(Page 3 of 3 pages for this article < 1 2 3)
Wednesday, November 25, 2009
Preview: Sony PMW-350 2/3” 3-CMOS XDCAM EX Camcorder
Adam Wilt | 11/25
XDCAM EX bulks up and goes shoulder-mount… for a price.
Pre-Production Prototype Performance
Running the PMW-350 through the wringer at Meets The Eye’s Studio B.
Because the camcorder I tested is a pre-production prototype, take the following information only as a preview or prediction of the final product’s performance. Both the good things I saw, as well as the problems, may be changed—for better or for worse—by the time the PMW-350 ships. If previous Sony preview prototypes are an indication, this 350 was probably about 90% representative of the final product… but it may not be. You have been warned.
Optics
The Fujinon VCL-816BYS 16x lens supplied with the 350K runs from 8mm to 128mm; in 35mm still-camera terms that’s 31.5mm - 503mm. That’s ever so slightly more telephoto on both ends than the lenses on the EX1 and EX3, which equate to 31.4mm - 439mm. Side-by-side viewing with an EX1 confirmed the basic equivalence of the two camera and lens setups in terms of viewing angles; the PMW-350 was just a bit tighter than the EX1 at both ends of the scale, but the differences were not so great as to make me choose one camera over the other.
The PMW-350 offers ALAC—Automated Lens Aberration Compensation—Sony’s version of chromatic aberration correction. The prototype lens didn’t have correction tables in its firmware yet, so ALAC wasn’t functional with it, thus some color fringing was visible in the wider half of the zoom range (you’ll see it in the corners of the test charts below). Even so, the aberration was no worse than average for low-cost zooms; the shipping lens with ALAC working should look at least as good as the lenses on the EX1 and EX3 in this regard, which is pretty darned good.
Distortion on the Fujinon follows the same pattern as on the similar 1/2” lenses on the EX1 and EX3 with some mustache distortion (barrel in the center, pincushion at the corners) present at wide angle, pincushioning in midrange, and minimal distortion thereafter. The good news is that distortion is milder throughout than on the 1/2” lenses; it’s visible, but never distracting (unless you’re shooting prominent horizontal lines, as in my test image).
Mustache at 8mm, pincushion at 15mm, distortion-free from 30mm on.
I didn’t note down specifics regarding portholing, vignetting, or exposure ramping; as I recall, the lens was fairly typical in this regard, with some light loss at the edges at wide apertures at telephoto. It certainly didn’t call attention to itself during testing, which is a good sign.
The zoom rocker offered fine-grained, smooth control, with a maximum speed end-to end of about two seconds, and a minimum speed low enough that it was difficult to differentiate between a slow zoom and no zoom. Smoothly-modulated zooms are easy to execute.
The assignable buttons on the camera, if assigned to zoom the lens, provide a single, programmable speed, with hard starts and stops.
Full manual focus with the focus ring snapped back allows focusing from about 31 inches to infinity. With the focus ring forward, in AF/MF servo mode, macro mode is available, allowing focusing down to about two inches from the front of the lens.
Autofocus worked well, with a minimum of hunting, though it can be a bit slow to lock in at the slower frame rates. Leaving macro mode engaged can slow focusing down a bit, as the lens has more of a range to explore.
Overall, the lens performed quite well for its price. Once its firmware is updated for chromatic aberration correction, I would expect most PMW-350s to ship with it, as it looks like it will provide very good value for the money.
Resolution
The PMW-350 offers true 1920x1080 sensors and backs ‘em up with true 1920x1080 recording. It also provides 720p and SD recording modes, which downsample their images from the full sensor. I recorded samples of each frame size using the HQ 1920, HQ 1280, and DVCAM formats, with edge enhancement off, and pulled frames from them, reproduced below in pixel-for-pixel sizes. Note that the 1080-line images have been tweaked by moving the lower, horizontal frequency sweep up in the image to save some space, and color / contrast has been roughly normalized, but otherwise these are as they came out of the camera.
1080i XDCAM EX frame grab, 1:1 (horizontal sweep moved up for space).
1080p XDCAM EX frame grab, 1:1 (horizontal sweep moved up for space).
To my eye, the 1080i performance is extremely good, and the 1080p only slightly less so, with a bit more vertical aliasing than I like—but when it comes to sharpness vs. smoothness preferences, I lean towards smooth more than many folks do, so the extra “snap” of the 1080p images may count for more than the extra aliasing as far as you’re concerned.
720p XDCAM EX frame grab, 1:1.
The downconverted 720p image looks very clean indeed.
480i DVCAM frame grab, 1:1.
480p DVCAM frame grab, 1:1.
Standard-def 480i has a surprising amount of vertical aliasing around 800 lines (twice the desired cutoff of 400 lines for a 480i image), but it’s of a fairly low amplitude and isn’t a serious problem in real-world shooting. Aside from that, the downconversion looks pretty good.
Sensitivity and Noise
I aimed the PMW-350 and an EX1 at a DSC Labs Chroma DuMonde chart, and set their exposures so that the midgray on the chart hit 50% on the waveform monitor. I then set the same shutter speed on a Spectra Pro IV light meter and varied the exposure index (a.k.a. ISO setting) until the meter read off the same iris setting. Both cameras were set up using standard, Rec.709 gamma settings, and the chart was illuminated with Kinoflo VistaBeam 600s with daylight tubes.
In 1080p, the EX1 came in at EI 500, while the 350 registered EI 800, 2/3 stop faster. In 1080i, the EX1 was EI 1000, while the 350 was EI 1600 (as expected: in interlace, dual-row readout is used, which sums two pixel together to reduce interfield twitter, so the effective sensitivity os twice as high). These same values held in 480p and 480i modes; the downconverted images are being created from 1080i and 1080p captures.
