Hvx200 Low light vs Sony Z1 and other questions

[geo]

I totally agree with your observations . . . +6dB looked funky, and +12dB was unacceptable. I feel your pain on your non-A DVX100 . . . I did virtually the same thing.

A friend of mine just bought the DVX100A.... a week before the DVX100B came out. And was kicking himself. I'll buy an HVX200...... if and when I need it. Not before. Because who knows what might be down the road as far as improvements or pricing?

That's one of the things I got from Henry Ford's autobiography: never buy anything for business before you need it. Even if it is a "good deal". Nothing is a good deal unless you need it. You always come out ahead paying retail at the last moment. I found this out when I DPed a show this summer: We needed Dedolights and Panel-lites. I found out that I would be able to buy the kits and rent them to the show for the 9 weeks we were going to shoot..... that was June. The lights came down last week. 9 weeks became 6 months. Even though I paid probably way too much with express shipping/rush order and all that, They have paid for themselves 3x already. I was toying with getting KINOs just before that... if I had they would have sat in boxes unused...

Sorry... that's a tangent. For those who have instant use for the HVX... have fun! I don't at the moment but will happily slap down the 7 grand when owning one becomes profitable. I don't want to get one just for the sake of having one and see the "A" version come out a week later....

Very cool little camera, though.

 

[Barry_Green]

Will the hvx200 shoot as good as the Sony z1 in low light?

Better. It will spank the Sony Z1 in low light. It'll be about a stop faster, with about 1 to 1.5 stops more dynamic range, at comparable noise levels. The HVX may be a tiny bit noisier, but it'll be less noisy than a Z1 with 6dB of gain (which is what you'd need to try to match the sensitivity of the HVX). Note: this is all subject to change related to the final balancing done to the production model of the HVX!

Will there be some sort of lens adapter to put on other lens
such as wide angel or fish eye.

Definitely.

And final question what software will work out with the uncompreesed hd? Vegas?

Vegas can work with uncompressed HD, yes. But the camera doesn't record uncompressed HD, it records compressed DVCPRO-HD.

How will I get the footage into the computer?

For uncompressed HD? You're on your own on that one, I'm sure some sort of Kona or AJA or BlackMagic card would work, but you'd need a SATA RAID also. For DVCPRO-HD? Easy -- you can plug the card into a reader, or you can firewire the data into your capture application, or you can link to your computer through firewire (on Mac) or USB (on PC) and copy the files over from the card.

firewire or do i have to get expensive reader of some sort too?

Well, you could use firewire, or you could use an expensive reader, or you could use a cheap $70 USB-to-CardBus adapter. Lots of choices.

 

[Barry_Green]

So i take it most of you guys think the panasonic
is a better camera than the sony z1.

Drastically better in oh so many ways that it's hard to count them all -- better picture, better color, more latitude, better sensitivity, better compression, sharper image, true 1080/24p instead of the absurd CineFrame 24 of the Sony, true 1080/30p instead of the resolution-losing CF30 on the Sony, true uncompressed audio instead of the Sony's compressed MPEG-2 audio, 4 channels of audio instead of two, plus there's the whole 720p mode, plus variable frame rates, plus DVCPRO50, plus a much much much better manual focus assist, plus true cinegamma, plus a wider lens, plus true manual zoom, plus tapeless acquisition (no dropouts)...

More expensive but way more features.

More expensive? It's $46 more. Granted you'll need to buy a card or two, but then the Z1 needs tape, so the cost differential depends on your workflow. I really don't see a market niche for the Z1 anymore. The Canon spanks it as far as 1080 HDV, and the Panasonic is a far superior camera in nearly every measurable way.

The only thing the Z1 really has going for it is NTSC/PAL switchability. But its price is way too high as compared to the HVX. Now, the FX1 has a nice market niche -- same image as the FX1, but at $3,000 it's half the price of the HVX. But I don't see too many more Z1s getting sold, when it's about the same price as the HVX but it's less than 1/2 the camera that the HVX is (when all features are considered).

 

[Barry_Green]

How is it that people decouple speed and noise like this? Isn't the SNR in low light the defining characteristic of its low-light performance?

-Spiff

How do you *not* decouple it? Sensitivity is what it is. We don't have infinite control over the gain, so we can't dial it in to match it exactly. If we could control gain in 1/2dB increments, then we could do what you suggest, and choose our SNR exactly. But we can't. So you look at the image in the given light conditions, at the given stop, then you look at the noise, and you pick your poison. And you can't get cleaner than 0dB on any of these cameras (well, the Canon used to allow -3dB, but I don't know if the new one does). So if you get a slightly cleaner image but a stop slower, how does that help you if the image is too dark?

The only thing we can do is see what footcandles it takes to render a scene to a certain IRE value, and then make a balanced judgement regarding the noise at that level. And then if a satisfactory IRE value is not attainable, you have to decide whether the additional noise from gain is worth it, or you decide to just say that no, you cannot acceptably render that scene with that cam at that light level.

