As CPUs get faster, machines are able to utilize more RAM and editing software gets more efficient, the need for intermediate codecs for editing HDV may fall away.
Not all of us are editing on the fastest computers, and a large percentage of us are shooting HDV. So what is an HDV shooter/editor to do? You can edit HDV natively or by using an intermediate codec. Editing natively means you ingest your MPEG-2 compressed HDV footage as the camera shot it and edit it in its native format. You can then output it without conversion as HDV or convert it to whatever format your video editing software will allow. Or, you could transcode your HDV footage before you cut your footage, using an intermediate format. In today’s personal computing editing world, there are advantages and disadvantages to both types of workflow. Before we compare and contrast the two workflows, we need to understand a little something about HDV.
When prosumers asked the HDV consortium – founded by JVC, Sony, Canon and Sharp – to make a more complicated encoding and editing process for them, the consortium invented HDV. Okay, maybe it didn’t quite happen like this, but compared to the compression found in the DV25 video format (typical Mini DV camcorders), HDV gets complicated. It needs to in order to go from the 480 lines of vertical resolution of DV25 to the 720 or 1080 lines of high-definition-quality video found in the HDV format.
The smart engineers who developed HDV figured out a way to compress all of those lines of high-definition video onto the same size tape we have been using in our Mini DV camcorders since 1996. By using MPEG-2 compression, a technology similar to that used on DVD discs, the engineers were able to squeeze 720 progressive vertical lines of resolution onto the Mini DV tapes at 19.7Mbits or 1080 interlaced vertical lines at 25Mbits. HDV is able to fit the same 60 minutes of 720p or 1080i HD video onto these small tapes by using what is known as a Long GOP MPEG-2 algorithm. To truly understand if you should be using an intermediate codec to edit HDV or not, you need to understand something about the Long GOP (hint: it has nothing to do with the Grand Old Party, a.k.a. the Republican Party).
Engineers developed algorithms to throw away information that is redundant between multiple frames, making video data streams smaller and thus easier to store on tape. This technique means that to decode an entire frame of video, the decoder must rely on information from other frames. The idea is that a group of pictures (GOP) can be used to make one frame. When we cut between two frames or add a transition between frames on our timeline, our editing systems need to access data from the frames on either side of it. Another way to look at it is to picture video as an animated flipbook. The Group of Pictures is similar, but a wee bit more confusing, because the individual “stills” are not independent of each other. Whereas each of the single frames of a DV25 data stream is a standalone, independent frame, a GOP has six interdependent frames for 720-line HDV and 15 frames for 1080-line HDV. GOPs can be double this amount for progressive footage.
The structure of the GOP and how the three different types of frames – the I-frame, the P-frame and the B-frame – work together is the subject for another article, as it gets a bit complicated. What we need to know is that this group of pictures is a combination of these three types of frames working together, enabling them to discard even more information than the standalone intraframe compressed video used in DV cameras. These six- or fifteen-frame GOPs use interframe or temporal compression, as well as intraframe or spatial compression, to evaluate the whole group, discarding more redundant information than intraframe accomplishes on its own.
Lossy vs. Lossless
By using both intra- and interframe compression, HDV is able to fit 4.5 times the resolution of standard-definition NTSC DV (3.75 times that of PAL DV) on both tape and the hard drive. The MPEG-2 compression is lossy, meaning it loses information, but the algorithms work very well to build a sharp picture. This discarding of data results in smaller file sizes with very little noticeable loss in quality.
So, if we are editing our MPEG-2 HDV natively, chances are we are slicing into one of these GOPs every time we make a cut. Thus, the computer needs to eventually rebuild this I-, P-, B-frame structure that we have just sliced up. In addition, just playing one of these clips could be difficult for a computer, as it must reference and decode many interdependent frames at once, just to show a single frame. Plus, your computer will need more RAM and processing power to run real-time effects.
That said, if you are not using many effects or transitions – in other words, if you are just doing cuts-only editing and need to output – editing natively is a good choice, if your computer is strong enough, with a good amount of RAM. It will need only to rebuild or conform those cut-up GOPs when it exports.
Transcoding to an intermediate format will let you use more real-time effects, and it can even preserve your footage, as cutting into the GOP in native editing can cause gradual degradation of the image. One of the main drawbacks with transcoding is space. Once you “unwrap” these files, they become much bigger, requiring much more hard drive space. Plus, you most likely will need to buy software solutions that range from $150 to $2000.
There are many companies making intermediate codecs with a variety of features. Some of these transcoding solutions include CineForm products for Sony Vegas, Premiere Pro, Ulead MediaStudio Pro 8 and Ulead Videostudio 11 Plus editing systems. Apple’s Final Cut Pro can use BitJazz’s SheerVideo. Check the specifications with both the transcoding codec and your editing software to make sure it is the best solution, and remember that some of these companies offer free trial versions on their websites.
HDV is an excellent solution to get consumer and lower-end prosumers into the HD market. It is only a temporary solution, as someday all cameras will most likely be HD. Until then, we need to figure out ways to make our present hardware and software work with us, as we bridge into the high-definition world. Transcoding HDV is one of these solutions.
Contributing editor Morgan Paar is a nomadic producer, shooter and editor, who is currently teaching high school video production.