That's interesting about the WORM cards, but unless the cost is reasonable few people would use them for video recording. We'll see how that progresses over time.

After thinking about recording and archiving costs more I can see how solid state is becoming competitive with miniDV tape prices, depending on how you run the numbers. But we still need something cheaper than $800 memory cards to make solid state fully practical for a wide range of users. For the fellow who wants to record for five hours out in the field without buying five hours' worth of SxS or P2 cards, the Sony Z7U offers another option using CF cards and the HMC150 will use SD. Figure five hours worth of those cards costs ~$750, and your payback compared to using DV tape is a modest 150-250 hours of shooting. Even for those of us used to DV tape costs this makes solid state start to look pretty good, so I guess we don't need another 3-4 years of price cuts - we just need more cameras which run on standard flash memory. (And for those who want higher performance, P2 and SxS are both functional solutions.)

That's exactly the thing Kevin. And why the HMC150 and similar cameras are so compelling right now. With Sony's XDCAM Optical and SxS and Panny's P2, it's going to be a slow adoption of CF cards I think. On the consumer end, I think the uptake will be much faster. The manufacturers can get the products in the hands of shooters and not worry about media. On the pro end, there is legacy inertia built into the older standards. It's those of us in the middle who are in limbo. Does Sony and Panny push us into the legacy world, or push us down toward consumers. With The EX1 coming from CineAlta and not Sony Electronics, I think that choice has been made clear. Not so much with the Panasonic HPX170 and other cameras.

P2 vs. SxS editing differences
 

I've believe the Sony SxS XDCAM format compresses a group of frames together in order to get the lower bitrate and increase the amount of footage you can get on a card. I've heard this can slow things down significantly when you do a lot of editing with this format. When you scrub the timeline to get the right edit point, the computer has to decode the entire group of frames in order to display a single frame on the time line.

By contrast, P2 DVCPRO-HD compresses each frame individually. That's why you only get 1/4 as much footage on a single card. For example, a 16GB P2 card only stores about 17 minutes of 1080p video, where the 16GB SxS card will store around an hour. If you're shooting a feature, news segment, or something like that, most clips only last around 30 seconds or so, and you need a lot of time to set up the next shot, so 17 minutes of actual footage might be a couple of hours on location. In this type of environment, swapping and downloading P2 cards isn't that hard to do.

So for features, short films, news, etc., the P2 format is probably better because it's faster to do lot's of editing. Keep in mind that with these types of projects, edit points usually occur every few seconds.

But for videography and other types of projects that have less edit points, having 4 times as much footage with the SxS format has obvious advantages.

In a perfect world, both Sony and Panasonic would have a menu selection to compress frames either individually (for faster editing), or as a group (for more footage), depending on the type of project you're working on. But as it is, you have to choose the format that's most appropriate up front.

As a point of reference, there's a table on the Wikipedia site that shows all the various digital video formats and their attributes. The table is about 1/3 of the way down here:
Digital cinematography - Wikipedia, the free encyclopedia

Any video can be converted to an I-frame intermediary for editing purposes, so that's a negligible issue other than the time required to do the conversion. But if you need to shoot several hours of footage out in the field without pausing to clear off your memory cards, SxS is more viable than P2 and neither is very practical compared to CF or SD.

In addition to conversion time, I believe there is also an issue with image quality. If you de-compress a group of pictures (GOP) and then re-compress to new format that compresses each frame individually, then you've gone through another compression cycle. This will add to mosquito noise and blocking artifacts. Two compression cycles (one for capture, and another for authoring edited output) is already too much, and it usually shows. The only case where it doesn't show is if the capture format is wavelet (i.e. Red 1, SI-2K), or the DCT capture compression is really light (i.e. higher bit rate formats).

You could de-compress a GOP based format like SxS and then create uncompressed files for editing, but the data size on the hard drive would be huge and this would slow down the editing as well.

From what I understand, it's generally best to edit in the format you use to capture, otherwise you will lose quality or speed.

Cineform works well.

Yes. Cineform uses wavelet compression, so it can be used as an intermediate without any noticeable loss in quality.

But from what I've heard, Cineform software costs more than most editors. Around $2000 if I remember correctly.

Typical editing intermediates use about the same bandwidth and storage as DVCProHD in some cases (e.g. Cineform, Canopus HQ) or somewhat more in others (Avid DNxHD, Apple ProRes), but not nearly as much as uncompressed footage. With hard drives so cheap these days decompressing for editing isn't a problem, and performance is improved significantly (not decreased) in the process.

In theory an extra transcoding step will reduce image quality slightly, but it's doubtful the results would be noticeable to most viewers. If you're really concerned about image purity you're arguably better off with a GOP-based recording format, because that can be delivered directly to viewers via Blu-ray or the internet where DVCProHD has to be transcoded for viewing purposes. In practice all the popular HD recording formats are good enough for most purposes, even DVCProHD. ;-)

This is true if you are not doing any editing. When you edit, your edit points will generally be within the GOP, so the video will have to be decompressed and then recompressed. As far as I know, all editors do this internally when authoring output.

Uncompressed 1080p video is huge. Moving that much data through the editor will slow things down. With uncompressed footage, it's not the processing intensity, but rather the I/O intensity that slows things down.

To add some dimesions to this, let's say you're doing a 2-hour feature and your shooting ratio is about 10:1. That's 20 hours of raw video. Uncompressed 1080p video is just over 300GB per hour, so that would be over 6 Terabyes for just the raw video.

That's what RAID arrays are for.

At today's prices, that's well under $1k in hard drives (but using something like Lagarith can make a lot of sense - if you want a completely lossless intermediate).

That's one reason it pays to convert to an I-frame intermediary. However, you can take raw GOP-based video and deliver it directly to end viewers, something you can't realistically do with DVCProHD. With any video format you're going to end up converting somewhere in the process unless you deliver the raw footage.

Please read my preceding comments again: I'm not talking about uncompressed video. Digital intermediates use ~80-160 Mbps and work very well for editing GOP-based source video on today's computers: I can handle 3 layers of HD 1080i on my laptop from an external bus-powered hard drive - and the codec for this comes free with Edius. It doesn't sound like you're familiar with how all this works, so I'd suggest you study up if you want to comment further.

Actually, the standard version of cineform which can handle anything up to 1080p is only $699.

Higher versions which can handle up to 4k at 10-12bit are more expensive, but I don't think any of them run as high as $2k.

The big advantage to wavelet compression is speed. Similar quality (at a given bitrate) can be achieved with DCT like compression (like AVC Intra), but it is slower to encode and decode.

Cineform NEO HDV, which is quite adequate for many purposes, is only $250. The only thing I really dislike about Cineform is how over the top they are about limiting you to using their products on only one computer per licensed copy.

I think you will find Cineform will readily agree to having one copy simultaneously installed on 2 machines. At least that has been my experience with AspectHD.

My experience has also been that the folks at Cineform are completely reasonable about having one copy on their notebook and one on their main machine. Or whatever else you need, within reason.

Just ask them for what you need and I think you'll be pleasantly surprised at how helpful they are.