At 0dB, the Z1 is a little bit cleaner, and a stop slower. Put 6dB of gain on it and it's about the same brightness as the HVX, and a lot noisier. So pick which interpretation you want, but there's not much else we can do about it.

And, again, the HVX isn't finalized -- they can still fine-tune the relationship between noise and sensitivity, so we won't know for sure what the actual comparison is to the Z1 until they finish the software.

 

[Stevet]

And, again, the HVX isn't finalized -- they can still fine-tune the relationship between noise and sensitivity, so we won't know for sure what the actual comparison is to the Z1 until they finish the software.

Barry, after having said that, what is your gut feeling there will be an improvement to noise level? Not that the current footage shows much noise at all , but it would be nice to gain a stop with 6dB and still stay clean. Hey, here's to hoping for the best!

Also, do you think a - 3dB option is handy?

Heck, if you've can pump up the light, you would be able to drop the noise level at that given exposure. Any thoughts on this?

Steve

 

[Zack Birlew]

Me personally, I think the more options the better. -3db gain? I honestly don't know exactly what that does, but I'll take it cause I'm easy like that! =)

 

[alisterchapman]

Noise is critical

Noise is critical

Noise, or rather a lack of it is critical. With any compressed format any noise is seen as picture information and uses up valuable codec bandwidth. With HD transmissions being compressed to hell any noise is going to push up the amount of visible artifacts, thats why Discovery won't accept 16mm film for HD programmes, too much grain (noise)! I would much rather have a camera with a low noise level at 0db than one with more noise. The Z1 is a good example of this, at least with the Z1 if you need more sensitivity you can bring in more gain, but if there is already noise at 0db then your stuck with it, like the old JVC HD1 a camera that could produce some stunning images, except for the noise which in turn was pushing the MPEG encoder to it's limits all the time adding all sorts of artifacts to the final output..

A Z1 shooting a reasonably static frame at 0db is not pushing the codec at all and I would challenge anyone to see any artifacts, but a more noisy camera shooting the same scene is going to be pushing the codec harder before you even add in any movement. OK the panasonic codec should be more robust than HDV, but I've seen some pretty nasty artifacts from varicams that have had gain switched in and hence more noise. If you can see the noise on a monitor at a demo or show than I think it is cause for concern, as in most cases your going to be a least a couple of feet from the monitor and I would not expect noise to be visible at 0db from the component or uncompressed output of a modern camera.

 

[xray]

More expensive? It's $46 more. Granted you'll need to buy a card or two, but then the Z1 needs tape, so the cost differential depends on your workflow. I really don't see a market niche for the Z1 anymore. The Canon spanks it as far as 1080 HDV, and the Panasonic is a far superior camera in nearly every measurable way.

No actualy it is $ 2.400 more. Without p2 cards. If you buy two 4GB P2 cards ( I say the minimum for a working flow) the HVX200 is $3700 more.

You can buy 2 Sony Z1s for that ( 3600 E for one)

The only thing the Z1 really has going for it is NTSC/PAL switchability. But its price is way too high as compared to the HVX. Now, the FX1 has a nice market niche -- same image as the FX1, but at $3,000 it's half the price of the HVX. But I don't see too many more Z1s getting sold, when it's about the same price as the HVX but it's less than 1/2 the camera that the HVX is (when all features are considered).

Not true, Sony sold 37.000 Z1 cams worldwide already. Its price as above mentioned is available by globalmediapro and very attractive. And you save the $ 1.800 P2 drive and the extra weight you have to carry. I think the HVX200 + P2 drive (AJ-PCS060G ) with battery is about 1.6 kg more than a Sony Z1 with tape.

 

[spencer]

Just askin again... Can someone explain Gain and stops to me? I gather faster stops are better, and gain probably does something with the look (the only gain i really know is amp related, so i'm going off of that), but I'd like to further understand the two.

 

[Barry_Green]

Barry, after having said that, what is your gut feeling there will be an improvement to noise level? Not that the current footage shows much noise at all , but it would be nice to gain a stop with 6dB and still stay clean. Hey, here's to hoping for the best!

Noise vs. sensitivity go hand in hand. Increase one, you increase the other. I don't know which way they'll end up, but I can say that the version 4 camera shown at DV Expo was dramatically quieter than the Version 2 cam that they shot the "Boom, Baby, Boom" music video with.

Also, do you think a - 3dB option is handy?

Oh, I'd love that, and I cornered the camera engineer and point-blanked him on it. I don't know if it'll make it in or not; they may be skittish about making menu-level changes in the operating system at this point. Fine-tuning the existing options is one thing, changing the menu structure is another. But I hope they add it!

 

[Ralph Oshiro]

Just askin again... Can someone explain Gain and stops to me? I gather faster stops are better, and gain probably does something with the look (the only gain i really know is amp related, so i'm going off of that), but I'd like to further understand the two.

I'm not the smartest guy on the block to explain electronic stuff here, but since no one answered you . . .

