Sony announces FX9 4K camcorder with 6K fullframe sensor
- Thread starter indiawilds
- Start date
Sony can call their AF whatever they want, it's just marketing jazz. Canon's Dual Pixel AF features only continue to improve and offer additional zoning and assist features. I highly doubt Sony is even up to par with the C300 II in regards to AF speed. Certainly not up to speed with the EOS R with its latest firmware update. But we'll see when both the C500 II and the FX9 are on the market.
NorBro
Major Contributor
lol. Try the a6400.
JPNola
Veteran
The natural tendency is to compare the C500 MkII and the Fx9 head-to-head, but the price difference makes that apples to oranges. The base C500MkIIis 32% more in price than the Fx9. ( do i have the math correct? )
For $5000 more than the Fx9 you gain internal raw and other features but you give up electronic ND, dual ISO, what appears to be the better AF system, a locking lens mount, and a full-size eyepiece.
Even if you decide the C500mkII is the better camera on specs, that doesn't mean it will out-sell the Fx9 that falls into a sweeter price range for many. If you don't care about internal raw there isn't much reason to spend an additional 32% to get the C500MkII, right? One rebuttal might be that for the additional $ you get 6k, but 6k isn't a deliverable resolution.
The color-science battle between the two has yet to be waged. We just don't have enough samples yet. But, even if Canon comes out on top there, is negligibly better color science worth another $5000 that could instead be put towards glass, modules, or aks? Even with the cost of the extension unit the Fx9 still comes in at a price under the C500MkII base unit, minus any modules.
This is less a camera-war and more of a price-war.
The remarkable thing about what Sony has done with pricing is that the Fx9 is only $2000 more than the camera it builds upon, the Fs7II. Whereas the C500MkII is a $7000 jump in pricing from the C300MkII. Almost twice the price.
For $5000 more than the Fx9 you gain internal raw and other features but you give up electronic ND, dual ISO, what appears to be the better AF system, a locking lens mount, and a full-size eyepiece.
Even if you decide the C500mkII is the better camera on specs, that doesn't mean it will out-sell the Fx9 that falls into a sweeter price range for many. If you don't care about internal raw there isn't much reason to spend an additional 32% to get the C500MkII, right? One rebuttal might be that for the additional $ you get 6k, but 6k isn't a deliverable resolution.
The color-science battle between the two has yet to be waged. We just don't have enough samples yet. But, even if Canon comes out on top there, is negligibly better color science worth another $5000 that could instead be put towards glass, modules, or aks? Even with the cost of the extension unit the Fx9 still comes in at a price under the C500MkII base unit, minus any modules.
This is less a camera-war and more of a price-war.
The remarkable thing about what Sony has done with pricing is that the Fx9 is only $2000 more than the camera it builds upon, the Fs7II. Whereas the C500MkII is a $7000 jump in pricing from the C300MkII. Almost twice the price.
PBloom's FX9 AF test looked good with the Zeiss 50mm 1.4 (woman dancer), however with different lenses (Sony 16-35, 70-200 2.8 GM) on the male dancer the AF jumped abruptly to the point of non-usable quality (i.e. would need to reshoot) for anything other than news/YouTube/IG level content. Reviewing again, it looks like the AF has trouble when the subject is farther from the camera and close to the wall. The 1.4 did better due to shallower DOF (still jumped near the wall, but less noticeable).
I noticed similar AF performance with the EOS R recently where the subject is over 10' or so from the camera and there are objects/walls near them. With good/tight AI the AF shouldn't be looking at the background (pixels / AF sensors). Canon's dual pixel sensors have an advantage here vs. Sony which has far less phase detect AF sensors (not per pixel like Canon), and contrast AF doesn't have enough information without hunt/shimmer tests.
Using a beam splitter and dual optical path inside the camera could do the hunting where the final control result is given to the main lens to prevent hunting/shimmer (this is a solution for contrast detect systems to work without visible hunting/shimmer. Beam splitter loses light + lens complexity = probably not very cost effective. The light field computational cameras will have no problem here- everything's "in focus" and DOF and focus plane can be chosen via AI or in post: this will be the cost effective solution as computational cameras mature).
