Are AVCHD camcorders the next HD lie?

Are AVCHD camcorders the next HD lie?

Summary: When I wrote "Don’t believe the low bit-rate ‘HD’ lie" with a comparison, I had no idea that these compromises in quality would apply to the latest consumer HD camcorder format called AVCHD as well.  When I first read about the AVCHD format with its use of MPEG4-AVC (H.

SHARE:
TOPICS: Hardware
53

When I wrote "Don’t believe the low bit-rate ‘HD’ lie" with a comparison, I had no idea that these compromises in quality would apply to the latest consumer HD camcorder format called AVCHD as well.  When I first read about the AVCHD format with its use of MPEG4-AVC (H.264) video compression at a maximum of 24 mbps versus HDV which uses the older MPEG-2 format at 25 mbps, I was very excited about the new format.  My enthusiasm dampened when I read the fine print that actual AVCHD implementation only uses 13 to 17 mbps MPEG4-AVC for compatibility with cheaper storage devices and it completely sunk when I read this excellent in-depth review from camcorderinfo.com.  Take a look at the screenshots below and it pains me to see how much detail is lost in the newer HD format.

Comparison of HDV versus AVCHD Image credit: camcorderinfo.com.  Lossless 1:1 crop.
Canon HG10 Bitrate: 13 to 15 Mbps MPEG4-AVC aka H.264 AVCHD camcorder format 1920x1080 60i Canon HV20 Bitrate: 24 Mbps MPEG-2 aka H.262 HDV camcorder format 1440x1080 60i (anamorphic)
The reason this comparison is so significant is that that both cameras have the same 2,960,000 pixel 1/2.7 inch CMOS sensor.  The only difference is the video compression, the bitrate, and the storage medium.  This flies in the face of conventional wisdom that says that MPEG4-AVC H.264 video compression can use 25% to 50% less bitrate to achieve the same quality level as MPEG-2 compression.  I've always suspected those numbers were inflated as an excuse to get people to believe that low-bitrate HD is the same as high-bitrate HD if better codecs where used, but even I didn't think it was this bad.

It's gotten so bad that I've seen whitepapers from vendors that try to push the idea that 4 to 6 mbps using MPEG4-AVC H.264 is equivalent to 18 mbps broadcast MPEG-2 HD video.  The problem is that these loose definitions of "equivalent" only compares perceived level of artifacts and not the loss of detail.  Now I'm perfectly willing to acknowledge that the above comparison may be partially attributed to a poor implementation of H.264 video compression, but the bitrate has to be a huge deciding factor.  The article from camcorderinfo.com also seems to suggest that they haven't found a single AVCHD camcorder that doesn't have low detail.

AVCHD is the second consumer HD camcorder format designed for DVD, Hard Drive, or SDHC flash memory recording designed as an alternative to the older HDV format which usually records to miniDV tape.  Because DVDs and flash drives are often capacity and write-throughput constrained, AVCHD camcorder makers have typically stuck with 13 to 15 mbps bitrates.  While Class 4 SDHC flash cards have write throughput of 32 mbps, cheaper Class 2 SDHC flash cards have a write throughput of 16 mbps and the camcorder makers don't want to exclude Class 2 SDHC flash cards.  Longer recording times on 8 or 16 GB flash cards obviously played in to the decision process.

It would seem logical that camcorder companies like should offer a max quality mode but perhaps they're gambling on the fact that most consumers will never know the difference.  This is really a shame since AVCHD random access storage is a much friendlier consumer format with drag-n-drop simplicity when transferring videos to the PC.  You simply put the SDHC in to your card reader on your computer or you hook up a USB 2.0 cable to the camcorder and you merely drag the files to your computer between 30 to 200 mbps.

With HDV, you have to deal with the hassles of something like Windows Vista Movie Maker which forces you to rewind the tape to the beginning and stream the video over in real-time at a fixed 30 mbps.  Buying an alternative video editing package will at least let you pick and choose what you want to record, but it doesn't solve the speed problem.  The other problem with miniDV is that it's very hard to preview what you've shot since you have to rewind the tape.  Sometimes if you don't put the tape in the right position, you can accidentally wipe valuable footage.  It's really a shame that AVCHD camcorder makers have botched the implementation because the format has real potential in the consumer market.

Note: JVC seems to at least be avoiding the low-bitrate pitfall with their hard drive based models like the GZ-HD7 which offers a high-bitrate MPEG-2 format.  This is essentially an HDV format for hard drive instead of a miniDV tape and the 60 GB hard drive provides approximately 4 hours of recording versus one hour on miniDV tapes.  This seems to be a good compromise since you get high-bitrate HDV format with the benefit of random access storage and fast drag-n-drop video transfers.  Camcorderinfo.com reviewed the GZ-HD7 here with quality lower than the best HDV camcorders but better quality than AVCHD camcorders.  While the recording format was better, the low light performance wasn't great.

