FX1/Z1 Image Quality Same as $100K Panasonic Varicam and Sony HDCAM, well suited for HDTV production

HDV wrote on 12/26/2004, 5:34 AM
Varicam costs $100K, configured with lens, viewfinder, etc. It is compressed 6.7:1; it's effective stream at 24p is 40 Mbps, frame size is 1.6 Mb. Recorded resolution is 960x720 pixels, off of aligned 1280x720 pixel CCD's. 1080/60i DVCPRO HD is equally compressed and records 1280x1080 pixels.

Enter new generation of a Sony camera. It costs $4K. The light sensitivity is couple stops less than Varicam but equal to the now discontinued DVCPRO HD camera that cost the same as Varicam. Compression varies from 4.7:1 to 15.7:1, depending on the amount of movement in the picture. The more movement, the less is the eye ability to resolve, so we don't see this degradation. The GOP is 12/15 frames (50i/60i). The I-frame size is 150-500 kB, again depending on picture complexity and amount of movement. That translates to 1.2-4 Mb. The CCD chipset has horizontal green channel offset, which creates effective resolution of the camera head / recorder 1440x1080 pixels. The CF mode is using the MPEG2 encoder to do deinterlacing, creating the most effective deinterlacer there is. While excellent deinterlacers drop vertical resolution in half at the moving areas of images, the Sony MPEG2 encoder based deinterlacer is capable of tracking motion so it's vertical resolution drop is a lot less on movement.

The lens is excellent; still not in the same league as the HD lenses used on Varicam; the lenses are made to resolve well even on Sony F900. These lower resolving HDV lenses lower the system resolution and make the images overall equal to DVCPRO-HD in interlaced mode and to Varicam, when the Sony HDV is in CF25 mode. Forget the CF24 mode.

The uncompressed analog component output, when combined with HDSDI adapter provides cleaner output, but then again other factors come in, like the lens resolution, etc. The quality of this uncompresed output is overall on the same level as HDCAM recorded output. One can use up to 200' coax to feed the camera to a MAC or to PC to record the picture. When recorded to PC with Aspect HD or Prospect HD, it can be compressed real time with a codec that is superior to HDCAM. HDCAM is compressed 4.4:1 and it records 1440x1080 pixels.

Overall the new Sony HDV cameras are very well suited for film out, via CF25 mode, or to record standard interlaced HDTV content. HDTV worldwide is some 90% 1080i; the recommended European standard was to be 720p but after Sony's lobying it is likely to change to 1080i. Sony is pushing 1080i as a stepping stone to future 1080p broadcast, with the same bandwidth as the presnt 1080i broadcast, but with 2x as efficient codec, on the MPEG4 level.

The new HDV camera image quality is superior to SD formats uprezzed to HD, and that includes Digital Beta, XDCAM, SDX-900. DV cameras like Panasonic DVX-100a and Canon XL2, when their PAL progressed images are uprezzed to HD, produce highly inferior image quality to the Sony HDV, even if the content was shot with PAL versions of the cameras. The CF25 mode of the Sony HDV cameras provides highly superior results, and the Zeiss lens is better.

Comments

farss wrote on 12/26/2004, 6:14 AM
I'd agree with everything EXCEPT your statement that with more movement the less the eye can resolve. You clearly, like many others do not understand mpeg-2 compression. The issue has nothing to do with movement in the way film people consider movement. The critical issue is the total number of pixels whose values change between frames.
This means a locked down shot of a car whizzing by at 300 mph does NOT constitute a lot of movement. However a locked down shot of a scene with a lot of detail but bad heat haze does have a lot of movement. In this scenario the entire frame can be affected, simply because the encoder just doesn't have the bandwidth to cope. I can assure you it's VERY visible, and that's not the only shot I've seen where the HDV systems cannot cope. Pretty well any other system from lowly DV25 up would in those circumstances have returned superior footage.
You also mention the lens. Certainly it's a fine piece of glass however out wide it does have noticeable barrel distortion on the edges.
This is an excellent camera, probably Sony's finest effort in a long time and yes it will have a major impact. The last thing it needs though is outlandish claims made about it, ones that even Sony themeselves are staying well clear of. If it really is such a wonderful solution for everyone shooting anything I'd imagine by now we'd be seeing a big drop in the prices for PD170s etc.
What isn't mentioned and people should take on board more: It's a damn fine DV25 camera in 16:9! You could throw all the HDV bits away and still have an excellent value for money camera, the Z1 even more so.

