Prosumer-Perception

[Bruce Dempsey]

How can anybody be taken seriously if you show up with an itsy bitsy camera like the cx550?
So change the perception for a couple of dollars!
this is my tricked out cx550v now it looks like an uber camera too

[Arkady Bolotin]

Hi Bruce,

Sorry I missed your post yesterday, but everybody’s celebrating Rosh Hashanah down here and I’m barely holding up.


Looking at your pictures I only can say: Wow!

But that is to say, to dress the camera like this costs more than a couple of dollars, isn’t it?

I can recognize some expensive pieces of your dressing: Manfrotto tripod fluid head, Rode VideoMic, matt box...; all in all it’s close to three hundred bucks, right?

By the way, why do you need the matt box?

P.S. Happy Hebrew New Year to you!

[Dave Blackhurst]

I just stuck a modded A1U lens hood on mine, got about 80% of the impact <wink>. Who needs the fancy tripod for handheld with the excellent OIS <wink, wink>. I've got a couple different shoulder rigs that make mine look pretty tough! I think I need to upgrade my mic though, yers is bigger than mine... (sorry, just couldn't resist!)

Seriously though, nice looking rig, although I think you've lost the entire stealth/guerilla film-maker vibe, they might take you seriously and ask for your permits!

[Bruce Dempsey]

Hey thanks for the Happy Hebrew New Year Arkady. Sounds like your celebrations were 'over the top'.
The half decent video head is of course de rigeur regardless of camera but If I may say the cx550v is simply amazing in its ability to smooth out hand held footage. I filmed my son at a skateboard contest and the resulting footage really astounded me. I'm definitely old school when it comes to tripod use for paying video gigs. Cameras are on video heads period but my boy wanted a low key recording as not to put him under more pressure so I handheld and it's beautiful, sort of floats along without the usual jarring associated with hand held shots and for several hours. Sony has done a beautiful job in the stabilization protocol on this camera.
Not In the pix is a 3 input passive microphone mixer (azden) mixing the audio from the shotgun and a wireless unit.
The matte box is for effect only. In fact it is a cheap barndoor set on a tulip lens hood which is mounted on a 37 - 52mm step up adapter and screwed into the lens
the video from this camera seems better than my FX7 and certainly superior to my HC1s
This rig simultaneously records hd-mpeg4 to it's internal memory and mpeg2 thru the s-video port to an attached DVD recorder and when I push the photo button, it records an 8megapixel still to the pro-duo flash memory card. Quite a feat for such an itsy bitsy camera. No other that I know of will do the three outputs as beautifully and at the same time.
It uses the same sensor as your AX2000 albeit just a single.

Dave -stealth mode by undressing the rig

[Ron Evans]

Unfortunately Bruce, the AX2000 and NX5U do not have a "R" sensor that is on the CX550 or even my older XR500 and consequently the smaller cams do a better job of grain control in low light. Side by side my XR500 has a cleaner picture than my NX5U or the FX1 . It lacks the depth of image of the NX5U and a little less resolution but it is very close for something that is 4 times less in cost!!!!! I had hoped the NX5U would have the "R" sensor as I changed from my FX1 because the XR500 had a better picture most of the time !!!

Ron Evans

[Arkady Bolotin]

Ron,

I wouldn’t claim the quality of the CX5500 image is wholly owing to the superiority of this new R (back-illuminated) sensor.

Theoretically speaking, back-illuminated sensors themselves are prone to problems such as noise, dark current, defective pixels and color mixture that lead to image degradation and may cause a decrease in the signal-to-noise ratio.

Therefore Sony has made something to improve R-sensors.

But Sony also advanced its “old” time-proof Exmor sensor used in NX5/AX2000. "Column AD conversion" function has been developed which further eliminates noise from the initial stage of signal conversion, thereby helping to minimize noise even at high sensitivity.

As a result, Sony has created a system capable of recording approximately 3.6 times more image data (3,680x2,070) than a 1920x1080i system and hence bright, high resolution images.

So, whatever the reason of the CX5500 image excellence, it’s not just the sensor.

[Ron Evans]

I can assure you that my XR500 picture has less noise than either my NX5U or my FX1. All my projects are in the theatre with high contrast lighting that most of the time is dark. Video noise in the dark areas is very noticeable. The SR11 with the original Exmor chip has the noisiest picture, the FX1 is next in line then the NX5U and the cleanest is the XR500. I assume the CX550 is very similar to the XR500 as I don't think they changed the sensor.

The whole marketing material from Sony on the "R" sensor was low light performance with less grain and I can attest that this is true.

Ron Evans

[Dave Blackhurst]

The low noise signature of these cameras is certainly a great feature, especially when shooting in bad/low light. Even the "low lux" mode stays pretty clean.

AFAIK, the entire -X-5-- series uses the same sensor. The menus changed between the XR5-0 and the CX5-0, and I think they squeezed just a tiny bit more out of the chip in low light, which is typical of the CX's when they were released behind the HC/SR/XR series cameras - they had a bit more "tweak time" before the CX release, and so the CX's might perform just a tad better - with the XR550/CX550 they released at the same time, so that wouldn't apply with this product cycle.