In 720p, the EX1 showed EI 640 while the 350 showed EI 1000; 720P modes on both cameras are about 1/3 stop faster than 1080p modes (but still 2/3 stop slower than 1080i, of course).
Observant readers will protest that I previously rated the EX1 a third of a stop slower than I do here. Those ratings involved setting an 18% gray card to 50% on the WFM, but the gray card is darker than the midgray on the DSC chart, and really should have been set to about 42% on the ‘scopes. Also, I should have leveled the exposure to set the midgray on the DSC chart in these tests to 58%, not 50%, according to DSC’s calibration information. Those factors combined should account for the difference; I apologize for the confusion.
Whatever the EI numbers actually are, the 350 is about 2/3 stop faster than an EX1 or EX3.
Noise on the PMW-350 looked to be comparable to noise on the EX1 at the same gain settings. As gain was increased, the 350 seemed to hold its resolution a bit better then the EX1 did; it looked a bit crisper at +18dB. The 350 can go all the way to +42dB, which made for a bright picture in near-total darkness, although that picture was quite noisy in both luma and chroma.
Going the other way, gain on the 350 may be set to -3dB for slightly better noise performance.
Latitude, Gammas and Knee
I didn’t do a black-to-white latitude measurement for lack of time, but I did compare the 350 to the EX1 by measuring how many stops I had to open the iris before a 50% midgray on the chart clipped at 109% on the WFM.
In standard, Rec.709 gamma, with knee set to capture as much as possible, the EX1’s midgrays clipped at 4-4.3 stops overexposure. The 350 clipped at 4.7 stops overexposure, for 1/3-2/3 stop of additional highlight handling compared to the EX1. As I previously measured the EX1 as having ten stops of latitude, that would put the PMW-350 at about ten and a half stops. Not too shabby!
The PMW-350 offers five standard gammas:
- DVW, to replicate the look of Digital Betacam camcorders.
- x4.5
- x3.5
- 240M, SMPTE 240M equivalent.
- R709, the Rec.709 setting (default).
- x5.0
These show minor variations in midgrays and black stretch/compression.
The 350 also has “HyperGammas” comparable to those on other high-end XDCAM HD and HDCAM camcorders, instead of the slightly different CineGammas of the EX1 and EX3:
- 1: 3250, which puts a 325% input signal at 100% output level.
- 2: 4600, which puts a 460% input signal at 100% output level.
- 3: 3259, which puts a 325% input signal at 109% output level.
- 4: 4609, which puts a 460% input signal at 109% output level.
(See my review of the HDW-650 for more details on these gamma settings.)
While the HyperGammas should offer the same or better latitude compared to the standard gammas, I only saw 3-3.3 stops of overexposure before grays clipped. Sony tells me that the HyperGammas weren’t yet dialed on in this prototype.
The PMW-350 does not offer user-definable gammas nor the S-Log curve found on Sony’s top-end HD cameras.
The standard gammas offer a highly-adjustable knee, with tweaks for knee point, slope, and saturation, as well as a separate white-clip level setting. Unfortunately, this prototype camera inherits the saturated-highlight knee cutout problem of the EX1 and EX3: if a bright area has a color, the knee “gives up” after its initial compression, and the bright area crashes hard into clipping, with the problem roughly proportional to the saturation of the highlight.
Highlight handling: color-balanced std gamma & knee; color-unbalanced std gamma & knee; color-unbalanced HyperGamma.
HyperGammas are immune to this problem since they don’t use a separate knee.
Sony may fix this before the PMW-350 is released. I hope so, because it makes use of the standard gammas much more problematic. While I use Cine gammas on the EX1 exclusively, and HyperGammas on cameras that have ‘em, many folks prefer the standard gammas for their more saturated highlights on skin tones. Having a dodgy knee circuit on a $6400 EX1 or an $8300 EX3 is one thing; you get so much performance in those packages that you’re willing to put up with a problem or two. In a $20,000+ camcorder, though, that knee circuit had better work, or those who depend on a properly working knee will stay away in droves (yes, I’m talking primarily about news shooters, and they’re the biggest single market segment for a camera like this). But again, I repeat: this is a prototype camera, and the shipping version may fix this problem. Stay tuned…
Etc.
The camera uses CMOS sensors, so it’s immune from vertical smear. However, it is subject to “rolling shutter”, which looks to be comparable to the same artifact on the EX1. Super-fast panning or frenetic shakycam will reveal it, but it’s not objectionable in everyday work.
Flicking on the power switch brightens the EVF within a couple of seconds, and the camera is ready to shoot in three seconds.
Sony claims about six hours of runtime on a 95 Watt-hour BP-GL95 battery; I saw 3-4 hours on the prototype with a 65Wh BP-GL65 battery, with a lot of zoom and focus servo action.
I didn’t test the different codec settings. DVCAM is well understood, and aside from the HQ 1440 mode, the XDCAM EX codecs are the same as in the EX1 and EX3. The new HQ 1440 mode (essentially the 35 Mbit/sec mode used on XDCAM HD camcorders) trades off 1/4 of the spatial resolution for a milder compression ratio; based on my experiences with the EX1, that tradeoff may be worth making in interlaced recordings, where the differences between fields makes the full-res codec throw off a few more artifacts than I’d like, but I’d stay with HQ 1920 for progressive material.
