Sony HVR-Z1U vs. Panasonic AG-HVX200
 

I'm getting ready to invest in an HD Camcorder. I'm very confused between the Panasonic and the Sony. This camcorder will be used to shoot a documentary of beautiful scenery. Any input is welcome. Thanks.

The Panasonic is not on the market yet. Neither is the JVC. Wait till the fall and you should be able to compare all three.

From the footage I've seen, from two of the three cameras, I'd say that for what you want it for, you'd be best fit with the FX1/Z1U for most things. If you need footage of something far away, go with the JVC HD-100U. The HVX200 is more for indie filmmakers at the moment, give it a few years.

The interesting thing about the Panasonic HVX-200 is that even though there is much less compresion and much better chroma detail the resolution is actually much lower.

For HDV 720p you get the full 1280x720
DVCpro-HD 720p only gives you 960x720 stretched.

HDV 1080i gives you 1440x1080 (Although this comes from a pixel shifted 960 pixel wide CCD on the Z1/FX1)
DVCpro-HD only uses 1280x1080 (who knows what kind of chips are in there right now)

For slow moving nature video I would think HDV would be better because you get more resolution detail. High 4:2:2 chroma detail wouldn't be as important and many would actually have a hard time to see the difference assuming no effects or color work will be done to the footage.

the biggest thing that sways me is the panasonic will only recrod HD to the P2 card, it will not record HD to tape. With P2 cards be so expensive and only being to hold four minutes, the Z1 seems to be the right choice.

Everybody is proclaiming the wonders of DVCPRO HD...but what people have seen is mostly from the Varicam (at $60,000 +). The HVX200 may be a very nice little camera, but it is still a 1/3" chip camera. Nobody really knows whether the recording format, ie., HDV or DVCPRO HD, will make all that much difference. Would a PD170 recording to Dibigeta look better than the same PD170 recording to DVCAM? Probably a little bit, but not day and night--it's still a 1/3" chip camera. I'm expecting the Z1, the Panasonic and the JVC to all be very similar in the end result. Whether the DVCPRO HD will be worth the difference in cost is the real question, in my mind. Until the price drops and capacity increases big time, it's not for me for the kind of stuff I do. For others it may be perfect, but remember--early adopters of technology almost always pay more, although that doesn't seem to be the case with HDV.

While we cannot yet compare the cameras, we can certainly compare the recording formats. And in that comparison, DVCPRO-HD proves to be far superior -- under all circumstances you get twice the color resolution and consistent performance. HDV always has half the color, and then you have to factor in best-case and worst-case pictures. In the very, very, very best case, shooting a completely still shot, HDV in 720p has slightly better mosquito noise performance than DVCPRO-HD. But in the worst case, lots of change and lots of motion, DVCPRO-HD is embarrasingly superior to HDV.

In 1080i, DVCPRO-HD is again consistent -- but HDV has even fewer bits per pixel of compression, so its mosquito noise performance is worse in 1080. And its 4:2:0 color sampling, combined with an interlaced scan system, makes for odd low-resolution line rendition.

Here are two pictures where I compared HDV vs. DVCPRO-HD, regardless of cameras, just comparing the compression system independently of any other consideration. You *cannot* extrapolate from this that one camera will be better than another, but you can certainly extrapolate that one compression system is clearly superior to the other.

Here was the test of 720p HDV vs. DVCPRO-HD. As outlined in the test methodology, HDV performs at variable levels depending on the type of data that it's asked to compress -- a still shot will compress very well, a moving/changing shot will compress poorly. Either case is likely to prove rare; you'll rarely ever have the luxury of shooting a completely static shot, nor will you likely be stressing the codec to its very limits, so both pictures represent the extreme, leaving it to the viewer to extrapolate a mid-case.
http://www.icexpo.com/dvx100/720vs720.png

For 1080i:
http://www.icexpo.com/dvx100/1080vs1080.PNG

I then later devised a test for HDV to try to get a mid-case level of performance. Using the same picture (the codec test picture from One River Media) I put it on the timeline with a mild 5% pan & 5% push. The results of that test were used for the "mid" case results. The picture is a codec stress test, so it really pushes the codec hard.

http://www.icexpo.com/dvx100/OneRive...traction-1.PNG
http://www.icexpo.com/dvx100/OneRive...traction-2.PNG
http://www.icexpo.com/dvx100/OneRive...traction-3.PNG

While the best-case for HDV looks good, and the worst-case for HDV looks atrocious, it's important to keep in mind that you're likely never going to encounter a situation where HDV looks as good as the best-case, nor as bad as the worst-case. The mid-case is probably a reasonable example of the caliber of results you'll likely get from HDV. (there is no "best-case" or "worst-case" presented for DVCPRO-HD, because it isn't variable -- every frame will perform exactly the same, regardless of what comes before it or comes after it, so in effect every frame is the best-case, and the worst-case. With HDV that's not true: what happens before it, and after it, will affect how much compression is available for the current frame, so you need to take into account best-case, worst-case, and mid-case results.)

For the full test methodology, etc., and discussion, see the thread at:
http://www.dvxuser.com/V3/showthread...ight=dvcpro-hd

What cameras have used? Was it JVC HD1 or 10 and Panasonic Varicam?

Radek

No cameras were used.

That was done using a still photo, and NLE applications. The DVCPRO-HD image was generated from compressing the image in Apple FCP-HD, the HDV image was generated from compressing the image in Vegas.