Here's something that could work if lens focusing was really fast and repeatable (to fix contrast AF hunting): do the hunting between frame captures. At 24fps and 1/48 (1/50) shutter you've got a lot of time* between frame captures to perform hunting to find the correct focus. Implemented correctly there would be zero hunting/shimmer. Perhaps a small low-mass lens in front of the sensor or being able to move the sensor (harder due to mass) would also work. Again, computational multilens camera systems make all this moot (and it will be cheap in mass production). With computational cameras, lenses will become software: you'll be able to purchase Cooke (my favorite for mojo), Leica, Zeiss, Canon, etc. with unique lens characteristics in terms of distortion, artifacts, color, sharpness, etc.
* Output a frame every 1/24 = 42ms, sample time is 1/48 = 21ms, that leaves 42-21 = 21ms to find focus. Higher shutter speed = more time, lower shutter = less time.
[EDIT] The male dancer AF issues could be solved with a simple non-linear ease-in/out: the jump/snap is what makes it not work, don't see hunting in that case.
I noticed similar AF performance with the EOS R recently where the subject is over 10' or so from the camera and there are objects/walls near them. With good/tight AI the AF shouldn't be looking at the background (pixels / AF sensors). Canon's dual pixel sensors have an advantage here vs. Sony which has far less phase detect AF sensors (not per pixel like Canon), and contrast AF doesn't have enough information without hunt/shimmer tests.
Using a beam splitter and dual optical path inside the camera could do the hunting where the final control result is given to the main lens to prevent hunting/shimmer (this is a solution for contrast detect systems to work without visible hunting/shimmer. Beam splitter loses light + lens complexity = probably not very cost effective. The light field computational cameras will have no problem here- everything's "in focus" and DOF and focus plane can be chosen via AI or in post: this will be the cost effective solution as computational cameras mature).
Here's something that could work if lens focusing was really fast and repeatable (to fix contrast AF hunting): do the hunting between frame captures. At 24fps and 1/48 (1/50) shutter you've got a lot of time* between frame captures to perform hunting to find the correct focus. Implemented correctly there would be zero hunting/shimmer. Perhaps a small low-mass lens in front of the sensor or being able to move the sensor (harder due to mass) would also work. Again, computational multilens camera systems make all this moot (and it will be cheap in mass production). With computational cameras, lenses will become software: you'll be able to purchase Cooke (my favorite for mojo), Leica, Zeiss, Canon, etc. with unique lens characteristics in terms of distortion, artifacts, color, sharpness, etc.
* Output a frame every 1/24 = 42ms, sample time is 1/48 = 21ms, that leaves 42-21 = 21ms to find focus. Higher shutter speed = more time, lower shutter = less time.
[EDIT] The male dancer AF issues could be solved with a simple non-linear ease-in/out: the jump/snap is what makes it not work, don't see hunting in that case.
Last edited:
NorBro
Major Contributor
IMO, there are three important factors to consider when evaluating AF online:
(1) Lenses. Self-explanatory.
(2) Firmware updates. This is Sony's first cinema camera with real video AF. As I/we have witnessed with Canon, video AF is better in its still cameras that shoot video. If Sony's is not currently optimized in this preproduction stage, it always can be.
(3) Human error. I 100% trust the notes on the screen when it says it's AF or MF, but please do note; there is always a chance for human error. Meaning the lenses could have been in MF instead and were being turned to find focus, even briefly.
(1) Lenses. Self-explanatory.
(2) Firmware updates. This is Sony's first cinema camera with real video AF. As I/we have witnessed with Canon, video AF is better in its still cameras that shoot video. If Sony's is not currently optimized in this preproduction stage, it always can be.
(3) Human error. I 100% trust the notes on the screen when it says it's AF or MF, but please do note; there is always a chance for human error. Meaning the lenses could have been in MF instead and were being turned to find focus, even briefly.