The EX1 and EX3 have an extended red response; under hot lights (those with a lot of red and IR energy, like tungsten lamps emit) they tend to see some black fabrics as shades of blue, purple, or brown, depending on the amount of far-red and near-infrared light they reflect. The PMW-350 reins in this tendency considerably, though it’s not entirely neutral:
Far red / IR sensitivity of three cameras.
Clearly, Art Adams is going to have to do a whole new set of filter tests… but for the rest of us, this means that we can shoot the 350 under tungsten lighting without wondering where those oddball colors came from quite so much as with the smaller 1/2” EX cams.
Conclusions
Really, it’s too early to draw conclusions: I only saw an unfinished, prototype camera. Still, we can discuss the camera in the abstract; it’s a grown-up EX1/EX3: CMOS sensors, great dynamic range, flexible images sizes and frame rates, interchangeable lenses, autofocus ability, SxS recording at 35 Mbit/sec, and an industry-standard shoulder-mount design. It’ll cost $20,500 without lens, or $22,000 with lens; that’s a lot more than the sub-$10,000 EX1 and EX3. Is it worth it?
For those who pick up a camera on a daily basis, often without a tripod, the answer is simple: the shoulder-mount design is head-and-shoulders above the lopsided EX1 Handycam or the EX3 chainsaw. There’s no contest; neither the EX1 nor the EX3 has even been in the running, ergonomically speaking, for those who spend extended hours with a handheld camera: news, nature, and documentary shooters especially. These folks aren’t deciding between cheaper half-inch EX Handycams and this 2/3” EX shouldercam, they’re trading off SxS cards against XDCAM optical disks, P2 cards, and DVCPROHD or HDCAM tapes; between CMOS and CCD sensors; and so on, all within the realm of 2/3” shouldercams. In this competition, the PMW-350 is very competitively priced.
For those who can work with the smaller, handheld form factor—because they only handhold intermittently, or for short intervals—it’s a more interesting question. Yes, the PMW-350 gives you larger sensors and shoulder-mount ergonomics, plus some improvement in dynamic range, sensitivity, and ergonomics—but these are incremental performance upgrades over the EX1 and EX3. What justifies the steep price jump going to the PMW-350? This question has a couple of facets.
From the supplier’s side, building a 2/3” shoulder-mount camcorder is a lot more expensive than building a 1/3” or 1/2” handycam, and that cost difference, combined with a smaller unit volume, makes the bigger camera disproportionately more expensive.
We see some of this at work with the HVR-Z7 and HVR-S270 camcorders, which share the same 1/3” imagers, yet the shoulder-mount costs substantially more than the Handycam: 50% more just for the privilege of being able the balance the S270 on your shoulder.
Panasonic’s Jan Crittenden, discussing the HPX-300 shoulder-mount camcorder, says that it uses 1/3” sensors to keep its cost down to $8,000; making a 2/3”, full-res sensor is hugely more expensive and would make the HPX-300 unaffordable. Panasonic’s own 2/3” HPX-500 uses a lower-resolution sensor, yet it still costs $14,000 despite both camcorders being shoulder-mount designs (and the 300 has AVC-Intra compression, to boot). That’s nearly twice the price to double the size of the sensor, even while dropping its photosite count.
Based on these two examples, the price jump from the EX1 or EX3 to the 350 isn’t that much out of line.
From the buyer’s side, there’s the impact a larger, more professional-looking camera has on a client’s perception (sorry, babe: size really does matter); there’s the comfortable ergonomics of the shoulder-mount design; there’s the immediate entree into the world of widely-available 2/3” lenses and accessories. 2/3” gets you slightly shallower depths of field than 1/2” (and you can put Zeiss DigiPrimes on the 350 if you need to get your indie-cine mojo workin’), and you reap the benefits of greater sensitivity and wider latitude.
Back in the DV days, a DSR-PD150 1/3” Handycam cost a fraction of what the shoulder-mount, 1/3” DSR-250 did, and the DSR-250 was cheap compared to the 1/2”, interchangeable-lens DSR-300. Sony sold bunches of all three.
Clearly, there will be a market for the PMW-350, even at 2x to 3x the price of the smaller cams.
But in the end, that’s a lot of abstract hand-waving. The PMW-350 is what it is (or, more accurately: it will be what it will be). If you’ve made it this far, you’re probably having one of the following reactions, all equally valid:
- Finally! An EX camcorder for grown-ups! About flippin’ time!
- This is a very interesting camcorder that competes well against other shoulder-mount cams. I’m interested to see what CMOS sensors, SxS recording, fully variable frame rates, HyperGammas, and an autofocus lens can add to my capabilities.
- Looks sweet, but I can live with my EX1 and/or EX3; the jump in cost hurts more than 350’s incremental benefits help.
- A 2/3” camcorder with an autofocus lens? 35 Mbit HDV-on-steroids recording? A color LCD finder? Did it come in a Crackerjack box? What were they thinking? I’ll use one only after they pry my HDW-F900 / HPX3700 / Varicam / Digital Betacam / BVW400 / Arri 535 out of my cold, dead fingers!
OK, maybe the last one is a bit extreme… nonetheless this is the most serious upward migration yet of features found in sub-$10,000 camcorders into the fully professional, no-nonsense realm.
I for one will be very interested in how the PMW-350 sells. Since I first started working with P2 cards on the Panasonic HVX200, I have been firmly in the solid-state camp, and the 350 is the first Sony shoulder-mount with solid-state recording. I’m also a cranky bastard about resolution, and this is the most affordable Sony shoulder-mount that captures and records all the pixels I think I’m entitled to in a “full-HD” camcorder.
