Okay, I finally got ahold of the rest of the JVC footage we shot side-by-side with the DVX, so I can post some more comparative clips. These include outdoors scenes as well as a spot from our 'matrix' spec spot. This should provide a variety of shots instead of just the same gal in the interview setting.

I'm not bothering with the standard-def versions because nobody seems to care, so here are the high-def versions:

http://66.78.26.9/~fiercely/DVXvsJVC/DVX-West-WalkUp.mpg
http://66.78.26.9/~fiercely/DVXvsJVC/JVC-West-WalkUp.mpg

http://66.78.26.9/~fiercely/DVXvsJVC/DVX-West-Gun.mpg
http://66.78.26.9/~fiercely/DVXvsJVC/JVC-West-Gun.mpg

http://66.78.26.9/~fiercely/DVXvsJVC/DVX-Matrix-720.mpg
http://66.78.26.9/~fiercely/DVXvsJVC/JVC-Matrix.mpg

http://66.78.26.9/~fiercely/DVXvsJVC/DVX-WestECU.mpg
http://66.78.26.9/~fiercely/DVXvsJVC/JVC-WestECU.mpg

For those tuning in late, these clips represent a side-by-side shootout between the JVC HD1 and the DVX100/anamorphic/24PA. None of the footage was touched in post in any way -- this is straight off the firewire. While the frames are not exactly identical between each shot, they're of substantially the same subject. Differences in framing are due to the fact that both cameras were rolling side-by-side (except for the 'matrix' clip).

The DVX was then up-rezzed to high-definition 1280 x 720 using S-Spline Pro, using the "photo" preset. That's the only post-processing that was done to any of these clips (and accounts for the "watermark" in the footage. The JVC clips were imported in Vegas and trimmed down to size, then output as a MainConcept MPG2 using the 720/30P preset. The DVX clips were exported using the MainConcept MPG2 720/24P preset.

The question is whether the JVC's higher resolution makes it a more compelling choice, or if the DVX's low-res footage could survive an up-rez to compete against the natively-higher-resolution JVC.

These are fairly large clips, 6 seconds or so each, so they'll take a while to download. They're 1280 x 720 x 24P (DVX) or 30P (JVC) MPG2 files, so to play them back at full speed you'll need a F A S T computer.

"The JVC clips were imported in Vegas and trimmed down to size, then output as a MainConcept MPG2 using the 720/30P preset."

What do you mean trimmed down?

Very good comparison of the two cameras, Barry. Thanks for taking the time to shoot and post the clips. This comparison has definitely altered my perception of the DVX. I know everything in this business is subjective, but just look at the Matrix comparison; (actually there's no comparison) the DVX clearly outputs a superior image (maybe not in resolution, but in every other aspect.) One question, Barry, was there any diffusion on the JVC camera? I thought I saw some flare from the pistol in the West-Gun footage.

The JVC footage is clearly sharper and in all the clips except the matrix footage has much better color. The matrix footage looks worse on the HD because it is trying to brighten the entire scene, this is one of the cases where knowing the camera very well and being able to let go of shutter control and man the exposure lock and lighting to make sure the shutter maintains what you want is the key. All of this is subjective and others have different views. But in response to Kenn Chrisenson's post saying "the DVX clearly outputs a superior image", you are baseing this off of 1 out of 4 comparison clips. Every camera has it's pluses and minuses, same with film cameras and film stocks. There is no perfect tool for all situations, as filmakers, videographers, etc. we have to see what tool is best used in each situation. Again, this is all subjective and everyone is allowed their own opinions on what looks good and what doesn't. How else would half the material being shown in theaters and at home today be let on the market.

Mark Jervis

I like the JVC footage better, but like I said before, I just have an issue with 30p’s limited distribution options (no reason to beat a dead horse there, though.)

Actually, I was basing my opinion on all of the clips. The JVC clearly demonstrated its characteristic green/red noise patterns throughout the exterior footage not to mention the banding in the blue sky during the walk up shot. Look at the actor's face as he moves toward the camera; facial details are muddied as soon as the actor moves, and since we are shooting "moving pictures," I would say that is quite a problem. Before I saw this footage, I was leaning toward the JVC camera, with its sharpness clearly a factor in its favor, but after seeing the color rendition and the effects of the JVC's compression, I'm definitely rethinking my purchasing decision.