Liam Hall
Veteran
Actually, I’d say the C500 mkIi is an $8000 decrease in price from the C700.
JPNola
Veteran
If that's the case, is the Fx9 a price decrease of $31,000 from the Venice?
My point is that if you were considering buying an Fs7 Mk2, the Fx9 is only moderately more- $3000 more, give or take. Whereas, if you were about to pull the trigger on a C300MkII, stepping up to the C500MKII is a jump of another $7000. Almost double the price of the C300MkII. Sony, I argue, is better than Canon and pricing that lures you into the next class up. Fx9 is a modest amount more than Sony positioned below it, C500MkII is a hefty amount more than the Canon positioned below the C500MkII.
The numbers here do not take into account the price of the add-on modules for each camera model. The C500MkII is arguably more complete without any add-ons. As long as you are okay with only having the prosumer style loupe that comes with the C500MkII.
It's a tough call, no? You gain features over the Fx9 if you buy the C500MkII but you also give up features. Primary among them being the electronic ND, a feature you'll put to use just about every time you shoot. As opposed to a feature you gain- internal raw, that you might only use once in a while or never.
(+ @Jaime) Here's the current progress- no 1DX II specific optimizations yet, just matched FS-CL to CLog on the EOS R, then loaded FS-CL onto the 1DX II and performed these tests: http://www.dvxuser.com/V6/showthread.php?365435-Matching-1DX-II-with-FSCL-to-EOS-R-with-CLog
This^ Sony Jumped way ahead with the A6400
Yeah I really wish the XT3 would do H264 10 bit simply so I could throw it in my editing program without transcoding it first.
puredrifting
Major Contributor
I agree with the numbers, the C500 MKII, I feel is priced about $3k too high but Canon always prices their cameras too high on intro so nothing new there. The Sony looks to be a solid deal for sure, there is a lot to like there.
For me personally, I've been shooting Cinema RAW Light about 80% of the time over the past six months, it really grew from only about 50% of the time when I bought the C200. So NOT shooting RAW with the FX9, for me
and what I shoot, would not be as appealing. We even have a corporate client who requests we only shoot RAW now, which is kind of amazing compared to a year ago. We have two studio clients and a corporate that all want RAW and we shoot all of our series in RAW.
6K, meh, I shot some 6K with the BMD PCC 6K this week, I really don't care about shooting anything in 6K, so for me that's not a handicap.
No anamorphic on the FX9? Once again, I think anamorphic is somewhat of a poseur thing, so many people shoot it and use it incorrectly and ultimately, I find blue lens flares and oval lights in the BG to be pretentious for 98% of projects. To me, anamorphic is a visual language with a specific purpose and if you are shooting it merely as an effect, I find that stupid. But to each their own. At least the C500 MKII permits that and the Sony doesn't. I don't care personally but it will affect considering the FX9 for some users who want to shoot anamorphic.
Variable ND is cool and better but I wouldn't NEED it, I would WANT it. Having 10 stops is essential for me, so many cameras stop at 6 stops, that's much more important to me but Vari ND is sweet, agreed. But not a deal maker either way for me.
Having to use lame Sony proprietary media bugs me, I much prefer CFexpress simply because you'll be able to use them in lots of other cameras. I hate proprietary media now, I've been down that road with Sony and years of P2 cards. Sucks.
I am on the request list for the C500 MKII and the FX9 for in depth reviews so I'll get a chance to get hands-on with both in the real world. I'm not buying any new cameras in the near future anyway so moot point for me.