Would I buy one? At Meets The Eye we own an EX1 (along with a Canon Vixia HF200 and three RED ONEs), and I’m in love with images the EX1 lets me capture, though I’m less enamored with the EX1’s challenging and sometimes painful ergonomics. The 350 would give us a flexible, very high quality 2/3” platform with the ability to genlock and to mount most 2/3” lenses, both of which are capabilities we value. If Sony fixes the knee in the standard gammas, the 350 might well be on our short list. But without a knee fix, I’d find it a harder sell, as the 350 would be put into service as a rental camera for studio operations, and I expect many of our clients would want to use standard gammas to better intercut with their other cameras. I’d also want to see the HyperGammas performing at least as well at the EX1’s Cine gammas, preferably better. We’ll just have to wait and see…
Pros
- Full-resolution sensors and full-resolution recording.
- 10.5 stops of dynamic range.
- Variable frame rates 1-30fps (1080p) and 1-60fps (720p).
- Progressive and interlaced recording, 60Hz and 50Hz formats.
- High-quality XDCAM EX 35Mbit/sec recording on SxS cards.
- Fully-professional shoulder-mount configuration with excellent, shoot-all-day-handheld ergonomics.
- Superb, huge, color EVF with real, variable peaking control.
- Interval, single-frame, and cache (pre-rec) recording.
- Autofocus-capable yet very affordable 16x lens.
- Works with wide-ranging ecosystem of 2/3” lenses, wireless receivers, tripod plates, remote controls, etc.
- BNC, full-size i.Link, full-size HDMI, and XLR connectors—no weird, proprietary plugs.
- Genlockable; remotely controllable; remotely paintable.
- No vertical smear.
- HyperGammas.
- Very tweakable image looks and operating characteristics.
- Lots of assignable buttons and switches.
- Some of the best EVF data displays around, including histogram, focal distance, and depth of field.
Cons
- A lot more expensive than the 1/2” EX1 and EX3.
- On-board recording is only 8-bit, 4:2:0 sampling; can show compression artifacts when stressed.
- SxS slots on right side of camera are away from the operator, and are more subject to unauthorized access by passers-by than card slots on the operator’s side of the camera.
- Stock lens has noticeable distortion up through 30mm.
- Rolling-shutter “jellocam” artifacts possible.
- DVCAM mode lacks any 24p option.
- No analog component or Y/C outputs.
Cautions
- Everything in this preview is subject to change.
- Prototype’s HyperGammas don’t appear to be working properly.
- Prototype’s knee suffers from EX-series saturation-induced blowout.
- DVCAM mode may not be standard.
- Prototype’s chromatic aberration correction not working, so it’s not possible to say how good the lens will look.
- HQ 1440 compression not tested.
- Lens ramping / vignetting not tested.
- Downconversion of E-E signal not tested.
- Recording media (SxS cards) not included.
- Tripod adapter plate not included.
- AC adapter, charger, and battery not included.
- Everything in this preview is subject to change.
16 CFR Part 255 Disclosure
This camera was previewed because Sony requested it, and I agreed to do so. Sony shipped the camera to and from me at their own expense, and provided pre-release PDFs of the operating manual, as well as certain background documentation not normally released outside of Sony. All hardware, software, and printed documentation sent to me for the review has been returned to Sony; electronic documentation provided to me will be erased once this article is published, though I’ll retain the PDF of the preliminary operator’s manual for future reference.
No other material connection exists between myself and Sony: Sony provides no compensation to me for reviewing equipment and has not influenced me with payments, discounts, or other blandishments to encourage a favorable review.
(Page 3 of 3 pages for this article < 1 2 3)
You must be registered to comment. This is an effort to reduce spam. Please REGISTER HERE.
Wow, thanks for this preliminary impression!
However, I was half in the bag until I read the part about still shooting 8bit 4:2:0 color space.
I’ve been ready and waiting to make the jump from prosumer to full-on pro with a 2/3” shoulder mount. But I’m having a real hard time trying to get past the negative of not having 10bit 4:2:2, like the HPX300 has, for example.
In my personal opinion 10bit is as much a pro feature differentiation as anything else - if not more so. I understand that you get 10bit out on the HD-SDI but that means an external recording device etc which jumps the up cost.
So, the HPX300 is shoulder-mount for that perception of professionalism, has 10bit, has internal rolling shutter compensation, DRS, scopes and is about half the cost?
I can get two for the price of this!
Posted by wsmith on 11/26 at 12:33 PM
Just to muddy the waters further: I shot a feature last year on the 8-bit, 4:2:0 PMW-EX1, and it looks flippin’ gorgeous. True, it can’t be pushed around in post nearly as much as 10-bit material, but 8-bit isn’t the end of the world.
Set against the HPX300 (which I am eager to get my hot, sweaty hands on for a proper test), the 350 has 2/3” sensors for shallower depth of field (the HPX300 is a 1/3” camera); fully variable frame rates in both 720p and 1080p modes; and hypergammas, which I can’t say enough good things about.
But yes: it’s between 2x and 3x the cost of an HPX300; it’s bigger and heavier; the HPX300’s AVC-Intra format allows finer-edged keying and heavier manipulation in post.
They are very different beasts and depending on what you are looking for, either one is a clear winner compared to the other, even including the cost differential. It’ll be interesting to see how they fare side-by-side in the marketplace—for that matter, I’m curious about how the PMW-350 will do against the EX1/EX3, which offer essentially the same image quality and functionality (albeit in rather different, 1/2” packages) at a fraction of the cost. In this sort of apples-to-apples marketplace battle, we’ll really see how much of a premium 2/3” sensors and shoulder-mounted ergonomics are worth, eh?
