New Sony 1/2" 1080p HDV camera, Sony's 6th HDV camera

[Radek Svoboda]

http://bssc.sel.sony.com/Broadcastan...p=141&id=80177

Street price 18,500 USD, with recorder, 35 mm adapter, lenses, etc.

Sony has this listed in HDV cameras. Is their modified X300 POV camera, has CCD's with 1080p24 mode. Writes progressive pictures into interlaced tracks, same as 100,000 USD Sony CineAlta.

Has many output options, including HDV. HDV avaialable through Firewire. Can get from it 1080i50-60, 1080p24-30, record into FX1, Z1, HDV recorder, or new one chip HDV cameras. Can record uncompressed footage to computer, hard disc recorder, or HDCAM or HDCAM SR recorder. Sound will have to be recorded separate.

True HD Canon lens for it lists 7,000 USD, has both manual, autofocus mode.

For about 25,000 USD street price you get 1080p system with HDV recording, LCD display, Canon HD zoom lens, W/A converter lens, battery, audio recorder and AT shotgun mic. Sony X300 with lens cost as low as 18,000 USD street price; this X310 about 1,500 USD more.

You be able get superior 10 bit uncompressed performance for film production by adding computer or portable recorder. Picture quality should be superior to 100,000 USD F900 recorded to tape, that is used for major motion picture productions.

For 25K USD you get HDV recording, for 30K you get 10 bit computer recording in superior quality .

Can save 6,500 USD by not buying Canon lens, instead using one of low cost 35 mm adapter, 35 mm SLR lenses.

That is 18.5 USD total for 1080p package with lenses, recording.

Next best thing would be Sony F950 1080p camera, costs some 120,000 USD without lens, LCD, recorder.

Sony now has 6 HDV cameras: FX1, Z1, Qualia, two single chip cameras and this high end 1/2" POV HD/HDV camera.

I wonder how long will be it for this camera become available in camcorder form. Sony has all ingredients already. Now they have to package them into one true 1080p unit.

Maybe this camera should have own subforum, is HDV, is available now, and would attract more knowlegable members that use higher end equipment.

Radek

[Radek Svoboda]

This camera could be addition for someone who has a Sony HDV camera and NLE. He can use this camera to record to his HDV camera, get true 1080p, or to NLE, get 10 bit 1080p. Would be for film productions mainly.

Someone could figure out how t replace camera mount with C-mount so low cost industrial HD lenses could be used.

Radek

[Jeff Sayre]

Originally Posted by Radek Svoboda

Picture quality should be superior to 100,000 USD F900 recorded to tape, that is used for major motion picture productions.

I do not think the picture quality of the HDCX310 will be superior to Sony's HDW-F900. The F900 has three, 2/3 inch CCDs each with 2.2 megapixles. It outputs higher resolution than the HDCX310. Besides, the HDCX310 is not a piece of equipment meant to be used for field production. Whereas I'm sure it could be done, there would be many undesirable shortcomings.

Originally Posted by Radek Svoboda

Next best thing would be Sony F950 1080p camera, costs some 120,000 USD without lens, LCD, recorder.

Again, Sony's HDC-F950 is significantly superior to their HDCX310. For starters, it also has three, 2/3 inch CCDs each with 2.2 megapixles, not 1/2 inch. This allows for much greater resolution. Second, it is provides full-bandwidth digital 4:4:4 HD RGB signals outputted through HD-SDI or a single optical cable.

There are clear reasons why Sony's two, high-end cameras cost so much.

Read the specs on these two high-end Sony cameras:
HDW-F900: http://bssc.sel.sony.com/Professiona...00_v11122b.pdf
HDC-F950: http://bssc.sel.sony.com/Broadcastan...s/hdc-f950.pdf

[Radek Svoboda]

X310 has extensive menues, adjustments.

All cameras are Sony, have 1080p24 mode.

1. X310: 1/2" CCDs, 4:2:0, 8 bit, 1480x1080 pixels, MPEG2 HDV compression 4.5-15:1 (varries in different shots).

2. F900 recorded to tape: 2/3" CCD, 3:1:1, 8 bit, 1440x1080 pixels, 4.5:1 compression

3. X310: 1/2" CCDs, 4:2:2, 10 bit, 1480x1080 pixels, uncompressed.

3. F950 recorded to tape: 2/3" CCD's, 4:4:4, 10 bit, 1920x1080 pixels, 2:1 MPEG4 compression

Qualitywise uncompressed X310 is between F900, F950 cameras. Actually native 1440 pixel count is better than 1920 converted onto 1440, which is F900's case. In addition pixels of X300 are nearly as large as in F900.