Bob.
winrockpost wrote on 12/26/2004, 6:24 AM
.............FX1/Z1 Image Quality Same as $100K Panasonic Varicam and Sony HDCAM


OK I havent had the opportunity to play with the new hdv cams, hell I dont yet own a hdtv,, but you are saying a 4k cam is going to have as good a pic as a100k cam with a 20 or 30 k (guessing) lens.
Call me a sceptic,
Cant quite believe sony would want that, even if they could achieve it.
Nat wrote on 12/26/2004, 9:31 AM
Well, I could take my dad's 8 megapixel consumer photo camera and my prosumer 6.1 megapixel DSLR and do some ab shooting and find out the consumer Cam does a great job. However, it has a fixed lens... This is the big difference, you can't change your lenses, when you pay 100k for a camera I guess you intent to change lenses once in a while...
Spot|DSE wrote on 12/26/2004, 10:59 AM
While I'm certainly an HDV proponent, that thread is fraught with errors, and unrealistic.
We compared images from the Viper, HDCam, Cinealta, Betacam, and other formats when writing the HDV book. Discounting lenses, HDV came closest to resolving with Varicam, but there are lots of other variables. A 5K cam isn't gonna knock these other higher end devices off their pedestal. (no pun intended)
For example, to resove an HDV cam to the value of a Viper, it would take 50+ HDVcams shooting to one output.
I've seen the HDV Cam using component out, HD/SDI. It looks stunning. And at that level, it somewhat competes with the higher end devices, discounting lens issues. But once it's on tape, it's compressed. You're making a moot argument. Some made this same argument about why DV was better than Digibeta, because they were using uncompressed component outs from the JVC 500 cam. Uncompressed will always look better than compressed. So it's not a fair comparison, nor an honest one.
ushere wrote on 12/26/2004, 2:07 PM
happy festivities and all that.....

now, without wanting to cause too much consternation....

i find these endless debates about how the quality of $5k cameras match that of $50k somewhat pityful. when dv came in i was shooting with an sp400, and people were running around with 1000's screaming the deathj of sp, amaxzing pics, and that's the end of betacam....

i bough a 1000, and regularly shot with it along side my 400. yes, it was a significant adavance on what had come before, but sitting inpost you saw the difference quite clearly.... i am now semi retired, and shoot with a 170 (having sold my 400 for a pittance). i am extremely happy with it, but, i still do post work for some of my old clients, who are shooting with digibeta, and there's NO WAY you can compare picture quality between a prosumer and pro camera - it's just a loke. the lens alone.....

as for the new batch of hdv - well, i've seen some footage from fx1 (i think, though it might have been the other) and was very impressed, but it isn't digibeta, and unless you've looked at the results side by side on a proper monitor, all this talk of equal too is simply wishful thinking....

do people really believe sony, etc., are going to produce consumer cameras that will kill off their broadcast depts? i think not, and i find the vast majority of proponents of the wonders of prosumer cameras have never shot with a pro camera, let alone seen side by side live footage on a pro monitor.

i believe the dv / hdv revolution has brought wonderful tools to the disadvantaged (financial / creative), but there is absolutely no correlation between domestic picture quality and that of broadcast.

that said, my old saw remains true:

it really doesn't matter what it's shot on, if the audience wants to watch it, they will....

leslie
farss wrote on 12/26/2004, 2:55 PM
Well down here SP is well and truly dead, DV25 replaced a lot of it but as you say DB is the format and camera systems that you should compare it against.
We are getting very good results going from the FX1 through the AJA converter to a DVW 250. This is a practical way to shoot BTW and means we're getting good mileage out of our 250, yes that VCR will even run on batteries.
But as those in the know keep saying, it ALL starts with the glass.
Bear in mind the FX1/Z1 uses a trick with the CCDs, pixel shifting. Although it technically delivers 1440 res that's only on the luminance and probably quite a valid trick when your target is 4:2:0. The chroma res out of the block and as recorded to tape is much lower.
And I'll say again. If you are dead set serious about shooting HiDef there's a large number of factors that affect production costs, the camera/post costs are not the biggest ones. We've already got a client shooting their first production in HDV and immediately watching the camera tapes on a HiDef monitor you can see these issues comes to light. This isn't a HDV issue, it's the same if you shoot Varicam or CineAlta. You need to start shooting like Hollywood does it, lighting is much more important naturally but focus is critical, if your talent needs makeup then, what you can fudge in SD will not work in HD.
Bob.
HDV wrote on 12/27/2004, 9:49 AM
I've tested the little Sony camera. It is very close to DVCPRO-HD / Varicam's image quality. And you've right, the DP is what matters the most, his ability to handle an HD format.
Coursedesign wrote on 12/27/2004, 10:20 AM
There is more to picture quality than lines of resolution.

