Working with AVCHD video files
For those accustomed to working with miniDV standard definition video tapes or the HDV 1440×1080 high definition format, the switch to AVCHD may prove challenging from an editing perspective.
“miniDV” and HDV formatted video can be edited directly on most computers with the caveat that editing HDV can take 4 to 5 times longer than mini-DV simply because there are a lot more bits to manipulate for the high definition video. For editing HDV and AVCHD you definitely want at least a dual processor (or dual core) and a minimum of 1 GB RAM (2 or more recommended).
mini-DV records video and audio at a rate of about 25 Mbps. Each video frame is compressed and then stored on the tape, individually using MPEG1 compression.
- miniDV image size is 720×480. Data rate is 25 mbps.
- HDV image size is either 1280×720 (few cameras use this) or 1440×1080. To display an HDV image on a 1920×1080 display, the image is stretched horizontally by 33%. Data rate is 19.7 Mbps (1280×720) or 25 Mbps.
- AVCHD image sizes are 1280×720, 1440×1080 or 1920×1080 – with a wide variety of data rates from as low as 5 Mbps up to 24 Mbps.
HDV records video and audio also at a rate of about 25 Mbps. To record a much more detailed image, HDV uses a compression scheme based on MPEG2. Unlike miniDV, which stores each individual compressed frame, HDV stores a compressed full frame, and then follows that with 14 frames recording only the changes from the first frame. Since the HDV codec only stores a full frame once every 15 frames, it can do a remarkable job of compressing a 5x larger image into the same space as standard definition recording. (There is also a 19.7 Mbps HDV version for the 1280×720 image size – however, there are few if any any consumer video cameras that record in 1280×720 mode on HDV so we will ignore that.)
The 15 frame sequence is known as a “group of pictures” or GOP. There is also a version with a 7 frame GOP. While HDV works very well, a major disadvantage is that a single tape drop out will typically ruin up to a full 15 frames – or 1/2 second of video. In the older standard definition encoding, a single dropout killed one frame or 1/30th of a second and no one noticed. With HDV, everyone notices a tape dropout!
AVCHD uses a version of MPEG4 encoding (H.264 or MPEG4 Level 10) that is twice as efficient as MPEG2. That means a 15 Mbps AVCHD encoding is roughly equivalent to a 30 MBps HDV recording (if such a thing existed). However, AVCHD can support multiple image sizes up to the full 1920×1080 HD resolution – plus 1440×1080 and 1280×720.. A 1920×1080 image requires 33% more bits than does a 1440×1080 image.
Most AVCHD based cameras record to Flash memory, and some to hard disk memory. Tape dropouts are then a “thing of the past”. This is a major plus for AVCHD. (Note – you can buy an expensive add on that will record HDV direct from the camera to hard disk to avoid tape drop outs – but these are expensive!) The major negative of AVCHD is that editing is very time consuming or requires huge disk space. And while tape dropouts are gone, you can still get corrupted file problems on the storage media (it happens to everyone eventually).
Editing standard definition video is straightforward since the original video contains each and every frame. Editing HDV and AVCHD is more complicated because the original video does not contain each frame. Further, AVCHD uses a complex encoding scheme that takes a lot more processor time to encode – and decode. Consequently, no matter how you slice it, AVCHD can take longer to edit.
At this time, there are very few editors available that can edit AVCHD files directly (Sony Vegas being a notable exception that can edit AVCHD). Thus, most solutions require that the AVCHD files be converted to an editable format.
On the Mac, the easiest solution is to connect your camera and let iMovie (or Final Cut Express 4 or Final Cut Pro 6) automatically import and convert the AVCHD format files to the Apple Intermediate Codec (AIC) or ProRes 422 codec format. However, converted files expand by 3 to 5 times larger than the original! Where as an hour of HDV or SD video might be less than 13 GB, an hour of AVCHD might be 49 GB! Fortunately, hard disks are now cheap.
So the disadvantage is disk space – but the advantage is very fast editing. If you have an older version of Final Cut Express (e.g. HD or 3.5) or Final Cut Pro (e.g. 5), you can use iMovie as your importer and then directly import the resulting converted files in to Final Cut (or just edit in iMovie if you want). If you set your time line sequence to AIC codec in Final Cut, you can import and edit all the files directly without a “rendering” step.
On Windows, the most straight forward solution is to use Sony Vegas – or the newest version of Adobe Premiere Pro CS4 or Premiere Elements 7 (update: or Magix Movie Edit Pro or Pinnacle Systems video editor). The advantage to direct AVCHD editing is very fast access to the files – the disadvantage is that each transition (dissolve for example) or title may require that the AVCHD be decoded into individual frames at edit or render time. So the time issue just gets pushed to a different part of the editing workflow.