Radek

[Thomas Smet]

I actually agree with Radek on this one. I almost recently bought one of these SD industrial cameras. SONY has one that uses the exact same image block as the DSR-390. Of course that is when I didn't plan on going HD.

When we have people today saying the FX1/Z1 looks almost as good as a F900 with only 960x1080i CCD's, 4:2:0 color and 1/3 CCD's you can expect a 1/2", 4:2:2, 1440x1080p, uncompressed to bridge that small gap even more.

I know 2'3" is better than 1/2" but at that level we are only talking a 2-3% quality change for a fraction of the cost.

The difference would be like comparing a SONY VX2000(1/3) to a DSR-390(1/2) and then comparing that DSR-390(1/2) to the DSR-570(2/3). There is a huge difference in the first comparison but a much smaller difference between the second comparison.

[Thomas Smet]

Does the optional computer XGA output board mean we could get 4:4:4 RGB video off of this thing?

[Mike Farrington]

I guess this means I'll have to cancel my order for the HC1. :rolleyes:

Edit: Hey, what is this? No emoticons on these forums? That's no fun. I was just going to delete the post instead of complaining about it, but I see I can't do that either. Blah. Oh well.

[Jeff Sayre]

Okay, we have two different issues we're debating here. The first, is input quality. The second is output quality.

I'm sure that members can (and probably will) debate all day long about whether output quality is more important than input quality. However, let's step back and look at the basic physics of what we do with our fancy cameras.

No matter how you slice it, CCD size and pixel density per CCD is the penultimate determinant of maximum, possible resolution--and that means maximum, possible image quality. That is it period.

Yes, the quality of the lens, the experience of the operator, the hardware and software features or limitations of the camera, do all contribute to what is finally outputted. But the CCD is king.

Here's a very simple example. Take two cameras with two different 3-CCD configurations. Everything else is constant during acquisition and signal processing: each camera uses the exact same lens, aperture and shutter settings are the same, shooting occurs under the same light conditions, both cameras shoot the same scene, and they both have the same 12 bit A/D converter. The only difference is that the first camera has 16 pixels on a 1/2-inch CCD and the second has 16 pixels on a 2/3-inch CCD. If camera one can output data at 10-bit 4:4:4 RGB and camera 2 can output only at 4:2:2, which camera do you think produces better, outputted video?

My guess is the second camera because each pixel is larger, thereby being more sensitive to light, experiencing less noise, and requiring less amplification before going to the A/D converter. Each of these factors help create a higher resolution image. Therefore, output options are not the most important factor in determining quality.

---
Why is this the case?

Remember, the light impinging on each pixel of each CCD is analog (of course). After the light excites a pixel on a CCD, the charge is transferred to a read out register. From there, it gets electronically amplified and then sent to the A/D converter.

So, once each of the 3 CCDs have been read (which occurs before signal amplification and A/D conversion), the total possible resolution of the focused image has been determined. No additional, true resolution can be created through hardware or software. It is CCD size and pixel quality alone that determine maximum, possible resolution.

Of course, resolution can appear to be increased through interpolation, but this does not add any new information. It just adds virtual pixels--virtual meaning that the data is created electronically and not from actual photons hitting actual, physical pixels.

---
Are there difference between pixels on CCDs?

Yes. A pixel is not a dimensional unit of measurement. It is a physical area that can vary in size depending on the manufacturer's design requirements for a given brand of CCD, monitor, etcetera. One pixel on a CCD can occupy a different area than another pixel on a different CCD. Therefore, you cannot calculate CCD square area by using the number of pixels on the CCD chip. You need to know the actual width and length of the image sensor.

Furthermore, there is something called a megapixel CCD which are traditionally used very effectively in digital still cameras. However, some manufacturers use them in low-end video cameras. The high end video cameras use non-megapixel CCDs.