The more expensive cameras have a much wider dynamic range, meaning highlights won't blow out as easily as on the current HDV cameras.

The expensive HD lenses have better color rendition and contrast rendition (the ability to resolve similar shades of any color without turning them into a mush).

The CCDs are more sensitive and larger so you can reduce DOF much more easily.

The DSPs are much more powerful, which helps with a whole bunch of real life video problems that most HDV-shooters have never even heard of, such as Low Key Saturation and Knee Saturation (raising color saturation in the shadows and highlights so the image doesn't look washed out there), Adaptive Highlight Control (multiple knees to help exposure control), and plenty more.

The high end HD cameras are very difficult to focus (because of the reduced DOF and minimal circle of confusion), the best way is with a $18,000 SceneFinder, a native HD color viewfinder with extraordinary technology that can focus even on a white wall...

That's not a joke, I have done it.

HDV cameras will just get better and better, but let's not kid ourselves that they are equal with "the good stuff" today.
HDV wrote on 12/27/2004, 5:09 PM
This new inexpensive Sony does not only have similar image quality as the 20x more expensive Panasonic; the low light sensitivity and the highlights handling is similar. It's the Sony 1/3" CCD's against Matsushita's 2/3" CCD's. It's Sony expertise against Matsushita expertise.

I've tested the camera to confirm this and one of the earler posts has a link to a similar test.
farss wrote on 12/31/2004, 2:56 PM
Again you are wrong. Low light performance is only a very small factor in CCD performance, in fact for anyone shooting a serious production it's totally meaningless, they light to suit.
What CCD size does affect, is DOF, that's just plain physics. It also affects latitude and smear. Even keeping the argument within the range of Sony cameras that's why their consummer stuff may have 1/4" 1M pixel CCDs and their broadcast cameras 1/2" or 2/3" 480K pixel CCDs.
Using your flawed logic the Sony PDX10 should be as good as the Sony 570 or 530 yet I can tell you the quality difference in the image is staggering.

And here's another example for you to consider, let's just keep it ALL in the Sony camp, and yes I've seen this myself, on a BIG screen and on a 2K res projector worth more than I'd care to think about.

Out of the camera the FX1 actually looks better than the out of the camera CineAlta footage, staggering isn't it! But then you start to look very closely and you realise that the CineAlta is giving more accurate color rendition, there's more depth to it. Well so there should be, it's sampling at 4:2:2 against 4:2:0 and it's recording much higher res. But wait, the HDV camera is WAY less noisy! Yipes, you could have bought a house with the money you would have saved if you bought the FX1.
Well no, you see Sony have a whole SYSTEM to process the CineAlta footage, they can get rid of the noise, they can grade the images better because they have more data to start with.

Now lets take another comparison, the FX1 up against the Viper, now there's a real camera. Yes I finally got to see Russian Ark! What a superhuman effort, I didn't realise the whole thing was shot off steadycam, two man crew, one with the camera and one with a backpack full of disks, wow!

Now to the untrained eye at SD the images aren't as stunning as the stuff from the FX1! But look at WHAT they're shooting and how they're doing it, they're using mostly low light and still getting a huge color depth and doing it with zero setup time.

So what I'm saying is this, with any camera firstly you need to see it as part of a system, a system that ultimately delivers images for humans to watch. Simply quoting number of pixels or bitrates or looking at single frame grabs tells us very little about the system and how it'll look to the viewer. And at the end of the day pros shoot with what they're comfortable with, if they have to go outside that comfort zone they spend a lot of time testing and checking. They run tests through the whole system to see how it's going to look in the delivery format. Many times I get amazed by some of the decisions DOPs make until I realise they're looking at a system and one driven by human factors. I heard of one who rejected using a DI because it gave them too many choices in post and that slowed down productiom. He wanted to make certain they couldn't change their minds in post, how they decided to shoot on the day was how it was going to look.

In the few years that I've been back in this game I've had to eat a lot of humble pie, I started out with a certain arrogance thinking my ability to manipulate a lot of numbers would hold me in good stead but slowly I've learnt that the numbers game only a small part of the story. We nearly invested heavily in the original HDV camera but decided against it. The numbers looked great, heaps of resolution and damn cheap too. What decided us againt it (thank god!), my SUBJECTIVE feel about the images we saw at IBC. To this day I don't know of any set of numbers that explains why it looked wrong but even though it was 720p up against the PD170 or the DVX100A it felt wrong. Now history has pretty much borne out my gut feeling, sure there are those who've managed to pull great stuff out of that camera, with a lot of work and with a LOT of understanding of its limitations.
Bob.