One saving grace that's less applicable to the consumer space is the HDMI output on newer AVCHD camcorders.  For small independent filmmakers, they can hook up a computer with a $250 HDMI capture card with very fast RAID storage capable of 1.5 to 3 gigabits per second and capture raw uncompressed HD video.  In this application, the recording format in the camera is irrelevant since the PC records the video uncompressed.  But when you're shooting at this level of quality, you're probably going to use a much more expensive camcorder than a cheap $1000 consumer variety.  The kind of HD video quality you get these days with relatively cheap equipment is unparalleled in the history of the video industry.  High quality video equipment is no longer cost prohibitive to independent filmmakers and the only thing that limits quality is imagination and talent.

Topic: Hardware

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.

Talkback

53 comments
Log in or register to join the discussion
  • Very interesting - consumer be aware

    Thank you George - it`s a shame that the industry still pushes around these low quality performance items/standards to earn cents by product unit. For cell phones and very compact handheld devices those specs can be acceptable, but not with alleged performance oriented devices.
    green alien
  • The problem might not be the bandwidth

    Bandwidth is not a pure quality indicator, as you can use part or all of the bandwidth to transmit garbage. The quality really reside in the encoder. sorry to state the obvoius, but we must always keep that in mind when talking video quality.

    That said, the problem with H264 is the processing power required to come up with a decent result. Just look at the requirement for H264 playback, and imagine that encoding requires one order of magnitude more processing capability than playback, and you havbe your answer. Yes h264 has many features MPEG2 does not have to enhance quality; but your camcoder simply does not have the guts to implement them, and you end up with a poor encoder that simply cannot do the jobs. Where you to enhance the bandwidth, the problem wouldn't be dramatically changed.
    s_souche
    • You may be right, and the lack of interframe compression is a killer for

      You may be right, and the lack of interframe compression is a killer for AVCHD. Even at the same bitrate as HDV, the lack of interframe compression in AVCHD will mean it won't match up well. The H.264 implementation in AVCHD is simply the worst case implementation of H.264 there is.
      georgeou
      • It has interframe compression

        Since when didn't it use interframe compression?
        BobTurbo
  • We live in a world where 192k MP3 is replacing CD Quality Audio...

    This comes as no suprise.

    We've become so receptive of Mediocrity from companies that they seem willing to use our own stupidity against us.
    BitTwiddler
    • No, it's like saying you don't need 128 kbps MP3s and that 64kbps is enough

      An optimized 192kbps MP3 stream is inferior to uncompressed audio, but the difference is much smaller and it's doubtful you can pick up the difference without special equipment.

      Broadcast HDTV, HD DVD, and Blu-ray are already compressed between 75:1 and 100:1 and they're trying to further compress that to 300:1 or even 500:1. This is like trying to say that you don't need 128kbps MP3s and that 64kbps MP3s are perfectly fine. The problem is that 64kbps MP3s are so bad that you only need your ears to hear the difference.
      georgeou
  • thanks George

    for exposing the latest 'HD' gimmicks.
    The CE industry is very good at spinning and comming out with new acronims to part you from your money.
    Linux Geek
  • The three cheapest 4 GB SDHC cards on Newegg are Class 6

    [i]While Class 4 SDHC flash cards have write throughput of 32 mbps, cheaper Class 2 SDHC flash cards have a write throughput of 16 mbps and the camcorder makers don?t want to exclude Class 2 SDHC flash cards.[/i]

    I don't doubt that when they made this decision, class 2 cards were cheaper than class 4. However, at this point class 6 cards appear to be the most economical. Don't the manufacturers know this happens all the time? What industry are they in?

    I've been looking at camcorders ever since we found out my wife was pregnant (3.5 years ago). At the time I was very surprised to see all of the models based on tape or mini dvd. Having watched the drops in price and leaps in capacity in sd cards at the time I was convinced that flash based camcorders had to be around the corner. I bought a new digital camera instead and have been using it for both photos and video. It amazes me that the industry still doesn't get it. Tape is obsolete for the consumer space, but the manufacturers still don't get it. Oh well. I'm still fine using my digital camera for short videos, and will in fact be replacing it with a slightly better model some time this year.
    enduser_z
    • Yup, you're right. It did play a factor when they were building this gear

      Yup, you're right. It did play a factor when they were building this gear. The problem is that AVCHD doesn't use interframe compression which means it will still be inferior at the same bitrate.
      georgeou
  • Like comparing JPEG with RAW

    If you record digital photographs in JPEG format then there will be some loss of detail compared to recording the same photos in RAW format.

    Does that mean the JPEG photo is not a 12 Megapixel photo? Of course not.

    Similarly, an HD video is an HD video regardless of whether it is compressed using a high bitrate MPEG2 codec or a lower bitrate H.264 format. Quality may vary considerably.

    HD is HD. Quality / bitrate is an independant measure. There is no lie here.
    interlocutor
    • You mean "High Definition Fuzz"?