So, what is a megapixel CCD? It is CCD that crams as many physical pixels into a given space as possible. The results for video acquisition can be less than ideal. I read this quote from Barry Green on a different forum that makes this point quite well:

"Megapixels hurt image quality, that's all there is to that. So using an anamorphic adapter on a megapixel CCD would return worse results than using the adapter on a non-megapixel CCD. Cramming megapixels onto a CCD makes each pixel tinier, which reduces its ability to gather light and also reduces its potential latitude. *If you want bad low-light performance and no latitude, get a megapixel video camera. Megapixel CCD's are good for digital still cameras, where you need the additional resolution for the still frame. *But they're lousy for video. A properly-designed camera would have enough pixels to create the frame, and no more. That way the pixels would be as large as possible (to gather more light, better s/n ratio, etc) while being as small as necessary to deliver maximum resolution. Look at broadcast cameras: you'll just about never find a megapixel ccd in a broadcast camera. They have just enough pixels to fill the frame, and no more. More pixels than necessary just leads to aliasing and resizing issues, which can result in *softer* pictures (i.e., less actual resolution!)"


Don't be fooled by the number of megapixels on a CCD. More pixels on the same size area is not a good thing. However, if the number of total pixels increases along with the size of the CCD, then you may be talking about actual increased image quality.

---
Let's look at the CCD chips in the three cameras were discussing in this thread:

What is the difference in the available light-gathering area between a 1/2-inch chip and 2/3-inch chip? Or, to put it another way, what is the maximum, potential resolution for each chip?

CCD chips are sized at a 4:3 (horizontal:vertical) dimensional aspect ratio or a native 16:9 dimensional aspect ratio. Since I cannot find anywhere the actual dimensions (width and length) for the native 16:9 CCD sensors, all my calculations are based on the 4:3 format. If anyone has the source documentation for the actual dimensions of native 16:9 CCD chips, please let me know!

[NOTE: The "inch rating" is a throw back to very old nomenclature used in the days of vacuum tube television sets. Actual dimensions of CCD chips are different than advertised. Visit this link for more information on 4:3 CCDs: http://www.dpreview.com/news/0210/02...nsorsizes.asp]


Here are the calculations for the 4:3 CCD chips:

Data:
- 1/2 inch 4:3 chip is really 4.8 mm by 6.4 mm with a diagonal of 8 mm
- 2/3 inch 4:3 chip is really 6.6 mm by 8.8 mm with a diagonal of 11 mm.

So, doing the math:

1/2 inch CCD: 4.8 x 6.4 = 30.72 square mm
2/3 inch CCD: 6.6 x 8.8 = 58.08 square mm


% Difference: ((58.08-30.72)/30.72) * 100 = 89% more surface area on the 2/3 inch CCD chip.

That's significantly more potential resolution!

---
Now, let's look at the issue of pixel density on each chip:

For the Sony HDC-X300: it uses a megapixel chipset, each with 1.5 million pixels/chip. That equals 48,828 pixels per square millimeter.
For the Sony HDW-F900 and HDC-F950: they each have 2.2 million pixels/chip. That equals 37,878 pixels per square millimeter

Therefore, there are fewer pixels per square millimeter on the 2/3-inch HAD chips. Each of the pixels is 22% larger which allows for higher-quality image acquisition.

Remember, these calculations are based on a 4:3 aspect ration and not the actual 16:9 aspect ratio so the numbers are only representative. However, the point is still valid that the larger pixel size on the CCDs of the higher-end Sony cameras equate to higher resolution input. This can result in higher quality output.

----
Now, what happens with the maximum, inputted resolution is all up to the camera's electronics and software. A great chip offering superb resolution may not result in a superior data stream if the camera designers compromised on hardware design. This point, is a heavily debated issue on this forum.

Why would camera designers settle for less than the maximum, potential outputted resolution?

Several reasons:

Technological Considerations
* Current technological limitations of their design and/or supporting products (capture devices, NLS systems, broadcast standards, etc.)

Sales-side Considerations
* Price point differences in their own model line where they want buyers to trade up to better quality output

Marketing Considerations
* Waiting to take full advantage of potential resolution in their product line until market conditions indicate timing is right

---
Camera comparisons --or, getting the facts straight:

Sony's HDC-F950
CCD Chipset: 3-chip 2/3-inch type 16:9 HAD FIT
A/D Converter: 12 bit A/D converter
Maximum Output: can output 4:4:4 digital RGB with a dual link HD-SDI (otherwise it's 4:2:2 with a single HD-SDI cable)

Sony's HDW-F900
CCD Chipset: 3-chip 2/3-inch type 16:9 HAD FIT
A/D Converter: 12 bit A/D converter
Maximum Output: The output to tape is 4:2:2, not the 3:1:1 that Radek suggests. It may allow for 4:4:4 output when used with a dual link HD-SDI capture card like a Blackmagic Design's DeckLink HD Pro 4:4:4 card but I cannot find any documentation support that theory.