<<<-- Originally posted by Kenn Christenson :
... Look at the actor's face as he moves toward the camera; facial details are muddied as soon as the actor moves, and since we are shooting "moving pictures," I would say that is quite a problem. ... -->>>

This isn't a problem, rather it was a choice made may the operator to set the JVC to a much longer shutter speed that the DXV. JVC is definitely shotting with 1/30th shutter speed, and I'm guessing the DXV is shooting at least 1/120th maybe faster. This would favor the DXV for sharpness. Not a good comparison and nothing to do with the compression used in the JVC camera.

Don't think the DVX footage was shot at 1/120th (too much motion blur for that) but that's something Barry could tell us.

<<<-- Originally posted by Ken Hodson :
What do you mean trimmed down? -->>>

I mean I took the 10-15 second capture, found the relevant parts, and exported that. I trimmed down the longer clip to be just the few seconds that would compare.

<<<Barry, was there any diffusion on the JVC camera? >>>

Well, that's a good question! I wasn't the operator on the JVC, Jay Nemeth was. I know he occasionally uses some sort of warm/soft fx filter, but I do not know if he was using it during these shoots.

<<<The JVC footage ... in all the clips except the matrix footage has much better color. >>>

Wow, that's one comment I never expected to hear. I think quite the opposite, of course. The DVX's color rendition is much, much closer to what we were seeing there. The blue sky, his blue jeans, etc. Obviously everyone's entitled to their own opinion, but the earlier clips had me questioning the JVC's color rendition, these clips have me completely convinced that the DVX is much more accurate and delivers a much more pleasing color palette. (and, of course, the DVX color rendition is completely customizable, don't forget).

<<<this is one of the cases where knowing the camera very well and being able to let go of shutter control and man the exposure lock and lighting to make sure the shutter maintains what you want is the key. >>>

I think it's safe to say Jay Nemeth knows the HD1 better than anyone around! He extracted the best the camera was capable of under those circumstances.

"and I'm guessing the DXV is shooting at least 1/120th maybe faster. "

The DVX was shooting at 1/48th. The DVX footage is destined for a transfer to 35mm, so the exact settings for the DVX footage can be found here:
http://dvfilm.com/maker/dvx100settings.htm

These shots are very subjective as to how the cam was operated. The shots are a great comparison (thanks a million Barry, no, two million) but ultimately depend on the cam operators choices. The JVC looks as though a different set up for white balance is in order, as well the exposure should have been lower as can be seen by the glowing teeth in the "Walk up clip". Underexposure with detail pulled from the blacks in post is almost standard order with the HD10, especially if you are going to be S spline pro'ing the DVX footage in post for comparison.
Lighting is also a major factor. You have to light specifically for the HD10, where as the DVX you have much more to play with in adjusting the cam to work with the lighting you have. As far as these clips are concerned both cams use the same lighting.
Again 3 out 4 clips are close-ups. Very hard to judge the value of the JVC resolution with such tight shots. The differance is there, but less impactfull than if we had a comparitive shot of talent against a barn from 30-40 feet out. Then it would be obvious that s spline pro can't pull detail from resolution that isn't there.
The matrix clip shows some differance in resolution as it is the only one that isn't a close-up. Again the JVC is over exposed in that clip and some work in post that equals the DVX's post work, would do wonders for the comparison. To compund problems in the comparison the DVX matrix clip is under exposed making it hard to see how much detail is lost in comparison to the JVC.

PS- how were scan lines from the prop monitors avoided on both cams?

Sorry I didn't realize it was the HD1 and not the HD10. This should be considered when comparing.

"PS- how were scan lines from the prop monitors avoided on both cams?"

They're televisions, not computer monitors, so 1/30th on both cameras took care of it.

We were running the DVX at 1/30th on the Matrix spot because most shots in that spot were shot with the mini35 and we needed every bit of light we could get. That particular shot is underexposed because it needed to intercut with the mini35 shot.


Sorry I don't have more wide clips. The walk-up shot shows wide detail all the way out to the mountains though. I have another old-west clip that is a wide shot, I'll try to work up the comparison and post it later.

Even if the scenes were lit for the JVC camera and the footage was slightly underexposed, the red and green noise seems like a very difficult obstacle to overcome. The noise is in every shot (look at the hat brim in the walking shot.) Then there's the motion compression artifacts I've seen in other footage (esp. the Japanese train footage posted a while back.) Would the noise, the blocky compression and edge enhancement have been less if shot with the HD10?

