Smear versus dynamic range?

[Lucia de Nieva]

Evaluating the martial arts footage around now I was quite shocked to encounter obvious vertical smear, a technical limitation that seemed to be overcome with the DVX. At least with our AG-DVX100AE we never had any trouble with it. This is really strange, since Jan stated that the HVX has a better dynamic range than the DVX, so column overflow in the CCDs should be even less likely. Too bad that we already sold the DVX in anticipation of its HD successor. Worse indeed that Juan's excellent idea originally intended to be a 50$ homebrew solution was sold out to become an outrageous commercial experiment.

When will the marketing divisions understand that progress means the addition and not the exchange of features?


Please find Jan's statement here:

http://www.dvxuser.com/V3/showthread...hlight=dynamic

[Jeff Kilgroe]

Originally Posted by Lucia de Nieva

Evaluating the martial arts footage around now I was quite shocked to encounter obvious vertical smear, a technical limitation that seemed to be overcome with the DVX. At least with our AG-DVX100AE we never had any trouble with it.

I have a hard time believeing that you never encountered vertical smearing with the DVX. Just about any shot with a saturated red against a high-contrast background would produce it. It was a limitation of the DV25 codec and color smears and color blocking is just a way of life for DV and present in the DVX and XL2 etc..

I haven't analyzed the martial arts clip hardly at all due to the poor recompression. It really isn't worth the time. On the latest 108024p "guitar" clip, there's a lot more to look at. Once again, it's hard to tell with the WMV re-encode, but there are definitely some issues that deserve attention. The most prominent, which could be confused in some situations with a vertical color smear, is the apparent low vertical resolution of the CCD block. Some vertical blurring/softening is present and on the oversharp contrast edges (especially in the beginning with the rocks and later with the guitar close-up), also the oversharp edges seem to exhibit a vertically elongated characteristic.

I will reserve final judgement until I see some native clips from the camera or get to play with an HVX myself. At times I wonder if the HVX200 is a wise purchase, but then I consider all my other options. None of the HDV solutions on the market work for me and my compositing needs. HDV is taking the color data the wrong direction as I have already outgrown my use for DV25. Even if the resolution of the HVX isn't quite as high as the resolution we get out of a Sony Z1, the 4:2:2 color, true progressive frame rates and higher data rate more than make up for it. At any rate, I'm confident that I'm making the right decision. If for no other reason that the HVX200 is by far the most affordable camera that meets my needs. If I could afford a CineAlta I would... That HD-950U with an 800Mbps DVCAM SR deck sure is sexy... And costs as much as 15 nicely loaded HVX packages.

[Lucia de Nieva]

Well, I referred to the perpendicular white line over the whole height of the image emitting from the large lamp on the right. This would occur also with an uncompressed data path, since it originates from the overloaded columns in the CCDs. With the DVX, you can capture strong and small lights (which mean more load to less cells) in pitch black environments without these "split" lines.

[Jeff Kilgroe]

Originally Posted by Lucia de Nieva

Well, I referred to the perpendicular white line over the whole height of the image emitting from the large lamp on the right. This would occur also with an uncompressed data path, since it originates from the overloaded columns in the CCDs. With the DVX, you can capture strong and small lights (which mean more load to less cells) in pitch black environments without these "split" lines.

Oh, I guess I missed something in your original post. Anyway, I think now were getting into the limitations of current CCD technology with these scanning rates in conjunction with the pixel/cell density in relationship to the 1/3" censor and limitations of the optics and fixed lens. I wonder how much Panny actually can do to combat this with the available CCD tech and they have to draw the line with the optics at some point to keep the camera within this price range.

[Barry Green]

That was a 1200 watt HMI shining directly into the lens. I dare say that any CCD camera would smear when pointing directly into a 1200 watt HMI! I mean, that's equivalent to about 5,000 watts of tungsten.

