HD100 vs. HVX on the Las Vegas Strip

[Barry_Green]

Okay, I apologize in advance. Things don't usually get under my skin, but tonight they did -- I got all rankled up by someone on another forum suggesting that the HVX is barely more than standard-def. When I demonstrated with the res charts that this simply isn't so, someone else weighed in and said "when the camera moves, it drops to standard-def." Preposterous notion (having been there when the basketball footage was shot and having used this camera every day for four months, I would think I'd know whether that happens or not), but I thought I'd go out and find out what the real "street truth" is. Lab tests are one thing, but what does the footage look like when you actually shoot something? Especially under less-than-ideal circumstances?

Strap on the gloves, boys, and let's go boxin'.

I rigged an HD100 and an HVX200 side by side and went down on the Las Vegas Strip. Settings were attempted to match cameras; on the HD100 I used Tim Dashwood's Cine-D scene file. On the HVX I used Cine-D+CineLike Matrix. I also attempted to match sharpness by using -6 on the HD100 and +7 on the HVX. The result made for some grainier HVX footage than the last time I did this; last time I was aiming for noiseless (so it may prove interesting to compare last time with this time). But, regardless, back to tonight: both cameras were wide open all the time, and I tried to match focal length as best I could, which worked for the street and the first part of the Sirens show, didn't work so well for the second part of the Sirens show... Both were in 24p; HD100 in 720/24p, HVX in 1080/24p and downrezzed to 720 for this comparison (because A, I refuse to tie one of the HVX's hands behind its back, and B, knowing what I was going to shoot, I wanted to get to the bottom of this "overexposure artifacts" question.)

Footage was brought into Vegas (because FCP won't handle HD100 24P, nor HVX 1080/24p) so I used Raylight, 'cause I still don't know how to run Edius! Timelines were lined up so that all the frames you see here are from the exact same moment in time; frame-accurate.

-----

Shot a little bit of the Strip, at wide angle, which should hopefully once and for all dispel the notion that the HD100 is in any way sharper:
HD100: http://www.fiftv.com/HVX200/HD100-n-...00-Strip-1.JPG
HVX200: http://www.fiftv.com/HVX200/HD100-n-...VX-Strip-1.JPG

and

HD100: http://www.fiftv.com/HVX200/HD100-n-...00-Strip-2.JPG
HVX200: http://www.fiftv.com/HVX200/HD100-n-...VX-Strip-2.JPG

The way I see it, they're about comparable sharpness, comparable detail rendition. This was mainly a static shot, although the cars were moving. Focus was set on infinity for this shot. HD100 shows tons of purple aberrations and vertical smear that the HVX just doesn't; look at the car headlights, or the streetlights, especially the blue aura on top of the streetlights -- what is that, anyway? Color resolution is also noticeably higher on the HVX -- look at the red & green traffic lights, look at the blue liquid in the cup against the red dress on the sign, look at the clarity in the word "ENEMIES" on the street sign in the HVX shot, vs. the blockiness in the HD100 shot. Look at the clarity of the word "Wynn" under the "ENEMIES" and compare. No doubt about it, the HVX is way higher in chroma res than the HD100. Comparable luma detail, and higher chroma res. And this was at full wide angle on the HD100, and 6.3mm on the HVX for equivalent field of view.

So, okay, that's a static shot. How about moving shots? Well, how about the torture test, the "Sirens of T.I." show? This is a worst-case scenario for HDV, and the Sony failed miserably, turning into lego blocks. The JVC did much better in that regard. Still fell apart when the explosions and strobelights overwhelmed it, but when watching the footage back on the monitor it was way more acceptable than the Sony footage. The shorter GOP really helps the JVC recover quicker. But it still had lots of issues, as you'll see.

So, I pulled a bunch of shots out of the timeline. This is, again, with the cameras matched as best I could get 'em in regards to settings, and using the same focal length and the same aperture (JVC actually a little more open, as its lens goes to 1.4 vs. the HVX at 1.6).

Note: ALL THESE SHOTS WERE HANDHELD, so there's motion in every shot. Not only handheld, but two cameras handheld (mounted on a bar so it was one big rig) so it's less stable than a single handheld camera would have been. These are all moving shots, and I tried to pull shots that represented movement as well as represented the biggest disparity between the two cameras, and a shot or two where nothing was going on at all which should show a "baseline." All these shots represent identical moments in the timeline, they are sync'd up frame-accurate (although obviously they could be as much as 1/48 of a second apart in "real time" because the cameras weren't genlocked).

Shot #1:
HD100: http://www.fiftv.com/HVX200/HD100-n-...0-Sirens-1.JPG
HVX200: http://www.fiftv.com/HVX200/HD100-n-...X-Sirens-1.JPG

Don't ask me what's with the purple on the HD100, that's just the way it decided to render the shot. As you can see in the HVX shot there's purple on the wall, but... um... the JVC went kind of purple-happy. The HVX shot is much more accurate as to what it looked like live. And no, focus is not "off" on the JVC, it is precisely focused. Any softness is due to the codec not resolving well; you'll see in other shots that the JVC was focused accurately.