A bit color correction, a tweek here and there, JVC still shapes up very well. Basically a bit of white balancing would have greatly improved the look. :)

http://members.cts.com/crash/d/dan/temp/DXVvsJVC.jpg

<<<-- Originally posted by Kenn Christenson : Even if the scenes were lit for the JVC camera and the footage was slightly underexposed, the red and green noise seems like a very difficult obstacle to overcome. The noise is in every shot (look at the hat brim in the walking shot.) Then there's the motion compression artifacts I've seen in other footage (esp. the Japanese train footage posted a while back.) Would the noise, the blocky compression and edge enhancement have been less if shot with the HD10? -->>>

HD10U definitely has less edge enhancement. However, any compression noise or artifacts you are seeing in these clips is due to the encoding Barry did; these clips are way below the camera's exported bitrate. More artifacts would be seen in the JVC clips vs the DXV clips similary because there is more detail (which harder to compress.) So again is it is very hard to have a comparison with two cameras with different setups (JVC 1/30th shutter, not white balanced -- see the yellow teeth and the DXV at 1/48th with white balancing), then scale one, then heavy MPEG compress both. The interesting facts that can be gathered are the color saturation and the suitability of upscaling. It is hard to draw any other conclusions.

Do the compression caveats you detail explain the large blobs of noise seen throughout the "Matrix" footage?

There is chroma noise that I believe is from the source. Anything else I believe is due to post compression. I clearly prefer the look on the JVC camera for the Matrix clip, yet is has some significant chroma noise that I would want suppress.

I've posted a few stills, which should remove the MPEG recompression question from the equation.

i.e., I took the .m2t file (as directly captured from the JVC capture utility), put it on the timeline in Vegas, and exported a few still frames.

I saved them as uncompressed .BMP's, rather than JPG's, so there would be no question about any form of recompression. That does mean that they're nearly 3mb in size though, so it'll take a few moments to download them. They're grouped in pairs: the JVC shot, followed by a roughly-comparable DVX shot (after up-rezzing).

Matrix moving pan shots:
http://66.78.26.9/~fiercely/DVXvsJVC/JVCStill-Matrix1.BMP
http://66.78.26.9/~fiercely/DVXvsJVC/DVXSource-MatrixPan036.bmp

http://66.78.26.9/~fiercely/DVXvsJVC/JVCStill-Matrix2.BMP
http://66.78.26.9/~fiercely/DVXvsJVC/DVXSource-MatrixPan116.bmp

Matrix shot at end of pan, camera stationary:
http://66.78.26.9/~fiercely/DVXvsJVC/JVCStill-Matrix3.BMP
http://66.78.26.9/~fiercely/DVXvsJVC/DVXSource-MatrixPan160.bmp

Old West, walk-up shot, at widest (detail all the way to the mountains):
http://66.78.26.9/~fiercely/DVXvsJVC/JVCStill-WestWalkUpWide.BMP
http://66.78.26.9/~fiercely/DVXvsJVC/DVXSource-SteveWalksUp003.bmp

Same setup, but at the end, close-up shot:
http://66.78.26.9/~fiercely/DVXvsJVC/JVCStill-WestWalkUpClose.BMP
http://66.78.26.9/~fiercely/DVXvsJVC/DVXSource-SteveWalksUp092.bmp

Interesting how these BMPs clearly show the JVC has more chroma noise, yet these also clearly shows the DXV has more luma noise.

Okay, one last set of clips, this time a wide shot outside.

http://66.78.26.9/~fiercely/DVXvsJVC/JVC-West-Wide.mpg
http://66.78.26.9/~fiercely/DVXvsJVC/DVX-West-Wide.mpg

(man, I've gotta be near my bandwidth limit this month!)

One can clearly see that resolution is not a problem with the JVC shots. Again, my problem with using the JVC is the large blobs of chroma noise. In trying to match the looks of the "Matrix" shots I noticed that as I adjusted a levels filter in Photoshop I came up with even more pronounced blobs of noise (something that's never occured in shots I've adjusted in DV.) I don't think I'd have as much trouble with the chroma noise if it was more "grain-like" as in the DVX shots, but then that's probably a result of the Bayer-type filter used to acquire the color information on the JVC.