Last edited:
cyvideo
Veteran
Going by what I have read and seen it sounds very much like the AF system that they pioneered on the A9/Z90/NX80/700 series. Had me confused initially but once I got my head around it it works surprisingly well. What helped me work out how it all worked is watching these clips below as it really explains the various settings that affect subject switching, subject tracking, subject lock on, focus field depth etc. Once I experimented with theses settings I can honestly say I have virtually not had an out of focus shot and when I have needed a rapid change of focus point the system has obliged. I hope Doug Jensen doesn't mind me quoting his following comments on the Z90:
"There are only a few instances where the focus slipped from the subject in this test shots, but I wouldn't have been 100% perfect even with manual focusing . . . not even close. Very impressive technology."
https://www.youtube.com/watch?v=Y5dHmNSLQFA
If this is how they are doing it in the FX9 then there truly is a new world of focusing coming with this camera. Seeing as the focus zone on the FX9 covers more that 90% of the sensor both vertically and horizontally and the Z90 only covers about 70% of sensor I think it looks quite promising. I've never been an AF advocate. Brought up in film and then 2/3" ENG, AF was something in still cameras. Alister Chapman had a similar experience in Norway where he relied very heavily on the Z90's AF to avoid taking off his mitts in -20 degree plus freezing situations.
"The Sony PXW-Z90 – a compact 4K camcorder with auto focus at it’s best!"
"While I was up in Norway it was between -24c and -30c. In those temperatures you really don’t want to take your mittens off for more than a minute or so. Being able to rely on the cameras autofocus allowed me to keep my fingers warm. Not one shot out of all my rushes from the trip has incorrect focus. That is truly remarkable and made shooting with this camera a real pleasure. I’m not saying that you should always use autofocus. When possible I love to be able to pick and choose how I focus. But in many situations or for less experienced shooters this autofocus system will be a game changer."
http://www.xdcam-user.com/2018/02/t...act-4k-camcorder-with-auto-focus-at-its-best/
As I say if this FX9 AF is based on the A9/Z90 developments I think a few people will be quite surprised by its performance.
Chris Young
The Sony Hybrid AF explanations and settings:
https://www.youtube.com/watch?v=R5-v6SuEUJ4
https://www.youtube.com/watch?v=YcNmJq88h_E
Last edited:
Yeah if the focus behavior is tunable- should be fine.
[EDIT]: watched the demo/example: for the FX9 and cinema/pro use it needs to be non-linear, e.g. ease-out, ease-in, e.g. a cubic curve, Beziers, Hermite etc. + rate (ideally fully adjustable, as with typical animation packages):
Last edited:
FX9 has XQD slots and XQD cards are not proprietary, with the majority of FS7 shooters already having a collection of those. With internal XAVC, XQD suffices anyway, so CFexpress is somewhat of a surplus to the requirements.* If/when Sony completes that Raw deal with Atomos, the off-the-shelf SSD's should do the trick there.
*Of course, they could have chosen X-OCN. They have the extension unit, so it wouldn't make sense to skip on the OEM external recorder. Then again, charging $7,000 for R7 seems steep but, they don't want an updated cheaper version to be used with Venice, F55 or whatever mid-tier camera they may release in the future.
This subject will undoubtedly be revisited once the camera is actually released in a few months time.
*Of course, they could have chosen X-OCN. They have the extension unit, so it wouldn't make sense to skip on the OEM external recorder. Then again, charging $7,000 for R7 seems steep but, they don't want an updated cheaper version to be used with Venice, F55 or whatever mid-tier camera they may release in the future.
This subject will undoubtedly be revisited once the camera is actually released in a few months time.
puredrifting
Major Contributor
Jeromy did say that the Shogun Infreno would get support for the FX9, but limited to 12-bit due to the data rates required. The new module for the Neon has different media as well as a higher data rate and probably also a bigger processor so that it can handle 16-bit raw.
I have a feeling that the higher data rate will be the big new feature for an updated version of the Inferno before too long. Since it's not possible to swap the recording module on those, there will probably be a higher priced one that requires newer media that can record 16-bit, but Jermoy did say that the Inferno would be able to record 12-bit raw from the FX9. Of course, so will the Neon, with either module so that even if you have the new module, you can still choose to record in 12-bit raw to save space.
I suspect that most people will be satisfied with the 12-bit raw, though I'm looking forward to seeing the inevitable joke pics with people trying to mount Neons on their cameras for laughs
Some high end Nikon cameras. When the XQD was announced, those two were the primary members of the XQD group/coalition.