Look at the detail in the bricks in the wall. Is anyone going to stand here with a straight face and claim that the HVX shot is "standard def" and the HD100 shot is "high def"? Didn't think so. Myth #1 shattered, thank you very much. But let's keep going, there are lots more examples:



Shot #2:
HD100: http://www.fiftv.com/HVX200/HD100-n-...0-Sirens-2.JPG
HVX200: http://www.fiftv.com/HVX200/HD100-n-...X-Sirens-2.JPG

No explosions this time, just some residual fires. But look at how much detail the HVX retains in the mainly purple/blue wall, vs. how blurry/soft the HD100 got. That should be proof enough of 4:2:2 vs. 4:2:0; solid colors make 4:2:0 softer. Look how the HD100 develops 8-bit gradient banding around the fires/smoke, vs. how well DVCPRO-HD handles the same issue. Night and day.




Shot 3:
HD100: http://www.fiftv.com/HVX200/HD100-n-...0-Sirens-3.JPG
HVX200: http://www.fiftv.com/HVX200/HD100-n-...X-Sirens-3.JPG

Another shot with no smoke or fire, this is some quiet time. More of a "baseline" shot. But still the 4:2:2 shows a whole lot more detail in the bricks than the 4:2:0 of the JVC. Framing is off, one is a lot higher than the other, but compare the crow's nest on the mast: relative magnification should be in the same ballpark.



Shot 4 (otherwise known as "the clincher"):
HD100: http://www.fiftv.com/HVX200/HD100-n-...0-Sirens-4.JPG
HVX200: http://www.fiftv.com/HVX200/HD100-n-...X-Sirens-4.JPG

I'm not gonna even comment, the pictures speak for themselves. Remember, exact same microsecond in time, exact same exposure, and comparable camera settings. Only thing I'll say is: notice how the detail in the wall came back in the JVC shot -- now that the wall's not blue, now that it's illuminated, the JVC isn't limited by 4:2:0 and can resolve the detail again. That proves it's not "bad focus" or anything like that; the lousy wall detail rendition in the earlier shots is because of 4:2:0.




Shot 6 (aka "the REAL clincher"):
HD100: http://www.fiftv.com/HVX200/HD100-n-...0-Sirens-6.JPG
HVX200: http://www.fiftv.com/HVX200/HD100-n-...X-Sirens-6.JPG

Hey, it's your choice, you can choose to shoot on either camera. I know which one I choose... and if anyone ever wants to see what SSE ("split screen effect") looks like on the HD100, this is a good example -- look at the crowd along the bottom of the shot, how they're gray/orange on the left side of the screen and rich black on the right side.



Shot 8 (aka "running up the score")
HD100: http://www.fiftv.com/HVX200/HD100-n-...0-Sirens-8.JPG
HVX200: http://www.fiftv.com/HVX200/HD100-n-...X-Sirens-8.JPG

Okay, okay, I admit, I'm piling on with this shot, but hey, I think the HVX looks flat-out amazing in this shot. Looks like a scene from "Pirates of the Caribbean 2". The HD100, on the other hand, has gone to macroblock hell, with a little SSE thrown in for good measure.



Shot 10 (or, "speaking of macroblock hell"):
HD100: http://www.fiftv.com/HVX200/HD100-n-...-Sirens-10.JPG
HVX200: http://www.fiftv.com/HVX200/HD100-n-...-Sirens-10.JPG

I didn't use any other shots from this side of the show because I blew the zoom setting (the HD100 is way more tele than the HVX is on this side of the show). But I gotta show this shot. This is what happens when HDV is not happy. When too much strobelighting happens, HDV falls apart, and this is what happens. To the JVC's credit, it did it a LOT LESS than the Sony did, JVC's codec is much more robust and resilient than the Sony, but it still happens. Would have happened more if I'd shot 30P, but I was trying to "keep it real"; shooting 24P stacked the odds in JVC's favor as much as possible.

Framing and telephoto were impossible to match on the fly: there's a stupid post in the way on the left of the HVX shot, and the telephoto's inequal -- with about 1,000 people in the crowd it wasn't all that easy to jockey for position so I had to settle for what I got.

-----------------


Long post, yes. Am I mad? Well, yeah, I guess I was; I was just outraged that someone would sit there with a straight face and say "the HVX is barely more than standard-def." It's just patently untrue. Say what you want about the pixel count, but when it comes to resolved detail the HVX more than holds its own. This is what the real pictures look like.

But folks, it's like this: if you want to buy an HD100 because you like its look, you like the interchangeable lens feature, you like JVC customer service, you don't like P2, whatever... go ahead. There are plenty of reasons to want an HD100. It's a good product. But please don't go saying that it's sharper, or that it resolves more detail, or that "the HVX is 'standard def' when shots start moving," or nonsense like that... otherwise I gotta go out and put it to the test and go proving what's what.

Whew.

I've calmed down now.

Thanks for listening.

