PDA

View Full Version : this cam good for vfx?



AhSing
06-05-2012, 10:02 PM
Thoughs on epic vfx with this cam ? Or not meant for that
?

Samuel H
06-06-2012, 03:35 AM
I would think this should be one of the BMC's strong points. Apart from the EPIC and the ALEXA, I'm not sure many cameras would be clearly better than this one for green screen. 2.5K RAW should be a great advantage here.

Brad Ferrell
06-06-2012, 07:35 AM
I routinely use VFX and the RAW shooting is what attracts me to this camera.

Kavadni
06-06-2012, 08:23 AM
I think for those like me with average pockets, it's a dream come true for VFX

Tzedekh
06-06-2012, 08:26 AM
The only issue might be the rolling shutter. I say "might" because we don't really know how severe it is.

Matthew Sonnenfeld
06-06-2012, 09:33 AM
In reality, no I don't think this camera is designed for VFX. If you are looking for green screen work, then yes this camera will perform fine but not to the extent that you are expecting. 2.5K resolution, even at RAW, is not what will make a difference in a chroma key situation. For green screen work the most important thing is the chroma resolution. Even though this is a 12bit RAW camera, that does not make it 4444 chroma. In fact, after the debayering, the chroma resolution will be above 4:2:0, but not quite 4:2:2. Barry Green has written extensively on this subject on this website, just do a search for his posts and you should find it pretty quickly.

Having said that, that does not mean that you can't get a good key out of a 4:2:0 camera, as many people have done lovely chroma key work with a 5/7D, but you have to understand that is pretty much what you are doing with the BMCC. The RAW recording gives you more latitude to use for grading in post and while this is indeed great, that does not translate to better VFX. Even BM will tell you, like they told me at the NY show, this camera is best suited to those who had been using, or considering switching to, DSLR's. Its main focus is towards feature or short narrative film work, episodic TV or web production, or commercials. Its biggest advantages are in its dynamic range, luma resolution (which when debayered will give you a pretty much perfect 1080p), and its design as a true cinema camera. Again, it's simply not made for VFX.

mhood
06-06-2012, 09:38 AM
IIRC, Barry estimated a 700,000 times improvement in my green screens over my 7D.

Tzedekh
06-06-2012, 10:10 AM
In reality, no I don't think this camera is designed for VFX. If you are looking for green screen work, then yes this camera will perform fine but not to the extent that you are expecting. 2.5K resolution, even at RAW, is not what will make a difference in a chroma key situation. For green screen work the most important thing is the chroma resolution. Even though this is a 12bit RAW camera, that does not make it 4444 chroma. In fact, after the debayering, the chroma resolution will be above 4:2:0, but not quite 4:2:2. Barry Green has written extensively on this subject on this website, just do a search for his posts and you should find it pretty quickly.

Having said that, that does not mean that you can't get a good key out of a 4:2:0 camera, as many people have done lovely chroma key work with a 5/7D, but you have to understand that is pretty much what you are doing with the BMCC. The RAW recording gives you more latitude to use for grading in post and while this is indeed great, that does not translate to better VFX. Even BM will tell you, like they told me at the NY show, this camera is best suited to those who had been using, or considering switching to, DSLR's. Its main focus is towards feature or short narrative film work, episodic TV or web production, or commercials. Its biggest advantages are in its dynamic range, luma resolution (which when debayered will give you a pretty much perfect 1080p), and its design as a true cinema camera. Again, it's simply not made for VFX.
Isn't this also true of the Alexa and Epic at their intended resolutions?

Wesley Byram
06-06-2012, 10:27 AM
For green screen work the most important thing is the chroma resolution.

Well, then let's do the math: The raw image is a 2432 x 1366 bayer pattern, which gives an effective chroma resolution of 1216 x 683 = 830528. 1080p at 4:2:2 (which is usually considered pretty excellent for green screen) has an effective chroma resolution of 960 x 1080 = 1,036,800. So, the bmc's raw image is not quite as good as 1080p 4:2:2, but both are much better than 1080p 4:2:0 (960 x 540 = 518,400). Also take into account the lack of compression artifacts in the the raw image and the BMC starts to look pretty good.