My Mac says, "Cannot open file because it is not a file Quicktime understands"?????

Anyone having this message?

Luckily- VideoLAN was able to open it.

The "blobs" of chroma noise can be greatly reduced with this VirtualDub filter. http://www.ifrance.com/freevcr/virtualdub/cnr-en.html

<<<-- (man, I've gotta be near my bandwidth limit this month!) -->>>


No doubt. Thanks again for your time and bandwidth.
Ken.

The JVC looks more 'film like' than the DVX, especially in these two. The DVX details look too 'modeled' and blobby. The JVC quantizing does look a bit bizy on my computer monitor, but still the material looks less 'electronic' because of the textures being represented more faithfully.

Thanks for posting these Barry! JVC should send you a cheque :)

-Les

<<<-- Originally posted by Barry Green : Okay, one last set of clips, this time a wide shot outside.

http://66.78.26.9/~fiercely/DVXvsJVC/JVC-West-Wide.mpg
http://66.78.26.9/~fiercely/DVXvsJVC/DVX-West-Wide.mpg

(man, I've gotta be near my bandwidth limit this month!) -->>>

hey,

ure right about the chroma noise on the JVC. but i dont think its a problem of the camera itself!!!!!

ive seen a lot of night footage with the jvc and the real `WOW` from the people who saw it (and they post it on the forum) was that it has NO NOISE at night. they dont produce noise like the dv produce - can you people remember it?

ciao

Guys DO NOT FORGET BARRY HAS THE HD1. You cannot take this footage as a basis to evaluate the HD10. Differences are too noticeable. The HD1 has way too much edge enhancement and lacks in certain bright areas some of the subtelties of the HD10. Maybe you should take a look at Darren Kelly's DVD witch made comparisons between the HD10 and other cameras.

"The "blobs" of chroma noise can be greatly reduced with this VirtualDub filter. http://www.ifrance.com/freevcr/virtualdub/cnr-en.html"

Unfortunately, chroma noise reduction (or luma for that matter) works at the expense of resolution, reducing the only advantage the JVC camera has.

<<<-- Originally posted by Kenn Christenson : Unfortunately, chroma noise reduction (or luma for that matter) works at the expense of resolution, reducing the only advantage the JVC camera has. -->>>

This is incorrect. With the suggested filter, luma is only used as a key channel to determine where chroma noise reduction should be applied (it is quite clever.) The resolution for chroma and luma is not reduced. I have used this filter with a fair degree of success, however it does inspire me to develop a version for Aspect HD. Give it a try.

Was the footage uprezzed from a Pal DVX100 or NTSC

Holly crap. the Chroma noise on the JVC is tremendous. This has to be the disadvantage of the single CCD. Anyone who said anything about single CCD being good for digicams thus good for vidcams is just wrong.

After seeing the wild west wide shots, the JVC shots contain a wee bit more detail than the DVX, but the edge enhancement on the JVC looks like crap. And what happened to that blue sky in the JVC?? I assume that clip wasn't color corrected? It just looks totally unatural and looks very much like its got a stylized color correction done to it.

"it has to be the single chip making that chroma noise..."

What basis you people have for this??? Don't you realize that there are a lot more variables in this cameras design than 'chip count'?
I highly doubt the extra blocking is because of the ccd. It is much more likely that it's the mpeg2 compressor. There is a way of masking that, by some post processing of the chroma channel. That does not change detail much, as the luminance channel holds what people see as detail.
I've done some restoration work on film where I used the luminance channel from the film ( at 2K res ) and the chroma from a video tape ( 720 res ). It looked fine. The film had some water damage, but it's lum. chan was still ok.

-Les

Yang Wen,
That is the look you get when you don't white balance. You should ignore the color of the sky, it has more to do with the camera setup than the camera itself. Edge Enhancement is on HD1 not on the HD10. These tests can only prove that the HD1 has more chroma noise than the DXV, and the JVC has better resolution and lower luma noise. Little else can be concluded.

<<<-- Originally posted by Mark Grgurev : Was the footage uprezzed from a Pal DVX100 or NTSC -->>>

NTSC.

Barry - Did all of the DVX shot use the mini35?

None of these clips used the mini35, they were all shot with the DVX and anamorphic adapter.

We shot the "matrix" spot on the mini35, but we took a few comparison shots with the anamorphic adapter, and those are the shots I used in these comparisons.