[Cees Mutsaers]

Very convincing set of shots. The only shot i thought the HD100 was a tat sharper was shot #3. When you look at the stairs it looks a bit sharper. On all the other shoots the HVX200 is the MASTER with respect to sharpness, vibrant colours, smearing, purple fringes. Actually I did not see any macroblokking at all on both cams but probably I am not trained enough to see it. Barry you conviced me that the HVX is very good Now just wait until we in PALL country can buy one. Shouldn't it be released over here one of these days?

[ullanta]

Very slick, thanks! I still wish I could get an image of you standing in the crowd holding that monstrosity over your head... in terms of detail level, look in the first set of pictures (strip-1) at the streetlight to the right of the bottom of the red dress... where'd the curved part go in the HD100 shot? That's not chroma, that's luma!


P.S. I like the way you lined up the SSE in shot 6 with the mast on the boat!

[ecking]

Anyone that says the hvx isn't hi-def is nuts.
I honestly think both cameras performed pretty good considering what you put them through.
I think in day to day operations(not exploding pirate shows) each of the 4 major hd cameras are good enough.

But I don't understand why you didn't just shoot the hvx at 720/24p instead of downconverting?

[stabwound]

Barry:

I admit to being one who's conceded ( based on various reports) the JVC has better resolution. But when I saw your pics was so happy to change my mind.

Here's a 300 percent pixel peep of a section in photoshop:



I wasn't sure what you were talking about until I enlarged the section... and fell off my chair. Viewed normally I couldn't spot the JVC's shortfalls. Then you notice the Jaggies on closeup. I'd also say the JVC was noisier.

I suspect the "Jaggies" and noise contribute to the illusion the JVC has a sharper image. Reminds me the coarsed grained Tri=X black and white film, which "seem" sharper than the fine grain (and higher resolution) films, which gave the impression of being "soft".

If some people call the HVX image "soft", well then, I can live with that.

Great comparison. Thanks.

Off topic sidenote: What's with my rubber eyepiece!?? It's always covered in dust no matter how often I clean it...

[Barry_Green]

Originally Posted by ecking

I think in day to day operations(not exploding pirate shows) each of the 4 major hd cameras are good enough.

So do I. I think in normal circumstances you should be choosing between them based on the feature sets, rather than obsessing over pixels.

But I don't understand why you didn't just shoot the hvx at 720/24p instead of downconverting?

Two reasons. One, because the HVX is higher def in 1080. In 720 the HVX records 960x720, in 1080 it records 1280x1080. 1280x1080 downrezzes to 1280x720 very nicely. Makes for ultimate 720p. And if I'm going to go comparing resolution, I'm not going to tie one hand behind the HVX's back, I'm gonna let it do what it does best. And 1080 is what it does best. I would only use 720 if I needed maximum recording time on the cards; otherwise it's 1080 all the way, all the time. It's what the camera does best.

Second reason: people have been saying there's "overexposure artifacts" in 1080 mode. I knew those explosions would push it into overexposure, so I wanted to see if I could find what they were talking about. I detect no artifacts whatsoever, so that's Myth #2 busted.

[steindj]

Originally Posted by Barry_Green

look at the clarity in the word "ENEMIES" on the street sign in the HVX shot, vs. the blockiness in the HD100 shot.

Better yet, look at the word "Mayweather" above "ENEMIES." It's completely illegible in the HD100 shot, just one red smear.

[Barry_Green]

Originally Posted by stabwound

I wasn't sure what you were talking about until I enlarged the section... and fell off my chair. Viewed normally I couldn't spot the JVC's shortfalls. Then you notice the Jaggies on closeup. I'd also say the JVC was noisier.

A lot of what you're seeing there is the 4:2:0 color resolution, vs. the HVX's 4:2:2.

In 4:2:0, a block of four pixels is all set to the same color; the only thing that differs among them is the brightness. That's why the red is bleeding below the lines; because the red is assigned to all four pixels in a block like this:
XX
XX

All four of those pixels (coords 0,0; 1,0; 0,1; and 1,1) are assigned the same color in 4:2:0. Whereas in 4:2:2, every line gets its own color sample. The HVX's 4:2:2 is assigning color like this: (0,0; 1,0); on the next line they get their own color (in this example, YY), so a completely different color can be assigned to (0,1; 1,1); or, graphically,
XX
YY

Where XX is one color, and YY is another color.

So you get twice as high color resolution.

Which should forever put to rest the myth that the HVX isn't a real 4:2:2. So that makes Myth #3 busted.

[Barry_Green]

Originally Posted by steindj

Better yet, look at the word "Mayweather" above "ENEMIES." It's completely illegible in the HD100 shot, just one red smear.

Again, due to the color sampling I would bet.

[SergejIvanovits]

Originally Posted by ecking

Anyone that says the hvx isn't hi-def is nuts.

The definition of HD has been rewriten. They says everything better than SD is an HD camcorder now. Remember there is a 880Mbit 4:4:4 recording possibility in the HD world as well.

Originally Posted by ecking

I honestly think both cameras performed pretty good considering what you put them through. I think in day to day operations(not exploding pirate shows) each of the 4 major hd cameras are good enough.

100% agree.