theory you could do a down convert without a transcode, such as an HDV Mpeg2 stream transcoded to a standard definition Mpeg2 stream, the codec stays the same but the resolution changes. There's no camera or system that will actually do this, nor would you want to, but you get the point.
Conversely, you can do a transcode without down converting the resolution. And indeed this is very common. So transcoding HDV files to Cineform, an intermediate codec, doesn’t change the resolution (unless you want it to) but does re-write the data stream out of Mpeg2 and into the Cineform codec which has a different colour space, bit depth and compression scheme.
Right Mike, I’ve got the picture but I’m sure some lesser mortals are starting to get confused. What I gather is the enthusiast simply needs to know that when the video camera is hooked up to the computer and they are down loading files, that’s exactly what they are doing, down loading. If at the same time they’re converting the file on the camera to a different resolution the process is called down converting. On the other hand, if they’re changing the codec in that process it is called transcoding. Got it! What exactly is the transcode process?
Jones:
In the transcode process the data of the video stream is re-written bit by bit into a new pattern dictated by the new codec, think like a translation from one language to another. In the case of a lossless codec like Cineform, the data is compressed in a way that looses no visual information - the data stream is packed up in a bigger box, more room to move, more flexibility when manipulated in post production. In the specific case of HDV to Cineform the key process that is taking place is a conversion from Long GOP to I-Frames.
Okay, now we’re talking about compression. Care to explain this process, Mike, and in easy to understand language, please? Pretty please!
Jones:
Long Group Of Pictures (GOP) is a compression based on discarding individual frames in favour of groups of frames that reference each other. This is done primarily to make your data smaller (compressed) so it will fit on the tape or disk. HDV and AVCHD are both Long GOP and are these days found in most consumer and a number of professional cameras. There is only one real whole frame (known as an I-Frame) every 12-15 frames. When such Long GOP formats are transcoded to a lossless format like Cineform, the Cineform codec separates and rebuilds the GOPs as single, self contained, individual frames and does so also in a 444 colour space with 10bits per pixel instead of
HDV's native 420 colour space with 8bits per pixel. When the process is finished the resultant file is placed in a wrapper, Cineform on Windows uses AVI as a default wrapper but the Mac uses MOV as a default. In either case they are just boxes in which to put the Cineform codec data stream.
It may sound a little confusing the first time you read this but it’s one of those topics you should print out and pin to the wall in your editing studio. Anything else you would like to add, Mike?
Jones:
Let me sum up. Aside from a change in resolution, if an HDV file is transcoded to DV the process is much the same as with Cineform in the sense that DV is not a Long GOP codec, it has individual frames. But, DV, unlike Cineform, is lossy (this means data is lost in the process of transcoding).
So, what this all means is that down convert means what it implies, a downward step from a higher resolution to a smaller resolution (HDV to DV, 4k to HD, etc). Just as up converting means to take a smaller resolution and artificially convert it to a higher resolution. A lot of docos are produced in high definition but use standard definition archive footage which is often up converted. Likewise, just about any broadcast production made these days in standard definition is up converted, also called up scaled, to high definition for broadcast.
Down convert is a change in resolution from higher to lower, where as transcode is a re-writing of data to a new compression scheme or codec independent of resolution. They are separate and totally different processes but sometimes done at the same time.
For more information on this process contact Redgum Television Productions, John Westwood www.redgumtv.com.au or redgumtv@tpg.com.au