Posted by Adam Wilt on 11/26 at 12:53 PM
I agree fully with everything you’re saying, Art. Now that Thanksgiving is over.
Is it just me or are you also surprised that Sony didn’t make this a 10bit camera? How can it ever really be considered a pro camera by professionals? Do you have a guess about how much that would have increased the price to?
Meanwhile where is Canon with a rejoinder to Sony and Panasonics latest moves? I’m told by some industry retail seers not to expect that Canon will ever introduce a shoulder mount pro cam because that’s not their market.
Still, a prosumer offering with a large single imager ala their DSLRs, along with their optics , and hopefully 10bit would be welcome now - if only in an annoucement.
I just don’t think I’m going to feel comfortable making any purchasing decisions on a couple of pro cameras until I see what Canon has up its sleeve.
Posted by wsmith on 11/27 at 10:44 AM
sorry for the non sequitur re Thanksgiving there. Was intending to say that now that it’s over I’m going to read all the stuff about Hypergamma you were talking about.
Still can’t make up for 8bit tho.
Posted by wsmith on 11/27 at 10:48 AM
Is it just me or are you also surprised that Sony didn’t make this a 10bit camera?
I’m not surprised at all. For most of the type of work that will be done with cameras such as this 10-bit is not needed.
Posted by Simon Wyndham on 11/27 at 05:39 PM
A $20K shoulder-cam without 4:2:2 @ 50 Mbit/s? Ok, shallower DOF than on the EX and the “size matters” form factor. For 3x the price? Seriously?
Posted by .(JavaScript must be enabled to view this email address) on 11/27 at 05:59 PM
Burn-E, this stays pretty much in line with the product hierarchy as before. If the 350 had 4:2:2 and 50Mbit/sec recording, why would anyone purchase a PDW-700? Already the 350 is slightly more sensitive than the 700, and the 350’s slow motion functionality is higher resolution than the PDW-800’s, so apart from CCD’s they would cannibalise their own market.
You can record higher bitrate pictures with the 350 using a NanoFlash. $20k for this camera is reasonable.
Posted by Simon Wyndham on 11/28 at 04:07 AM
Thanks for this comprehensive review Adam. One advantage you left off was weight. After years of lugging around the BVW550 and DSR500 we own (not to mention the BVP 350 and UMatic portapack remember those!), my back is acheing for lighter cameras. Unfortunately the CMOS sensor and on-board processing are a big worry for me. I provide imaging support in a laboratory and sometimes use our cameras for imaging instrumentation where we take measurements off the picture. The CMOS shear, onboard processing and weird “mustache” distortion make measurement problematical.
I guess we will have to cough up the extra dough for a PDW-F800. Unfortunately 3x cost, more weight and more power consumption means the suffering will continue for people like me.
The really intersting question in my mind is how the news people will accept this system. Will the CMOS shear be too much for them?
I would be interested in their opinions.
Jim
Posted by .(JavaScript must be enabled to view this email address) on 11/28 at 04:45 AM
Hi Simon,
Maybe this camera really targeted to users who don’t need 10bit or care about it. But there is the 10bit HPX300 now so I think shooters are bound to consider 10bit very, very carefully.
But I can really only point to myself and my own experience. I started shooting with the advent of DV, then moved through HVD and now shoot with AVCHD. Buying 2 cams a time I’m now on my fourth set and have owned Canon, Sony and Panasonic.
I don’t pretend to understand the masterminds at Sony and who they’re targeting with this cam bit I would think I’m part some market that they are targeting - maybe not with this cam but some other future offering. I’m looking to increase my capabilities and I’m willing to spend up to 20-25k each on a couple of cameras if that’s what it takes to make me more pro in terms of what 10bit can do for me.
I’d be more than satisfied with external recording to get 10bit if this camera is better than the HPX300. But is it?
With the help of sites like this and various print publishers I’ve always strived to understand what I need to be more professional and open up more opportunities with my gear.
Now I’d like to do high-level greenscreen work, more high-level compositing, etc etc. My observation is that these things really do call for 10bit.
I feel like I’ve eaten from some ‘forbidden tree of knowledge’ about 10bit and now that Panasonic delivers that seeming promise.
All of the attention I’ve paid on this subject hasn’t been lost on me. The messsage I’ve gotten loud and clear from this site and others is that 10bit is pro. With the advent of the HPX300 I think that’s becoming crystal clear that I can have it. Unless I misunderstand Panasonic’s message and who they’re aiming at with that camera.
I understand that we can record 10bit externally on this cam and that’s good. Now I wish to know how this camera and an external recorder will perform against the XPX300 in terms of picture quality.
Does 2/3” chips win over the HPX300’s 1/3” chips?
Which one does better with greenscreen work?
Which one comes away with the best pictures and in the widest range of conditions?
CODECs like AVC-I, or long GOP AVCHD don’t really factor in too much for me personally. I’ll just transcode to Cineform intermediate and get on with the work. I’m confident that I can devise a workflow around whatever camera I choose.
If I am to go for this camera, x2, I’ll need this site to conduct the ‘mother of all shootouts’ between it and the HPX300.
A big hats off to Adam Wilt for his objective analyses and professional insights!
I’m indebted to Art, this site and the more experienced pros around here for helping. It’s no small amount of money to me and I’ll agonize until I understand everything there is to consider about these two cameras and exactly how they’ll suit my work future intentions.
I hope Art can give us the ultimate technical cage-match between these cameras on the merits of their respective picture quality.
Thanks!
Posted by wsmith on 11/28 at 12:14 PM
I’m not really sure what 10-bit footage would gain you with keying. Certainly going to 4:4:4 colour space would give you a huge advantage in keying, but 10-bit colour wouldn’t really give any advantage. Further more, any 10-bit recording system is only going to be as good as the system feeding it.
