What are the differences between HDV and AVCHD?
 

Can anyone articulate the primary differences between the two? Do both use long GOP? Is one easier to edit? Are there any inherent image quality differences?

You need to read this article:

http://governmentvideo.com/articles/...icle_982.shtml

Few clips:

AVCHD uses MPEG-4 AVC/H.264 compression and encoding for video
recording."

"HDV, on the other hand, uses MPEG-2 Main Profile @ High-1440."

"Anyone who has compressed an HD video feed to H.264 and to MPEG-2
High-1440 at the same bit rate will tell you that the H.264 version is
noticeably better."

"And this is the first 'rub' to HDV users -- AVCHD uses an almost
identical data rate (24 Mbps) -- in an MPEG-2 transport stream
wrapper."

"There are other video advantages in the AVCHD format."

"For example, it allows for 16:9 aspect ratio pictures in HD with a
raster size of 1920x1080 in addition to the HDV raster sizes of
1440x1080 and 1280x720."

"At the larger raster size, AVCHD has the potential to produce higher
horizontal resolution than HDV."

"AVCHD recording supports 1080i/24/50/60, as well as 720p/24/50/60."

"Plus, the new format supports 16:9 and 4:3 SD raster sizes of 720x480
at 60i (NTSC) and 720x576 at 50i (PAL)."

"Digging a little deeper into the two specs, AVCHD has an advantage
over HDV in luminance sampling of 1080 video."

"With the 1920x1080 raster, AVCHD uses a luminance sampling frequency
of 74.25 MHz, compared to HDV's 55.7 MHz."

HDV is still regarded as a much more pro format, primarily IMO because Sony and Pana havent produced a pro-level camera in that format yet....

As a more mature format, there is much wider editing support for HDV right now. In FCS at least, HDV is a breeze to edit, AVCHD is a PITA. That could change over time but is the situation currently.

MPEG-2 is a lot less taxing on a processor, to encode and decode, than MPEG-4 AVC. Image quality, at any particular bitrate, should be considerably better with MPEG-4 AVC though.

I'd take a lot of that article with caution as the writer clearly doesn't have a full grasp of why sampling frequencies are different when you have a different number of samples! Nor does he understand the differences between spacial and temporal resolution. The writer is also not aware that one of the default codecs for Blu-ray is almost exactly the same as HDV so encoding for Blu-ray from HDV is a breeze.

Mpeg4/AVCHD/H264, whatever you want to call it has some pros and cons. Subjective 1st generation picture quality can be better than HDV/Mpeg2 for the same bit rate, but in multi-generation tests AVCHD does not hold up as well as Mpeg2 using current codecs. Perhaps as the codec matures this situation will improve.

As has been said AVCHD is incredibly processor intensive to encode and decode, especially at high bit rates.

I've got to agree with Alister, that the author of that article does not seem to comprehend what he's writing about.

Both use long GOP encoding with AVCHD being more heavily compressed than HDV, and hence more difficult to edit. The workflow is also different, with AVCHD requiring more time to transcode to a useful editing format than it takes to capture HDV from tape. AVCHD offers long continuous recording times using large memory cards or hard drives, while most HDV cameras can only do this by purchasing a separate external recording device. AVCHD has slightly higher recorded resolution, but this is unlikely to be noticeable in most situations and is dependent on other camera features like sensor resolution and lens quality.

In practice the biggest difference is the shift from a tape-based to tapeless recording format and corresponding workflow changes, which will take some getting used to. If you shoot a lot of footage in a short period of time you'll want plenty of memory cards and a convenient way to offload those while out on a shoot.

Another consideration is that there's quite a gap in quality between the theoretical improved performance of AVCHD versus what can actually be achieved in real-time by the encoding chip in a consumer-level cam.