Sony's HDC-X300
CCD Chipset: 3-chip 1/2-inch type 1.5-megapixel HAD CCDs (note: these are megapixel CCDs and it is not clear if they are native 16:9 or masked 4:3 chips)
A/D Converter: Cannot find information
Maximum Output: Unclear from what I've read
24p requires 2:3 pulldown

End of big, long post.

[Radek Svoboda]

F900 records 3:1:1 to tape, 4:4:4 is not available

X310 has 16:9 CCD's, 24p requires pulldown on F900 too. Sony has nice brochure on X300, maybe too on X310, check Sony site. Check brochure. You'll be impressed.

Whatever I said originally is correct.

Radek

[Jeff Sayre]

I was not clear in my intended statement of the F900 and I apologize.

What I meant to say after my long-winded post, is that due to the superior resolution of Sony's CCDs in their HDW-F900, the luma to chroma sampling rate of 3:1:1 is actually as good of output as 4:2:2 in a few other HD cameras.

In fact, here is an excellent article explaining how the CineAlta F900 (at 3:1:1) has identical color space output as Panasonic's Varicam (at 4:2:2). Just visit this link and go to the "Riddle#2' section: http://www.jkor.com/peter/riddles.html

Don't get fooled by output sampling rates. They do not necessarily imply superiority. What matters most is the resolution of the CCD. If that is superior, it can allow for a lower sampling frequency and still produce the same high-quality output as cameras with higher sampling frequencies but lower CCD resolution.

The CCD is where maximum, potential image quality is determined. Not at the output. After the photon has hit the CCD, quality starts dropping. The better the CCD, the more room there is to let the quality drop as the signal gets passed from CCD to output channel.

Sometimes 3:1:1 is as good as 4:2:2. Don't get caught up in these rates thinking they indicate absolute quality. These numbers are simply ratios that you can use to help determine what the final output quality will be. The final quality depends on how much original light was sampled. And that is determined by the quality of the CCDs.

(If you want to be confused even more, when referring to HD sampling rates, the use of 4:2:2 and other rates is technically not correct. In HD, the correct sampling rate for the above example would be 22:11:11. It's a long story but it is another example of why sampling rates are not an absolute indication of quality. 22:11:11 sure looks a lot better than 4:4:4 doesn't it!)

[Thomas Smet]

3:1:1 HDCAM is only comparable to the 4:2:2 Varicam if you look at just the number of chroma samples. Yes 960x720 4:2:2 = 480x720 chroma ; 1440x1080 3:1:1 = 480x1080. So there are the same number of horizontal samples. The issue with chroma is the ratio of chroma to luma pixels. HDCAM has a much higher luma value but the chroma in relation to that is fairly low. With HDCAM you basically have 960 interpolated or duplicated horizontal pixels of chroma. Varicam only has 480 interpolated or duplicated horizontal chroma pixels.

Another thing is you cannot really compare chroma depth between a 720 and 1080 format. Now if you looked at the DVCPro HD 1080 version you get 1280x1080 and still 4:2:2. That means you now get 640x1080 chroma instead of 480x1080 if you were to compare apples to apples.

When dealing with very small details such as hairs and other thin objects the ratio between luma and chroma can make a huge difference. Yes the Varicam is lower resolution therefore softer but when you try to pull a key on a thin object you will get a more even result.

Now how did this post turn into how much better a F900 is? Radek just pointed out a new camera that gets us even closer to high quality HD. The last time I looked this was a HDV forum meaning 1/3" cameras. If these cameras can look amazing imagine how much better this 310 will look. Maybe when shooting high detail landscapes or very dark scenes the F900 will have the advantage but the 310 is getting us very close. Much closer than any other current HDV camera for about 1/4 of the price of the F900. If we are talking about the F900 recording to HDCAM tape then I feel as though the uncompressed output from the 310 would be better. If I had to choose between a 1440x1080 4:2:2 image that was slightly softer due to chip size and optics or 1440x1080 3:1:1 that had lots of detail I would always choose the 4:2:2 softer image for keying. You can add as much detail and sharpness into an image as you want but at the end of the day when dealing with keying I need chroma not luma detail. Heck I would even rather have a 4:4:4 960x1080 image from the FX1/Z1 rather than 1440x1080 4:2:2. The image would be softer but the keys sure would look damn good.