Abilities such as the sensitivity and noise levels are more important to me than 10-bit recording. I feel that in reality the marketing of 10-bit with regard to cameras has become a bit of a marketing buzz word, much like megapixels on DSLR’s, that doesn’t actually address the real technical ins and outs.
In identical perfect systems 10-bit would be better than 8-bit. But camcorders from different manufacturers are not all equal. They have advantages and disadvantages that affect the final picture outcome.
So taking the HOX300 as the example, Panasonic give you on one hand (the 10-bit recording) but take away with the other (1/3” chips). Same with Sony. You get given 2/3” chips with one hand, but 8-bit recording on the other. See where I’m coming from? No single camera is ‘better’ than the other. The only determining factor is what works best for the work that you do.
Posted by Simon Wyndham on 11/28 at 12:34 PM
Simon,
Let’s suppose we can safely forget about 4:4:4 for now given that 4:2:2 is really quite satisfactory for keying. I think it’s safe to say that if the other aspects like lighting are handled properly.
I wasn’t aware that there was such a thing as an 8bit camera that provides 4:2:2 color for internal acquisition. I thought that you require 10bit if you want 4:2:2 in a digital camera. Am I incorrect there?
I do understand that any 10bit recording system (external device) is only as good as the system feeding it. That’s why I’m paying attention this camera at all. It’s HDMI out is apparently 10bit, like the EX1, EX3 and the newly announced Sony NXCAM. Were it not for that fact, I’d have dismissed this camera as a potential candidate for me.
I’m a longtime habitue of these sites. Nowhere have I gotten the impression that 10bit capability is being used hyperbolically or as a new buzzword - as was indeed the case with ‘megapixels”.
I think your final observations are worth addressing to the utmost. One mfg give this takes that away. I’m willing to reconsider my extreme desire for 10bit but only after exhaustive comparative analysis of these 2 cams.
I think it’s a bit oversimplified to say that “No single camera is ‘better’ than the other. The only determining factor is what works best for the work that you do.”
I want to see some info on what the experts say about 8bit, 2/3” chips versus 10bit 1/3” chips in objective terms that are completely separate from a particular camera’s ergonomics or lense quality or roling shutter compensation, or DRS etc.
I do undestand that people like me need to understand what we wish to accomplish and make our camera decisions predicated on that basis.
Maybe it’s only in my mind but despite the price difference, Sony and Panasonic seem to be competing for the potential user of these new camaras.
After all, here I am looking at both cams and I’m in the position to spend 2-3 times the cost of the HPX300 if the Sony is actually ‘better’ (x2 cameras).
But maybe they have two completely different user markets they are targeting. Does anyone really understand what those two different markets are?
Posted by wsmith on 11/28 at 01:42 PM
You are correct in that you were incorrect. You don’t need 10-bit to have 4:2:2. Sony’s PDW-700 for instance records 8-bit colour, but at 4:2:2 resolution. It is also worth noting that when the EBU did all their torture tests to determine a minimum recommended spec for HD broadcast they found both the XDCAM 4:2:2 50Mbit/sec codec and Panasonics AVCIntra at 100Mbits/sec to perform equally through the broadcast chain.
But again, with regards to colour space, if you are recording in progressive scan the differences between 4:2:0 vs 4:2:2 are not as great as you might imagine. Interlaced is a different matter though.
From my info I don’t believe that Sony have nailed down the price of the 350. It is also worth noting the included lens in one of the packages. Considering how expensive 2/3” lenses are, this is something that should be considered when weighing up the value for money of the camera too.
As far as the market goes, Sony sees the 350 as a direct high def replacement for the DSR-500/450 cameras. It is what a lot of people have been crying out for. There simply hasn’t been a full 1920x1080 camcorder with 2/3” chips at this price range before. Remember the HPX500 had low res chips and in reality was only good as a 720p camera.
From my own personal perspective I would not go for the Panasonic 300 for a few reasons. The first is the small chips. I shoot a lot of sit down interviews and head shots so I need to be able to get shallow depth of field even in smaller rooms. This is not possible with the 1/3” chips on the Panasonic.
Then there is the sensitivity. In 50i modes the Sony 350 is rated at f13. That’s pretty damn sensitive. Far, far better in low light than the Panasonic, and in fact even better than the PDW-700. Panasonic seems cagey about S/N ratios in their 300 specs sheet, but I can pretty much guarantee that the 350 will suffer much less from noise than the 300, especially if you find yourself in a situation where you need to use the dreaded gain switch.
Power draw. The Panasonic, despite having much smaller chips, draws 18w of power. The 350 on the other hand, even though it has much larger chips actually draws less power at 15w maximum.
The weight of the two cameras is about the same. The Panasonic is 3.6kg and the 350 is 3.5kg (without lens and battery). Both much lighter than previous shoulder mount cameras.
The Sony has four totally independent channels of audio with manual control of all of them. There are 2 XLR’s on the back and one stereo 5 pin XLR at the front. This gives a lot more versatility for many situations.
All that said, I believe that the cameras are aimed at slightly different markets. You pays yer money, you takes yer choice.
Posted by Simon Wyndham on 11/29 at 05:04 AM
Simon, thanks for your insights.
I’m beginning to feel that this cam, with it’s ability to record 10bit externally, is superior to the HPX300.
I also shoot a fair number of interviews, for a well-known performing arts video label’s bonus tracks. In all the time I’ve done that I’ve never used shallow depth of field. I don’t really see this used on TV interviews much, if at all. If I were to do one outside I might wish I had it. Everything I do is well lighted. I don’t fancy ever doing indie film shooting but who knows on that.