Compare that to the clip of the XL H1 where a tiny flashlight shined into the lens caused full-height vertical smear, and worse smear than the HVX exhibited out of a (let me say it again) 1200-watt HMI! Far from being concerned, I'd say this is extraordinary performance. I'll have to shoot some car headlights etc., but I think the HVX will prove to be the most smear-resistant of all the HD cameras. I've seen obvious vertical smear from the FX1 and HD100 off of car headlights.

[Lucia de Nieva]

Honestly Jeff, I would say the question is not what they can do, but what they want to do. For instance, a decent digital still camera with a fixed 7x wide angle zoom lens by Zeiss or Leica is around 750 $, with the technology used being basically the same like that found in digital video cameras. Panasonic as well as Sony offer both video and still cameras, and it is more than likely that the R & D of both product lines is done mainly by the same people. And yet the customers are supposed to swallow that an 8MP lens for 2/3" comes cheaper than an 1.1MP lens for 1/3". A P2 card is around 650 $ for 4 GB, consisting of four 1 GB SD cards at 50 $. The HVX uses a Linux OS, but can't send the same data written to the P2 cards via IEEE1394 or USB using ATA commands. Come on, no technical limitations here.

And Barry, you are right concerning the FX1/HD100 smear, but would't you expect also that the successor to a good product at least preserves the standard set by the lower model? Again, the DVX gives no vertical streaks in similar situations. Maybe the ABG (anti-blooming gate, a current overflow foil) hasn't be adapted properly to the new chips yet.

By the way, the .m2t files from the H1 look horribly in terms of latitude and noise. Reminds me of the L1/L2. Retro style really seems to be en vogue at the moment.

[Barry Green]

You're arguing against the laws of physics though. Smear sensitivity is directly related to pixel size. If you want more pixels, you're going to have to give up in other areas. You cannot make such a fundamental change and expect other items to stay the same. The XLH1 has 1.5 megapixels, very very very tiny pixels -- so the tradeoff for high resolution is that you get much poorer dynamic range, and increased sensitivity to smear. The DVX's pixels, by comparison, are massive -- and thus much more smear-proof.

If the DVX100B was more prone to smear than the DVX100A was, then yes I would clearly say you had a point. But when you go to a unit that has anywhere from 2 to 5 times as many active pixels, those pixels are going to be necessarily smaller, and the smaller they are, the worse your sensitivity, signal-to-noise ratio, smear performance, and latitude are going to be.

[Jeff Kilgroe]

Originally Posted by Lucia de Nieva

Honestly Jeff, I would say the question is not what they can do, but what they want to do. For instance, a decent digital still camera with a fixed 7x wide angle zoom lens by Zeiss or Leica is around 750 $, with the technology used being basically the same like that found in digital video cameras. Panasonic as well as Sony offer both video and still cameras, and it is more than likely that the R & D of both product lines is done mainly by the same people. And yet the customers are supposed to swallow that an 8MP lens for 2/3" comes cheaper than an 1.1MP lens for 1/3". A P2 card is around 650 $ for 4 GB, consisting of four 1 GB SD cards at 50 $. The HVX uses a Linux OS, but can't send the same data written to the P2 cards via IEEE1394 or USB using ATA commands. Come on, no technical limitations here.

You're just making assumptions cross-referencing spec numbers between different products. Unfortunately, there are many more factors in delivering these products than what you have touched on. Yes, the CCDs in still cameras have higher resolution, but they must make trade-offs to get there... For instance, still cameras aren't scanning the CCD at 60Hz... The pricing of P2 cards has been discussed at length here and until the recent price drop, I was one of the biggest complaintants here. Now comparing 1GB SD chips that actually meet the requirements for P2 - zero defect rating, 3.2MB/s sustained write speed, 1,000,000 minimum cycle life across the entire chip, suddenly you find yourself looking at SD chips that cost $125 from the cheapest of the bargain vendors and carry MSRPs of $180 or more. 4 * $125 = $500 and if you figure in a fair market value for a decent interleaveing, embedded memory controller and PCMCIA packaging that will put the price about $525. Considering that all the currently available P2 cards are from a single supplier (Panasonic) and stamped with their own brand name, we can expect to pay a premium... In this case it's about 15% which brings us to the $625 price tag that most vendors are quoting right now, with MSRP still being $649.