      Like I've stated before "Peak Power" in stereos was no lie, but
      still deceptive!
      kd5auq
      • HD doesn't mean High Quality

        HD doesn't mean High Quality. It simply indicates how many pixels there are in the picture. The quality of the image is determined by two things in addition to pixel count: (1) The codec used and (2) the bit-rate. Thus you can have low bit-rate HD and high bit-rate HD and they are both HD. The high bit-rate HD is not necessarily better quality than the low bit-rate one, because it also depends on the codoc / compression technology used.
        interlocutor
        • Oh here we go again

          "HD doesn't mean High Quality"

          Tell you what. You get the manufacturers of HD camcorders to put this slogan on a promenent place on their product packaging and I'll stop writing these blogs on the HD lie.

          "The high bit-rate HD is not necessarily better quality than the low bit-rate one, because it also depends on the codoc / compression technology used"

          Really? Did you look at the screencaptures in my blog? Didn't seem like H.264 helped the lower bitrates there.
          georgeou
          • point me to a high quality video

            and I promise I will generate two video, one high bitrate, and one moderate bitrate, with the last way better than the first
            s_souche
          • Define high versus moderate

            Define high versus moderate. Are you talking about 24 optimized MPEG-2 versus 12 mbps optimized H.264? The standard PSNR comparison tests I've seen call for at least 18 mbps H.264 to match 24 mbps MPEG-2. In the case of AVCHD, they didn't even bother to implement interframe interpolation which means it will pretty much stink at any bitrate less than 50 to 100 mbps.
            georgeou
          • Don't move the goal posts

            Previously I said high bit-rate HD is not necessarily better quality than low bit-rate. You
            indicated that you disagree categorically and now someone is calling you on it. But suddenly
            you want to insert additional parameters such as a large difference in bit-rate? Notice my
            comment that you disagreed with didn't specify a large difference in bit-rate.
            interlocutor
          • I didn't move the goal posts, and I'm not the one claiming more than a 2:1

            I didn't move the goal posts, and I'm not the one claiming more than a 2:1 advantage on compression ratio. It's the IPTV vendors that want us to believe that 8 mbps H.264 is just as good as 18 mbps MPEG-2 and I'm saying that's nonsense. I've also had people claiming that web content 1.4 mbps 720p using H.264 compression is just as good as broadcast ATSC HD because it's using H.264 instead of MPEG-2. I'm not saying you ever claimed this but without the proper disclaimers on the limitation of lower bitrate higher-end codecs, you risk people coming to the wrong conclusions.

            The white papers I've seen claim that you only need 2/3rds the bitrate when using optimized H.264 versus optimized MPEG-2 to achieve the same PSNR (Peak Signal to Noise Ratio). But that could easily mean you have more signal and more noise with MPEG-2 than H.264 but have the same ratio of signal to noise. However, I would prefer having more signal even if it means having more noise. This is why I actually prefer a slightly noisy analog TV signal over a low 2 mbps MPEG-2 stream even if the MPEG-2 digital video is perceptually "cleaner".
            georgeou
          • Manufacturer warnings

            <<< You get the manufacturers of HD camcorders to put this slogan on a promenent place on their product packaging and I'll stop writing these blogs on the HD lie. >>>

            We are gonna need a lot of warnings in that case. What about the megapixel counts of cameras? Digital camera makers are keeping on increasing megapixels without a corresponding increase in sensor size. For example the new Fuji F40fd has more megapixels than the older F31fd, but pictures are not as good as F31fd. We should have them display a warning.

            And how about ISO ratings. Canon and other digital camera makers advertise 1600 ISO on their Point & Shoot cameras when anything above 400 ISO is totally useless. We should have them display a warning about that too.
            interlocutor
          • You can't increase sensor size and keep a small form factor

            You can't increase sensor size and keep a small form factor. If you increase the size of the sensor, you need to increase the size of the lens and the camera body and consumers don't want that. On the other hand, increasing megapixel count does improve image quality when there is sufficient light. That means higher megapixels count helps quality when you're in daylight or when you have sufficient flash to light up the objects you're shooting. For lower light scenarios, you get more graininess per pixel but you can offset that by downsizing the resolution and you end up with the same noise count as the camera with the lower resolution but same sensor size.

            So bottom line is that higher megapixels DO help in situations with sufficient lighting but they don't help NOR hurt in darker lighting. I know David Pogue likes to claim it hurts you to have higher pixel density but he failed to factor in the fact that you can reduce the resolution and make it a wash.
            georgeou
          • And Hinders Quality

            "That means higher megapixels count helps quality when you're in daylight or when you have sufficient flash to light up the objects you're shooting."

            It also hinders quality because even though you increase resolution, you decrease dynamic range because the smaller pixels fill up too early. This means you will get blown out highlights or shadow clipping.

            "For lower light scenarios, you get more graininess per pixel but you can offset that by downsizing the resolution and you end up with the same noise count as the camera with the lower resolution but same sensor size."

            Only if you resize it to a postage stamp. The noise at ISO 400 on the best P&S cameras will not just go away with a little resizing.
            BobTurbo