I have used a SONY DSR-500 which is a 2/3" camera with native 16x9 chips and amazing quality. At the end of the day however DVCAM is still 4:1:1 and when it comes to keying it isn't really any better than my other Canon XL1. There are more details in the image but the damn chroma blocks still look the same.

If we are talking F900 and HDCAM tape I think the 310 could be better depending on what you are shooting.

If the F900 is also using uncompressed output well then yes the F900 would be slightly better.

(Quick question here: Is the uncompressed output on the F900 the full 1920x1080 raster or 1440x1080?)

If the uncompressed output is only 1440x1080 well then pixel wise the two cameras output the same 1440x1080, 10 bit, 4:2:2. The only thing that would give the F900 the advantage is a tiny amount of extra detail due to larger chips.

If the F900 does have true 1920x1080 chips and outputs that uncompressed well then it has a bigger advantage. Although I do have to say it gets a little hard trying to tell the difference between a 1440x1080 stretched image and a 1920x1080 image.

[Jeff Sayre]

Oh boy, I think Thomas has me here.

Originally Posted by Thomas Smet

Now how did this post turn into how much better a F900 is? Radek just pointed out a new camera that gets us even closer to high quality HD. The last time I looked this was a HDV forum meaning 1/3" cameras.

You are correct, Thomas. This is a HDV forum. But that is where this thread started so I stayed on topic by discussing the camera comparisons that Radek brought up in his first posting. The HDCX310 does have a HDV output option, so from that standpoint, Radek was on the money to post it here and compare it to other cameras. It has been good, if not long-winded, discussion.

However, having said that, it sure does seem that I may have stretched this topic a little too much into the technical side of things! That is what happens late at night when the mind starts wondering.

I have not intended to imply that the F900 is the best camera out there other than the F950. In Radek's first post, he set up a hierarchy when discussing the X310.

Originally Posted by Radek Svoboda

Picture quality should be superior to 100,000 USD F900 recorded to tape, that is used for major motion picture productions. Next best thing would be Sony F950 1080p camera, costs some 120,000 USD without lens, LCD, recorder.

My point was that the X310, while a wonderful camera that has great potential, is not superior to the F900. They each have their benefits but the X310 is not a clear winner.

Originally Posted by Thomas Smet

When dealing with very small details such as hairs and other thin objects the ratio between luma and chroma can make a huge difference. Yes the Varicam is lower resolution therefore softer but when you try to pull a key on a thin object you will get a more even result...You can add as much detail and sharpness into an image as you want but at the end of the day when dealing with keying I need chroma not luma detail.

Thomas brings up an excellent point that helps reinforce the fact that, no matter how wonderful a camera is from a technical standpoint (chip size, A/D bit sampling depth, output sampling rates), the true determinant of the value a given camera brings to a project depends on the requirements of that project. So one person's great camera may be another person's sort of okay camera.

So, if each of us can afford a $25k to $100k camera, then what are we arguing about anyways :)


(Quick question here: Is the uncompressed output on the F900 the full 1920x1080 raster or 1440x1080?)

I'm not clear on this myself. Whereas I know people have claimed that in tests of the F900 they could output 1440x1080 8 bit compressed 3:1:1 to tape (HDCAM) and uncompressed 1920x1080 10 bit 4:2:2 to disk, I do believe that the F900 actually samples the entire 1920 x 1080 pixel array but then filters it down to 1440 x 1080.

Here's another link that may help to begin to answer that question: http://hd24.com/panasonic_v_sony.htm

[Radek Svoboda]

As long picture quality is similar to F900 camera that cost 150K USD with lens, etc., and you pay 1/5 cost (30K USD) and picture is 10 bit uncompressed 1080p, it is revolutionary. I say is better than 150K camera, you Jeff say is worse, but we both agree is quite close to F900.

1/2" 1080p HDV camera, equipped for production, for 18.5K USD is revolutionary too.

Radek

[Jeff Sayre]

Originally Posted by Radek Svoboda

I say is better than 150K camera, you Jeff say is worse, but we both agree is quite close to F900. 1/2"

1080p HDV camera, equipped for production, for 18.5K USD is revolutionary too.

I do agree with that! And, I appreciate you bringing this up for discussion, Radek. It's good to debate ideas and products in a professional and productive way :)

[Radek Svoboda]

If Sony would make this into 1/2" 1080p 10K USD camcorder, it be riot.

HDV now extends from 1,700 USD to 27K USD for upcoming 2/3" JVC HDV camera. 1/2" 1080p camcorder for 10K USD would be nice.

Radek