Increasing your aperture increases the amount of light passing throug the lens. Increasing gain is just like your "amp related" knowledge. The signal is electronically "boosted" and with that, you get an increase in NOISE. That's why gain is bad. Every +3dB of gain equals 1-stop of "exposure" increase. Most cameras have three default gain-up settings of (typically) +6dB, +9dB, and +12dB.

 

[Ralph Oshiro]

Me personally, I think the more options the better. -3db gain? I honestly don't know exactly what that does, but I'll take it cause I'm easy like that! =)

You know how +3dB makes your picture noisier? Well, when you have ample light, many pro cameras have the ability to enable a -3dB gain for the opposite effect: 1-stop less sensitivity traded for reduced noise.

 

[doccutter]

Just askin again... Can someone explain Gain and stops to me? I gather faster stops are better, and gain probably does something with the look (the only gain i really know is amp related, so i'm going off of that), but I'd like to further understand the two.

I think perhaps a way to think of gain v stops is the idea of an optical vs. a digital effect, even though that's not technically correct. F-stops, by opening and closing a mechanical aperture, control the amount of light that passes through the lens to hit the CCD's. Aperture settings should not effect noise level at all if you're set to 0db gain, the noise you get, regardless of aperture is not going to change. Gain works in the realm of electrical or digital amplification. If your image is too dark, not only are the details difficult to see, but you've lost latitude.

Imagine a scale, like a ruler, that goes 0 to 100. If your image is too dark, underexposed (or too bright; overexposed) then instead of having your luma information spread nicely all over the scale, it's bunched up towards one end. What gain does is simply take the information from the CCD and add numbers to the values produced by the CCD. So if the brightest part of the image is 70 IRE, gain will simply mathematically add IRE to the image, stretching the data over the scale. You can see this effect in Photoshop if you take an image that is dark, then apply levels or brightness/contrast, then look at the histogram. Sure, now the data is spread all over the scale, but now you've got these big holes spread up and down the histogram. Effect? You lose detail, since you can't get back the latitude you lost by having your image crushed to one side of the scale. Simultaneously, since amplification is never perfect, it adds random noise to the image. (Photoshop, when working with clean images, is pretty good about not adding noise, but it only has to work on one image at a time, and has an unlimited amount of time to do so). Additionally, the gain setting, by pushing up the IRE values of pixels, is making the inherent noise of the CCD more obvious. The result when recorded to digital compressed formats is noise and heightened compression artifacts. Ewww.

 

[Stevet]

Oh, I'd love that, and I cornered the camera engineer and point-blanked him on it. I don't know if it'll make it in or not; they may be skittish about making menu-level changes in the operating system at this point. Fine-tuning the existing options is one thing, changing the menu structure is another. But I hope they add it!

I'm hoping Panasonic was listening to you Barry. I think the addition of the -3dB option would be great!

Thanks,
Steve

 

[Anders Holck]

many pro cameras have the ability to enable a -3dB gain for the opposite effect: 1-stop less sensitivity traded for reduced noise.

Most Camras will also loose a stop or two dynamic range in the -3db setting, which is bad.

Also note that most cameras use the CCD amplifier to do white balance, so some extreme color temps will boast more noise even in 0db setting.

 

[Antoine_Fabi]

Let's hope the HVX final product will have a clean and quiet image.
It is critical for professional use. IMHO
Much much more important than artificial sensitivity.

 

[lpcvideo1]

. . . Much much more important than artificial sensitivity.

In marriage it is very important. :grin:

 

[spencer]

Ohhh, so gain increases noise but increases light at around the same time? That makes sense.

 

[Stevet]

Ohhh, so gain increases noise but increases light at around the same time? That makes sense.

Yes, It's signal-to-noise. As the signal goes up, so will its noise floor.

Steve

 

[jbeale]

3 dB = 1/2 f-stop

3 dB = 1/2 f-stop

> Every +3dB of gain equals 1-stop of "exposure" increase.

1 stop is 6 dB (see * below). I just tried it on my FX1. f/5.6 at +18 dB looks the same brightness as f/2.0 at 0 dB .

For any straightforward camera design, -3 dB gives you 1/2 stop less sensitivity, and 1/2 stop less dynamic range *if* your sensor was noise-free. But whether it actually affects the useful dynamic range depends on whether your DNR is constrained by ADC quantization, or the sensor noise on the dark end. With a noisy sensor, the effective DNR may be the same.

*Note: dB is a power scale: +3 dB is twice the power. But the output of a photodiode (used in both CCD or CMOS sensors) is a current proportional to light intensity, which is passed through a transimpedance amp to become a voltage. Increase the aperture one f-stop for twice the light, and you get twice the voltage from the sensor. In electronic systems, Power = (Volts)^2 / (Resistance). Twice the voltage gives you four times the electrical power (6 dB). The camera gain operates on the video signal, in the post-CCD-sensor electronic realm, where 3 dB is only 1.414 times the voltage.

Now, there are amplifiers which operate in the optical realm (eg. a laser is a self-oscillating optical amplifier). If you had a 3 dB optical amp, it would give you one f-stop more sensitivity.