There was definitely a boost in OIS performance between the XR500 and the later models - they really smoothed out the "roll" axis significantly, while the other two were already quite good.

SO far, I still like the CX550v the best overall for features/size/weight, although I kept a couple CX500's around since the image quality is almost identical (sometime I've got to try side by side 17Mbps vs. 24mbps to see if I can see any difference!).

[Arkady Bolotin]

Ron,

As I mentioned before, compared to conventional front-illuminated sensors, back-illuminated ones commonly cause problems. It is so because moving the active matrix transistors to the back of the photosensitive layer normally leads to cross-talk, which causes noise, dark current, and color mixing between adjacent pixels.

To overcome this Sony has developed a novel on-chip lens optimized for back-illuminated sensors, that achieves a higher sensitivity of +6dB and a lower random noise of -2dB without light by reducing noise, dark current and defect pixels.

Second, it’s Sony’s BIONZ image processing engine installed in CX550v. Working together with an Exmor R-Sensor (back-illuminated), BIONZ noticeably eliminates noise during data conversion and image processing.

That was I meant when I said, “it is not just the sensor”.

[Ron Evans]

Arkady,
I accept that one must always look at the whole system rather than just a single element.
The sensitivity will be higher in the R sensor because the sensor area is greater than on the similar sized Exmor. Instead of the space for connection tracks between the sensor elements there is only a guard band which is smaller. This may account for all the sensitivity increase. For a single chip design with Bayer filters this may be very different than on a 3 chip design.
The AX2000 and NX5U came out after the consumer range with R sensors and with the Bionz processor that is now used in almost all Sony camera products though of course programming will be different. Under almost all conditions that I have tried, in side by side comparisons my XR500 has less grain noise than my NX5U. Since they both have a Bionz processor I therefore drew a reasonable conclusion that the R sensor does make a difference that Bionz programming on the NX5U could not achieve with the standard Exmor chip. In fine testing it may be that there is a resolution difference that accounts for the grain etc. There is a big difference in black level settings at default. SR11 and XR500 are close to 0 IRE as expected but the NX5U is more like 5IRE without correction and at default will give a washed out grainy image in comparison to consumer cameras.

Ron Evans

[Arkady Bolotin]

Ron,

You have touched upon an interesting topic concerning video noise in the dark areas. This is the actual problem plagued especially HD video acquisition.

The reason for that is quite simple and lies in the nature of high definition: the more photosensitive pixels across the sensor you have the higher the probability of photocurrent from them without light.

As of year 2010, three principal ways to overcome that noise (or at least minimize it up to a bearable level) are known.

First, we can cool the sensor to the degree of virtually no current. Such approach is used in astronomy systems (with astronomic costs, for sure).

The second way is to enlarge the size of each single pixel and consequently the size of the sensor itself. Such method is used in the high-end digital cameras capable of recording the highest resolution video.

The third way is to increase the photon cross-section within sensor pixels, i.e. the probability of photon capturing. This way is realized in back-illuminated sensors, which can capture over 90% of descending photons in comparison with only 60% for front–illuminated sensors.

As you can guess, no one of those techniques is perfect.

As to back-illuminated sensors, this method works well when the size of a sensor is relatively small (for example, iPhone4 has a back-illuminated sensor). With increasing sensor size, the effectiveness of the back-illuminated technique is dropping fast. For instance, 1/2 inch R-sensor would have almost the same low-light performance as the usual one.

[Cliff Totten]

Size is king!

It's not about the size of the entire sensor, it's the physical size of the "photsites" on the sensors that is going you the sensitivity to keep noise down. The optimal amount of sensor photosites ("pixels") for 1080 video is around 2.1 megapixels. If you have less?...you need to interpolate to arrive at 2.1. If you have more?...you still need to interpolate and "throw away" pixels to get down to 2.1 megapixels.

With all things being equal (lens and post processing) this is what you get;

Sensor A - 1/3rd inch with 5 million photosites = TONS of interpolation & TINY photosites=not very sensitive and shows artifacts on resolution charts.

Sensor B - 1/3rd inch with 1 million photosites = Interpolation with BIG photosites= VERY sensitive but shows artifacts on resolution charts.

Sensor C - 1/3rd inch with 2.1 million photosites = Clean, full raster image with moderately sensitive rating and shows little interpolation on resolution charts.

Anyone know that exact "pixel" (photosite) count on the Sony CX550V? Is it 5 million? If so, that's 3 million too much for 1080 video. (prolly nice for stills though)

The Sony CX550V IS a nice camera but I dunno,...I seriously doubt that it can compete in sensitivity with a 1/3rd x3 sensor block. I believe the AX2000 has three 1.1 megapixel sensors and much larger glass in front.

The problem with "consumer" models is that they design the sensors for BOTH video and high megapixel stills. (and that is a sensor design contradiction)

CT