HD100U and XL H1

[goldyprog]

Any ideas as to what Canon's HDV 24f going to look like compared to the HD100. I have not seen either of their raw footage, except for online clips of JVC's. I think this is definitely worth thinking about and analyzing, since both cameras have their positives and negatives.

But how will the footage look compared to one another? Guesstimate time . . .

[Barry_Green]

No way to know. The only guesses I'd offer are that the JVC is true progressive, and we're not sure exactly what the Canon is yet... the JVC offers the most efficient HDV compression method, the Canon will likely exhibit more compression artifacts... the Canon should (operative word: SHOULD) have a better lens on it than the JVC, and that'll make a difference in the footage too.

In a couple of months we should be able to answer these questions definitively. As of right now it's all just guesswork.

[Spiff_2]

I'll add to Barry's guesses.

It's a battle between:
1440x1080 15-GOP with interlaced processing only @ 25 Mbps
1280x720 6-GOP progressive processing only @ 19 Mbps.

15-GOP is more efficient than 6-GOP, but progressive processing will offer better efficiency over interlaced processing. It is my guess that if the interlaced processing doesn't overwhelm the codec efficiency, the 1440x1080 stream will look signficantly better at the same magnification (meaning - put both up on a 50" display, I think 1080(i/p) will look better).

If Canon implements a progressive processing algorithm it will be no contest. Simply put, 25 Mbps vs. 19 Mbps.

-Spiff

[Barry_Green]

I don't think it can be summed up that easily though. Talking about the numbers in the compression system so thoroughly discounts everything else: the lens, the chips, the sampling, the DSP, etc...

As far as real-world observable detail, if the Canon is using a frame-mode type of recording, then both cameras will deliver about the same amount of detail. If Canon is using genuine legitimate progressive-scan like they did in the XL2, then the Canon could deliver up to 50% more resolution -- and that will show.

Getting back to the numbers: comparing 25mbps to 19mbps discounts the effect that compression has on the signal. 24P HDV is vastly more efficient at compressing its signal into 19mbps than 1080 HDV is at putting its signal into 25mbps!

More bandwidth = better, yes, but only when that bandwidth is put to proper use. Examine it like this: the JVC has 19.7 megabits in which to encode 22.1 million pixels. The Canon has more bandwidth, but has to encode more pixels: 25 megabits to encode 46.6 million pixels. On a bits-per-pixel basis, the JVC is much more efficient: .89 bits per pixel for 720p, vs. .54 bpp for the Canon. JVC's system delivers almost twice as much bandwidth per pixel as Canon's.

But, then again, that discounts the effects of efficiency in the long-form 15-frame GOP of the Canon, vs. the 6-frame GOP of the JVC.

So, let me sum it up like this: this is a question that cannot be answered by appealing to the numbers. It must be seen on the screen to know.

[mezelf27]

Well, I can't get the idea of implementing 15-frame GOPs. The motion estimation artifacts will get the upper hand every time, coimpared to a shorter GOP system.

JVC has done the better thing as far as we've seen (compared to sony that is) AND at a lower bitrate (which is surprising because of the shorter GOPs, but obviously caused by the resolution).

At a datarate just under 25, I think 720 is the only standard that can remain a high quality standard....

[robroysyd]

I don't think you can generalise any of this. As Barry said it "It must be seen on the screen to know", I'd go further than that, it depends on what your shooting style is. If you shoot in the old school Hollywood style where the camera rarely or barely moves then a high res long GOP makes a lot of sense, if you use the camera to hose down the scene then you're sure going to be in trouble.
There's more fundamental issues at play here too that shouldn't be overlooked. All of these cameras now give us an affordable way to produce footage that can hold up on a very big screen but quite regardless of the limitations of HDV putting an image on a big screen is a very different game to putting one on the small screen, that's where I think we all need to start from in this discussion.

[goldyprog]

One of the threads compared the snapshots from "The Hitchhiker" to the JY-HD10, and they did look rather similar to one another.

These snapshots I think have a negative impact on the camera when compared to the one-chip: on the flipside, however, I have seen some very impressive footage from the Gy-HD100. My guess is that the Canon may deliver a non-CF24 shooting mode without the CF's problems, but will still cast a video-like 1080i image.

To put it another way, Canon's version of CF24, but a lot better.

[Walter_Graff]

"I don't think it can be summed up that easily though. Talking about the numbers in the compression system so thoroughly discounts everything else: the lens, the chips, the sampling, the DSP, etc..."

Once again Barry is a voice of reason.

[Graeme_Nattress]

Yes, you can analyse the numbers to death, but there are so so many factors to take into account, which gets especially hard when you add motion and camera shooting style into the equation. Barry's right that you need to do a proper controlled test and shoot the same thing on both cameras to actually get a real answer, and even then, it's biassed by the very material you're shooting.

[mezelf27]

I don't think you can generalise any of this. As Barry said it "It must be seen on the screen to know", I'd go further than that, it depends on what your shooting style is. If you shoot in the old school Hollywood style where the camera rarely or barely moves then a high res long GOP makes a lot of sense, if you use the camera to hose down the scene then you're sure going to be in trouble.

Actually, even motion in the picture is enough to cause the motion-estimation-search to create artifacts. So you'll need to film with a steady camcorder and never shoot anything with with lots of motion... (so not only shooting style contributes: also the subject.) I'm not saying 15-GOP is bad - actually IT IS a more efficient compression, hence I said "surprisingly it has a lower data-rate" - it is just very limited in use for most footage. It doesn't sooth lots of camcorder movement, nor action-shots with steady POV's....