If you’re reading this, you’ve probably read my first writeup on cameras for in-car video. If not, you might want to, as it presents my arguments for why I’m not going to deal with a tape-based camera, which basically comes down to the fact that convenience is my biggest criteria, more than absolute video quality. A lot has changed since that first writeup eight months ago, and two very appealing new contenders have come from an unlikely source: Sanyo.

Sanyo’s new family of HD, solid-state camcorders (Sanyo HD700 on left, HD1000 on right)

In my previous writeup, I mentioned that the quality of mpeg4 tends to suck. Well, the times have changed. H.264 is a newer encoding method for MP4 used by the Sanyos, and it rocks. Both these cameras use the same chipset to encode video, and both encode at 12Mbs max. Now that’s half the theoretical rate of some of the tape-based AVCHD or HDV camcorders out there, but it’s PLENTY for most users. I would also point out that at the time of this writing, 17Mbs is about the max consumer HD camcorders are recording. At 12Mbs, which is the data rate whether shooting 1920x1080i or 1280x720p, you get about 10-12 minutes of video per 1GB of storage, or about 95-95 minutes on an 8GB SDHC. There is no maximum file size, apparently, though I haven’t tested this. I would think there would be a 4GB limitation due to the fat32 formatting of the SDHC, but it’s not mentioned in the user manual. Still, that’s about 45-50 minutes worth of video. It should be also be mentioned that the Sanyo HD700 doesn’t do 1080i video, it maxes out at 720p. Ok, enough about encoding.

Sanyo HD700 Video still, 720p. Note the grainy quality of the video. Click for actual resolution (1280×720)

Sanyo HD700 ($300 – $450)

The first camera I tried was the volatilely-priced Sanyo HD700. At the time it was $300 from Amazon (at the time of this writing it’s $400 at least), and I jumped on it despite not really needed a new camera. I played with it for a couple of weeks, but sadly I didn’t have a track day during that period to give it a proper test. I only had 30 days to send it back, and I wasn’t going to miss that opportunity. The reasons I sent it back were basically it has a very small aperture, slow lens. This means it performs poorly in low light. This isn’t such a big deal for a track day cam, but I was looking for a multi-purpose camcorder, one I could use recording kid’s birthdays, etc, and the HD700 is not that cam. Now don’t get me wrong: for $300 it’s pretty good…if you can find it for $300…but only for an outdoor camera. Once you get to sunset, you’re not going to have enough light for this lens, and that pretty much kills this camera for general use. If you’d like to see a video sample from it, here is a 19MB video shot at dusk driving home. This shows pretty effectively the low light problems with this cam. It would, however, provide a very cost-effective way to get HD on-track video.

Sanyo HD1000 video still at 720p of T6 at WSIR (using a wide-angle lens). Click for actual size.

Sanyo HD1000 ($650 – 750)

On A much happier note, Sanyo also offers us the Sanyo HD1000. While this camera shares the encoding chipset, general layout, and menuing system with the HD700, the similarities end there. The lens is a very fast f1.8, with a 10x optical zoom (not that we care for in-car use), and, thank god, an actual threaded lens for add-ons. This is fortunate, as both the HD1000 and the HD700 have an odd problem; they’re both very zoomed-in at their widest setting. In the car, this means you really need a macro lens to back the camera up, which is no problem with the HD1000.

Sanyo, of course, offers a wide-angle adapter for the HD1000, but at about a $119 street price, it’s pretty damn expensive in my opinion. Instead, I turned to Ebay, where cheap knockoff lenses of dubious sourcing abound. I picked up a .45 wide-angle adapter for $30 with shipping from Hong Kong, and as you can see below, it does the job. A note on wide-angle adapters: the 40.5mm threads of the HD1000 are somewhat unusual, so you’re likely to find 37mm lenses with a 40.5mm conversion ring rather than a true 40.5mm lens. That’s what I bought. The downside to this is that when you’re shooting stills with the camera, you have to zoom in about 1.3x to get a square picture, otherwise you get rounded edges from the smaller wide-angle lens. It doesn’t affect video mode, but the camera starts at a wider setting when shooting stills, so the problem shows up there.

On the left, the HD1000 at its widest setting (Look, Ma; no hands!). On the right, with a .45 wide-angle adapter. Note: these have been compressed and recompressed. The quality sucks. It’s just to show the difference in fields of view.