Even so, having those capabilities and 10bit external recording seems compelling.
I wish Canon would get with it and show us something new.
Posted by wsmith on 11/29 at 10:27 AM
10 bits means that you have 1024 gradations for luma (minus foot room and head room), while in the 8-bit system you have 256 gradations. Seems like a lot, but in fact you barely can have 50:1 contrast ratio with all gamma tricks without having banding. With 8-bit system you have to shoot precise and have less options for grading. With 10 bits you have more freedom for error. Why do you think pro still shooters use raw? Because it allows to get data back from an image which otherwise would be ruined if recorded into 8-bit JPEG. 10-bit video is nothing like 14- or 16-bit raw still photo, but it is a step forward. Obviously, greater recording latitude must be supported by greater sensor latitude.
Posted by .(JavaScript must be enabled to view this email address) on 11/29 at 09:09 PM
Adam, Very nice and thorough as always!
I even picked up info about the EX1 and EX3’s that explains things I’ve seen in footage that I’ve not been able to explain.
After reading this, I’d REALLY like Panny to send you a HPX300 to check out.
Anthony Burokas
http://IEBA.com
Posted by IEBA on 11/29 at 09:55 PM
I shoot for various Producers and Production companies for their projects and it usually is a segment for A roll or B roll. I bought a Z7 which does tape and CF cards thinking that since the footage i shoot is sent to them this was the best of both worlds. However, I’m getting requests for HD (not HDV) so that means upgrading. Your opinions will be appreciated. Between the EX3, PMW350 and HVX300, which camera would best overall to meet the majority of my clients needs? Are any of those formats lining up to be the more desired one?
Any thoughts here???
Posted by bobkat on 12/13 at 06:31 PM
Hi bobkat,
That’s the big question being pondered here. For me the EX3 is not in the running with the other two so I don’t address it (but good for B cam of PMW350). You need to ask yourself:
Is material I shoot going to be subject to heavy post? PMW350 can shoot 10bit externally; Panasonic’s 10bit internally.
Am I ever going to be shooting in really low light? Or is everything I shoot well lighted which make the Panasonic good enough?
Will I use the built-in scopes or the DRS (Dynamic Range Stretch on the Panasonic?
Do I need easy depth-of-field control? Apparently even the 1/3” chip Panasonic has “great depth-of-field” control.” But I cannot vouch for that personally.
Panasonic is the only one to deal with “flash banding” and has released compensation software to remove it. (slight bug with CMOS…)
Oh and panasonic costs a lot less. I suppose if I were running and gunning news or indie filmaking where I need that sensitivity then Sony wins.
My latest question was workflow related: If sony expects the EX3 to be used as a B cam, will we be able to transfer settings from the 350 to the EX3? Apparently not as things stand now. This is a big problem. They changed the menu of the 350 to emulate their pro camera menu system so now you’d be dealing with two different menu systems and no settings transfer if the EX3 is B cam. Got to fix that.
I was more hot for the Sony but now I’m leaning toward Panasonic for the moment. I think it’ll make me happy and cost a lot less. I wanna see some material in post. I’m a sucker for quality though so Sony has to bring it on!
See my comments above also if interested.
Posted by wsmith on 12/13 at 10:31 PM
Given enough bandwidth 10 bit is without doubt better for grading and manipulation, but with these highly compressed formats the advantage is much more questionable. You need to record 30% more data to get the same quality from 10 bit over 8 bit. AVC-I and Mpeg2 are different codecs so you can’t assume too much from just the numbers. My experience is that in many cases what I can do with an image in post production tends to be limited more by noise and compression artifacts than the bit depth of the source material. If I take the 10 bit HDSDI output from my EX (or PDW-700) and record it using a NanoFlash at 100Mb/s 4:2:2 Mpeg 2 and also record using 10 bit ProRes HQ I find that I can grade both by equal amounts before I get banding. From what I can tell the banding is stemming from imager noise and macro blocking rather than a limitation caused by 8 bit. If I then grade the 35b/s footage I do reach the limit of what I can do sooner which reenforces my view that this is a limit imposed not by the bit depth but by noise and compression artifacts.
Posted by Alister Chapman on 01/19 at 12:17 PM
Alister,
I think it’s commonly accepted that banding is caused by low bit rate acquistion which in turn results in low I/O precision in post (unless transcoded up)
Compression just tends to aggravate the appearance even more.
Posted by wsmith on 01/19 at 12:46 PM
Low bit rate or low bit depth?
Perhaps banding was the wrong description, blockiness is a better description of what I see in my tests and it’s clearly related to compression artifacts and noise. Now perhaps there is a tiny difference between 10bit and 8bit, but the high compression ratios being used are negating the bulk of any advantage 10 bit may give. To really take advantage of 10 bit IMHO you need to go to much lower compression ratios or better still uncompressed.
This was born out by the EBU codec test as Simon said earlier.
Posted by Alister Chapman on 01/19 at 12:59 PM
Alister,
I see. Blockiness is attributable to compression.
Most of us agree re the benefits of working in uncompressed. I wouldn’t want to hastily deprecate the benefits of a 10bit workflow - all other things being equal of course.
Best,
Posted by wsmith on 01/19 at 01:40 PM
As you say, all other things being equal. Which they are not. All the common current acquisition codecs are compromises where the manufacturer has to balance bit rate, bit depth, luma and chroma sampling, processing requirements and marketing hype 
Posted by Alister Chapman on 01/19 at 01:49 PM
Alister, True but if you are capturing uncompressed - even after shooting on location - you wouldn’t have to worry about that stuff.