Out of all the complaints you have, the only one I agree with is the last one. You are correct that there is absolutely no technical reason why direct recording to a firewire or USB2 hard drive can't be implemented. However, as much as I hate to grit my teeth and type this, I actually think Panasonic made the right choice by not offering such a thing. Why? Because it introduces far too many unknowns. If Panasonic allowed this, then suddenly they would have to deal with countless irate and irrational customers who are unhappy with the HVX200 becasue it can't record to their generic HDD that has poorly implemented firewire interface and someone esle can't record to their iPod (yes, I'm sure there are people dumb enough to try). Hard drives also don't provide the security and stability of a solid-state device and of the currently available solid-state technologies, SD was the right choice and it still takes at least 2 or 3 SD chips in sequence to achieve the necessary write speeds. Put 4 chips into a commodity interface (PCMCIA) and suddenly it's a format that most any computer is capable of reading or obtaining a reader for without introducing a new standard.

[John Jay]

Here are some smear values for typical cams

2/3" DSR 570 Smear -120 dB
1/2" DSR 390 Smear -115 dB
1/3" FX1/Z1 Smear -107 dB

all manufacturer quoted values

[Lucia de Nieva]

First of all, thanks for the numbers, John. The DVX's smear resistance was announced around -100 dB as far as I remember. Very interesting too, that the differences between 2/3 SD and 1/3 HD are only marginal.

Pixel count of the HVX is around one megapixel, that's twice the number of the DVX in NTSC, and I was speaking about the experience with the PAL model. So we are dealing with a 1/4" pixel density in SD, that is to say two 1/4" SD CCD together would give one 1/3" HD CCD, if you take my meaning. And then Juan's work proves that the CCDs in the DVX are very close to HD by default.

I am always puzzled how easily people are willing to burn their currency-converted lifetime for technology already outdated upon its release. Seriously, regarding the last decades it is pretty fact that you can't buy anything which hasn't been in the drawers for at least five years. It may be difficult to help this principle, but the minimum should be pushing the margins by using means like this forum for example. As long as you give the excuses for crippling functionality yourself, you can't expect anyone but Jesus to do better. Indeed there is always one good reason for the industry, named profit, printed in bold letters on the doomsday ticket.

Back to the specs my comparison stands. Any of the decent digital still cameras mentioned before record 15 FPS progressive from a 2/3" inch area four times the size of a 1/3" CCD which hence can be scanned in a quarter of the time needed for 2/3" giving 60 FPS. Any objections? However, this is just an approach, but without these we probably could not communicate the way we do now, being busy with collecting firestones. Anyway I would speculate that plugging in an USB drive is less prone to user errors than messing around with a NLE suite on a laptop. Also, potential HVX users are not exactly everyday consumers, are they? Then Jan assured that 7200 RPM 2.5" drives would be fine for direct streaming. Panasonic could recommend housings and chipsets the same way, couldn't they? Problem is, who else but the ENG guys would buy the P2 cards then?

Remember one thing: When one day there is a really fine camera without a catch, the cause is not some executive's paroxysm of acute benevolence, but the customers' - our - constant articulation of constructive yet demanding criticism. This was what made the DVX what it should have been in the first place - the A version. So keep on nagging. !;-)

[Michael Struthers]

C'mon Luis, are you kidding? These corporatons have to make money - this isn't a socialist world. They have to make a unit that will sell in the hundreds of thousands and still make a small profit on. You want a varicam level quality for 6K? What is the motivation of someone to sell you one for 6k?

I'm sure you will write your story so well that no one will be noticing the vertical smear in your smash Sundance feature.

[Lucia de Nieva]

.. just check out the price for the AG-DVC180A which is the name for the AG-DVX100A in China. And they still make good money from it. No, this is no socialists' world, but a continuum of global mass inertia where you have to evoque impacts to get things moved. Well, let's skip the politics. So Mike, you have a point. In the end it's just another tool, becoming high-tech junk in a few years. (sigh)