Other useful info on the HD1000 are that battery life is very good, and despite the fact the specs say it’s rated for 85 minutes of 720i video per charge, I’ve taken 2.5 hours of 720p video in one track day, along with many non-flash stills, and never run out of battery on the camera. The optical image stabilization is digital, meaning it’s not good. Testing with stabilization on and off, I’d have to say just leave it turned off. If you have a decently isolated mount, or a relatively soft car, your video will be pretty smooth. The Willow Springs 1-2008 HD videos on this site were taken with stabilization turned on, while the Laguna Seca 2-2008 videos were taken with stabilization turned off. For this reason, the Laguna videos are much less jerky. For both events, the HD1000 was mounted using the little Chasecam clamp mount I discussed here. My videos would probably be even smoother with the i/o port mount, but I don’t really have room for it in the Exige.

Optimizing settings for the track – trial and error experiences

You may be wondering why the video of the fast lap at Laguna looks so bad, and all I can say is I screwed up. I had used the cam the night before to shoot some stuff, and left the camera on “high sensitivity” mode (aka night mode) in the early sessions of the day. That meant the white balance was way too bright, and I had to try to salvage the video by screwing with the settings in the software when I edited it. It should not be taken as an example of what the camera can do. For that, look at the other 2-2008 videos (but not the splitter cam, which was taken with my Canon SD700).

Half the battle of getting good video is using the correct settings for what you’re doing, so I’ll mention here what I’ve found using the HD1000. For one thing, you must use manual focus mode. If you leave the cam on autofocus, it’ll constantly focus-hunt as you go around the track, come up on cars to pass, etc. It’ll drive you nuts. I finally settled on setting manual focus at 20m (meters). I shot the Willow Springs videos at 8m, and that’s why the distant objects are so out of focus. I used 20m at Laguna, and I think it’s a better compromise. You should also put the camera into “sports” mode, which tells the cam to optimize its compression for fast-moving objects. If you don’t, objects in transition, like passed cars or changing landscape in turns, will be less smooth. Turn off video stabilization, as I mentioned earlier. And perhaps the suggestion you may find most surprising, shoot at 720p.

1020i v 720p

1020i is interlaced video shot at 30 FIELDS per second. 720p is progressive, at 60 FIELDS per second. That is not the same as FRAMES per second, and if you’re curious about the technical differences, you can look here. Basically, shooting in a progressive, or non-interlaced, mode gives you more data for a given resolution, and thus better quality. It also gives your video more data to work with when you edit it down and convert it to other formats, use effects on it, whatever. This is illustrated by the fact the HD1000 uses 12mbs of bandwidth to shoot 720p or 1020i; there’s a greater data density at 720p in a smaller area. Anyway, there are two reasons I choose to shoot in 720p over 1020i. First of all, I’ve got fast-moving objects, and having the greater data density allows the codec to make better selections during compression to make the video smoother. If I was shooting still or slow-moving objects it would be less important, but for our use, this is a major consideration. The second reason is 1020 video is hugely intensive to play. My core2 duo 2.4GHz machine can barely play the stuff, and you really need a .264 hardware decoder to deal with it. What it all comes down to is 1020 is so intensive to decode it’s ahead of its time for most of today’s computers, as they’re using software and non-optimized chips to attempt to decode it on the fly, and they’re overwhelmed. Some of you with slower computers may find you can’t even play the 720p videos I’ve put online smoothly. Now consider that what I’ve put online is compressed down to 3mbs, and imagine it at 12mbs with over twice as many pixels to deal with. It’ll bring your computer to its knees.

Memory and the HD1000

A quick word about memory for HD recording use. You are going to need fast SDHC cards to keep up with the write speeds of these cameras, as 12mbs is well beyond what the bargain-basement cards will manage. While I have, admittedly, had good luck in the past with Corsair flash cards in my TZ1, I’ve decided to quit risking my data and just buy the best. The way I look at it, is it worth it to save $20 on a card, and take a 20x greater likelihood that you’re going to lose an entire day’s video? It’s not to me. That’s why, for this cam, I bought two of the
Sandisk Ultra II 8GB SDHC cards. They’re only $62 as of this writing, and they’ll give you piece of mind. Note that SDHC cards are not readable by the older-standard SD card readers, so if you don’t already have an SDHC reader, you’re going to need something like this SDHC Reader, which is only $5. Whatever you do, and I’m completely serious, don’t buy Transcend flash cards. I have owned 4 or 5 of them, and I’ve had to send 3 of them back for replacement. I’m convinced at this point they don’t actually do quality control at all; they just figure a lot of people won’t bother to send the cards back to them for replacement (they do have a free lifetime replacement warranty). I wouldn’t be surprised if they actually bought chips that fell out of other vendors QC, actually. Their record is that bad, at least in my small sample.

Notice: This work is licensed under a BY-NC-SA. Permalink:

Leave a Reply

Your email address will not be published. Required fields are marked *