Personally, I wouldn’t be considering this camera at all if I couldn’t externally record a 10bit uncompressed signal by SDI or HDMI.
One question I have about this camera is does the SDI or HDMI output a 10bit signal even if the recording was originally acquired as an 8bit signal to memory card. (The camera’s internal acqusition bit rate is only 8bit)
Posted by wsmith on 01/19 at 02:44 PM
The SDI output is 10 bit.
Posted by Alister Chapman on 01/19 at 02:59 PM
I did see that in the review. We have more or less come to expect that any reasonably good camera will output a live signal externally, even if only 8bit.
Inasmuch as this camera outputs a live 10bit signal externally, I wonder if will also output a 10bit signal even if the signal was originally recorded internally to a card as 8bit.
Posted by wsmith on 01/19 at 03:16 PM
All very interesting, but what are all these personal projects that require 10-bit uncompressed? Not wanting to sound pompous, but surely if you were the camera op or DP shooting on a production that required 10-bit uncompressed you’d be arranging hire from somewhere and it wouldn’t be your money you’d have to worry about.
The 350 is a DSR-500 replacement for the HD age. It fulfils that job as a primary. The fact you can attach a NanoFlash is the icing on the cake. Yet everybody seems, to me at least, overly concerned with all this 10-bit business.
If a production team came to me as a DP to find the best tool for the job for 10 bit, or to be perfectly honest more likely much higher, then depending on the budget I certainly wouldn’t be looking at an ENG style camera
If the BBC came knocking on my door for a docu that required 10-bit, then more than likely they’d be hiring in their own kit anyway.
Why does everyone worry about this stuff so much?
Posted by Simon Wyndham on 01/19 at 05:12 PM
All true. Unless you are trying to avoid banding…
I only speak for myself but 10bit is on my list of must-haves in deciding on new cameras.
This one may be an ENG style cam but I don’t mind that at all. And I know 10bit uncompressed is only going to make life easier on intensive compositing, color correcting, greenscreen work, and, of course, banding.
Accept for Panasonic’s HPX300 capturing 10bit internally, this cam seems to have lot’s going for it in my opionion. I still want to see what Cannon has up its sleeve too. They are taking their time coming to the tapeless party and I just hope they offer something with 10bit internal acquisition too. Their DSLRs are all 8bit which is fine for Canon’s stated target market: photojournalists.
Posted by wsmith on 01/19 at 06:55 PM
I’m not saying 10 bit is bad, but it is only one element of the complex chain that makes up a camera system. That system needs to perform as a whole, much like a car. Stick a $20k powerful engine in a $1K chassis and it will go fast, but only in a straight line. Put a $1k engine in a $20k chassis and it won’t go fast, but it will corner well. The best all round performer would be the $10K engine with the $10k chassis. Not the fastest, not the ultimate in handling, but good all round performance (unless your a drag racer).
The camera chain starts at the front with the lens and optics, any limitations here will restrict the rest of the system, the best codec in the world can’t make a bad lens look good. Next comes the sensor, then the image processing, then the codec, then the recording medium. In a perfect world these will all be in balance with the appropriate elements at each stage. If anything the codec and media are the least significant elements as with most cameras these are the easiest to bypass and replace with an external recorder. So if your going to spend the money IMHO it’s the front end that counts, not the rear.
While Canon may well bring out a new camera, it’s unlikely to have a 2/3” sensor. The HPX300, while highly regarded by many , has a 1/3” sensor system which for many ENG users is a non-starter due to diffraction limiting and lack of lens choice. At the moment the PMW-350 is pretty unique in offering industry standard form factor, full 1920x1080 HD and a very capable codec at a price thats significantly lower than any other HD camera with the industry standard 2/3” lens mount.
As for 10 bit v’s 8 bit, well at the kind of compression ratios were looking at here my tests have indicated to me that blockiness due to compression artifacts or camera noise tends to be a much bigger issue than banding in the majority of shots. Banding tends to only show up on flat surfaces and other large expanses of even brightess while artifacts and noise are in every single shot. For me a clean, noise and artifact free image is my first priority.
Posted by Alister Chapman on 01/20 at 10:36 AM
Regarding optional 2/3” lenses for the PMW 350. What about legacy glass from the DSR 570 cameras? Will our 2/3” Fujinon (A17x7.8 BERM-M28) being used on the DSR 570 work with the 350? This might help us decide in our purchase. Thanks.
Posted by .(JavaScript must be enabled to view this email address) on 03/08 at 10:04 AM
Well your lens of the 570 will certainly fit and it will produce a picture. The 17x7.8 is a nice SD lens and for SD it would be very good. I would guess that if you keep it between f4 and f8 it would be OK for HD, perhaps a tiny bit soft in the corners. More open than F4 and it will go soft and the 2x extender will be softer still. Until you try it on an HD camera it’s difficult to know how it will perform. There is huge variation from lens to lens and any knocks or bumps over the years may have put the collimation out. Try before you buy I say.
Posted by Alister Chapman on 03/08 at 10:13 AM
Hi Adam and other posters here re this cam.
I just got off the phone with Sony Operational support for this and other pro cams. My questions was:
“Why does Sony seem to promote the EX1R and EX3 as good B-cams for use with the PMW350 when the menus are so different that one cannot use a memory card to establish any common user-profile settings whatsoever?”
The support technician agreed and said it wasn’t the first time this question had been asked and that he’d send it over to the product management masterminds.
Posted by wsmith on 05/19 at 09:26 AM
|