The point was to test the format, regardless of camera. Once you involve the camera you involve so many other variables -- lens, chips, DSP, etc. This little comparison was strictly on the format, not on cameras.

It is still a somewhat misleading test (and I'm not a huge fan of either MPEG2 or DVCPRO-HD compression for post.) These images have very hard chroma transitions that are designed to show the nature of chroma sampling (important) yet these image types do not occur in the real-world (more importantly optics can't resolve this detail.) Compression tech is designed for natural images, or continuous tone compression -- discontinuities don't occur in real camera source data. Although you have show-cased very well why neither codec is great for post-production, which does introduce discontinuities with the image (from titles, transitions, etc.) If you redo the test, start with a 3MPixel image from a digital camera -- resize this to 1440x1080, 1280x1080, 1280x720 and 960x720 (i.e. as appropriate for your test methology.) Your results wil be more subtle and more (significantly) accurate.

Also you are currently showing interlace HDV compared with 24p progressive DVCPRO-HD (as indicated on the other forum.) The DVCPRO-HD benefits from progressive DCT process whereas the HDV implementation of MPEG2 is interlaced only. I suggest you repeat your test with 60i processing (moving the image each field not each frame as you have shown here -- again real world comparisons.)

Finally you are testing particular encoder/decoder implementations --- every implementation is very different. MPEG uses motion estimation to find matching macro blocks, yet this takes time, so each implementation will/can have different search policies like how far into the image to search. You show HDV breaks down quickly, yet in my own tests with the same footage does not show this. I did a series of pushes to black of lengths 10sec (slow), 5sec, 2sec, and 1sec (very fast -- like a whip pan.) This are pixel displacements of about 5pels, 10pels, 25pels and 50pels per frame -- only the 1 sec whip shows any visible increase is distortion and no where near and badly as you show.

In the long run I do believe 1080i60 DVCPRO-HD is superior to 1080i60 HDV, because 100Mb/s vs 25Mb/s will win out (and DVCPRO_HD only has a 12% res drop over HDV.) It will be worth some real world tests to show whether the difference is significant (for purchase decisions.) Yet for cameras shooting 720p24 I think HDV wins for practical applications as 19Mb/s MPEG2 is a match for 40Mb/s DVCPRO-HD, and the resolution gain of HDV is significant (33% more pixels.)

"It will be worth some real world tests to show whether the difference is significant (for purchase decisions.)"

That was the point I was intending to make. Obviously DVCPRO HD is a better format. Whether it will make a serious difference in actual shooting with a specific camera will have to wait for the camera to be on the market. It may look superb, the camera itself may be better than JVC's or Sony's, who knows? The only meaningful testing is to line up the 3 cameras and shoot the same things under different conditions with them.

So for 1080i60 DVCPRO-HD is better, for 720p24, HDV is better, is what you guys are saying?

Radek

That wasn't really point of my write-up. I was pointing out that the compression analysis and examples where too similistic and not realistic to draw conclusions. Some compressors will have some advantages over others, yet in the end camera differences (and pricing) will have a bigger impact on the purchase discussion.

The compression format alone is only one component of the overall imaging chain. You can't decide that the JVC will be superior to the HVX, or that the Z1 would be superior to the HVX, or that the HVX is superior to either of them, based on the recording format. The lens, the chips, the DSP, and the recording format all play a role.

I understand David's concerns. The test pictures were indeed designed to be hard on the codec, as I said. Real-world, live footage would likely not stress the codecs so hard, but real-world, live footage moves as well, so still-shots are not very representative. I'd love to get ahold of some never-compressed footage and use that to test. Something piped from an HDCAM SR camera, through HD-SDI and captured raw uncompressed... that'd be a good source of footage! My digicam only shoots compressed 7-megapixel, no RAW format... If I had a digital camera that could shoot raw, that might be a source for do-it-yourself progressive-scan footage, but I don't have any way to get ahold of never-compressed HD-caliber interlaced footage. Interlacing could be simulated in post, but I'd rather have genuine camera-generated interlaced footage to start from.

And, it is a purely intellectual exercise, with not a lot of applicability to real-world circumstances until there are cameras around the recording formats. The lens, the chips, the DSP, or other factors may play more of a role in end picture quality than the recording format does, for all we know. I don't know how many would have predicted the level of purple/green chromatic aberration we're seeing in the FX1 lens...

Thanks guys for all your insights - my understanding of the relative merits and limitations of the different low-end HD options (or soon-to-be options) has increased exponentially over a week of visiting this site.

The only trouble is - who can wait another nine months for the release of all these machines, and subsequent evaluation of them?

I hired a Z1 yesterday, and found that shooting doco footage with it presented a number of mechanical challenges after being so used to the PD150. The auto focus override button is not as easy to locate while keeping your eye to the viewfinder, and the external audio options are squirreled away in the menus etc etc - just a matter of getting accustomed to the changes. But it is very easy to focus and handle otherwise.

More importantly I found the images played out of the Z1 through S-video to my widescreen CRT to be very impressive. No doubt watching it on an appropriately resolving screen would be much more revealing.

My point is this - these cameras are now so economical - if you have a doco or other job to shoot now, buy the best available option, then sell it again if you want when the results of the others finally come in. I'm sure a 6 month old Z1 would retain a lot more value than my PD150.

No doubt by the time the Pana is released Canon will be hinting that it's XL-3 will deliver the ultimate budget HD solution - all we have to do is wait another 6 months for its arrival.

Cheers