Other wise, you’ll need to buy an application like VoltaicHD, which will convert the AVCHD files to AVI files that can then be imported to other editors (but which may require a rendering step in the editor). The files will also become a lot larger after conversion.
Some AVCHD capture programs enable you to view thumbnail clips of the original AVCHD that is still on the camera – and then select to import only the clips you want to work with (iMovie, Roxio Toast 10). Many also provide a way to quickly copy the compressed AVCHD files and archive those files – either by writing them to hard disk or, for some, burning to a DVD data disc for long term storage.
UPDATE TO NEXT PARAGRAPH: ROXIO TOAST 10 IS DEFECTIVE AND IS UNABLE TO CORRECTLY PROCESS AVCHD FILES. IT WILL PRODUCE FILES WITH UNSYNCHRONIZED AUDIO AND VIDEO.
Another method that works on the Mac is to use Roxio’s Toast 10 software. It can read AVCHD files from your camera and convert them to any format – including HDV or NTSC DV for subsequent editing. Obviously, if you decode a compressed AVCHD file and then re-encode into another compressed format like HDV, you will lose a tiny bit of image quality. Depending on the original source material, most people won’t notice.
Finally, there is the issue of dealing with 24p – or 24 frames per second mode. Depending on the camera and the editing software used, you may have to perform an “inverse telecine” operation on the original 24p video prior to editing. See footnote 2.
Update Fall 2012
Since this post was originally written, another popular solution has emerged called Neo Scene which converts videos to the Cineform codec (now owned by GoPro). Neo Scene converts your AVCHD into the Cineform codec – files do become much larger but editing becomes much quicker and smoother. The Cineform codec is better suited for editing than is AVCHD and many editing programs can play Cineform coded video in real time, smoothly.
I do not – yet – own this product but have used the trial version and have used the free GoPro Studio (which also uses Cineform but does not support so many cameras). Cineform really works.
Cineform can also handled 24p files embedded in a 60i video stream. Versions are available for both Windows and Mac, and based on my experiments on Windows – Cineform coded files work fine in Sony Vegas Movie Studio 11 and in Magix Movie Edit Pro latest versions.
Working with 1920×1080 AVCHD
On the Mac, if you are using iMovie as your editor, you need to be aware that the AIC codec only stores 1440×1080 images. So when you import your 1920×1080 images, they are re sized to 1440×1080 and then stored in the AIC format.
Does it matter?
Reading online forums, I see many people are focused on their 1920×1080 pixel resolution of their new Flash-based camera – but neglect that many consumer lenses cannot deliver the full resolution anyway. And after transcoding and editing, they then generally output their final production to 1280×720 video for online upload, or display via AppleTV, Sony PS3 or XBox 360. Or they burn a standard definnition DVD at 720×480.
What that means – 1920×1080 may not be as important as you think. Yes, 1920×1080 recorded on your camera, and then hooked up to an HDTV via an HDMI cable will look super! But only if you play back from the camera.
If you edit, transcode and ship as 1280×720 or standard DVD, no one will be able to tell the difference if you recorded your original in 1920×1080 or 1440×1080 or 1280×720.
What is more important to image quality is the person behind the camera making good decisions about what to photograph and how to photograph – and using a tripod of monopod wherever possible for steady images. Avoid excessive fast pans and fast zooms – which can make your highly compressed media turn to mush trying to deal with so much rapid motion.
 There are a bunch of Windows-based programs that claim to convert AVCHD to other formats. Oddly, the programs are nearly identical except for minor changes to their user interface. Most seemed to be based on the freely available ffmpeg program (but which does not support AVCHD) with some AVCHD decoder added on. The programs are so similar, that I suspect they are all distributed by the same organization under different names (but all priced at $35) as some sort of market strategy. Who knows? Anyway, VoltaicHD is its own program and not part of the mass distribution of similar programs. I’ve run a demo of VoltaicHD and it worked fine.
 Working with 24p. Some of the newer cameras offer 24 frame per second shooting mode. Some, like Canon’s HG10, HG20, HV20/HV30 stuff the 24 frames into a 60-interlaced fields sequence using a technique known as 2:3 pulldown (which I will not try to explain here). To edit this type of 24p, you must first pre-process the video (which otherwise resembles normal 60i video) to remove the pulldown frames prior to editing. The VoltaicHD program has an option to do this automatically during the transcoding from AVCHD step – which will save you time in the longer run. Otherwise, for example, I have imported using iMovie, but then run the imported videos through the JES Deinterlacer using it’s pulldown removal option (also known as “inverse telecine”), which adds another step and more disk space to the issue of shooting in 24p!