Long story short, if you are getting good enough results from 1080p 4:2:0, then the BMC's image should be significantly better, at least on paper.

Wesley Byram
06-06-2012, 10:33 AM
The only issue might be the rolling shutter. I say "might" because we don't really know how severe it is.

Indeed, this is the main reason that I have not pre-ordered yet. Keying and compositing BMC footage should be a snap, but if the rolling shutter distortion is very strong, then it could make motion tracking problematic.

Tzedekh
06-06-2012, 10:50 AM
Indeed, this is the main reason that I have not pre-ordered yet. Keying and compositing BMC footage should be a snap, but if the rolling shutter distortion is very strong, then it could make motion tracking problematic.
It's been rumored that the BMC's sensor is the BAE/Fairchild CIS2051, which has a global-shutter mode. As present, the BMC's sensor has a rolling shutter, and even if it is the CIS2051, we don't know when or if the global shutter will be enabled. But it would be nice.

Wesley Byram
06-06-2012, 11:39 AM
It's been rumored that the BMC's sensor is the BAE/Fairchild CIS2051, which has a global-shutter mode. As present, the BMC's sensor has a rolling shutter, and even if it is the CIS2051, we don't know when or if the global shutter will be enabled. But it would be nice.

If I'm not mistaken, that chip loses 2 stops of DR when in global shutter mode. It would be cool if it were possible to switch shutter modes. You could leave it in rolling-shutter mode most of the time for the DR, and switch it over to global-shutter mode when you need moving objects to be rendered correctly.

On the other hand, VFX heavy films are shot on Red and Alexa all the time, and those cameras have a rolling shutter. Perhaps it's not always a big deal in practice?

Tzedekh
06-06-2012, 12:01 PM
If I'm not mistaken, that chip loses 2 stops of DR when in global shutter mode.
I had not heard that. Do you have a link to that information? I was aware that in global-shutter mode, the maximum frame rate was cut in half.


On the other hand, VFX heavy films are shot on Red and Alexa all the time, and those cameras have a rolling shutter. Perhaps it's not always a big deal in practice?
True, but I still occasionally read about how tracking in VFX shots is made harder because of rolling shutter. And we don't know how the BMC's sensor stacks up against the Epic's and the Alexa's with respect to rolling-shutter wobble. Because the CIS2051 has a global-shutter mode, Fairchild may not have done much to minimize artifacts in rolling-shutter mode. If the BMC does use that sensor, wobble may be more pronounced than in the better-optimized Epic and Alexa sensors.

Macalincag
06-06-2012, 12:31 PM
I had not heard that. Do you have a link to that information? I was aware that in global-shutter mode, the maximum frame rate was cut in half.

http://www.fairchildimaging.com/downloads/download.php?file=sCMOSWhitepaper.pdf

Wesley Byram
06-06-2012, 12:34 PM
I had not heard that. Do you have a link to that information? I was aware that in global-shutter mode, the maximum frame rate was cut in half.

I tried to find where I read that, and all I could come up with was a post in the original epicly long dvxuser thread: http://www.dvxuser.com/V6/showthread.php?280186-BlackMagic-Cinema-camera!!!!-2-5k-for-3k/page54

Edit: According to the white paper Macalincag linked to, it appears the global-shutter mode does increase the noise in the image, which would lower the usable DR.

At any rate, we'll just have to wait for some controlled tests to see how fast or slow the sensor readout is. If it's as bad as the 5D MKII, for example, that could be a deal breaker for me. If it's more like the C300 or AF100, then that's probably good enough for tracking, as long as the amount of movement in the scene is small. We'll see.

Barry Green
06-06-2012, 01:27 PM
Isn't this also true of the Alexa and Epic at their intended resolutions?

Yes, it is. But the Alexa is 2880 pixels wide and is intended for a 1080p display, and the Epic is 5k intended for 4k, so both are oversampled. The BMC is oversampled when intended for a 1080p output, but not when intended for 2.5k.

I think the BMC will do fine for VFX work, and will be 700,000 times better than a DSLR. It won't be as good as an Epic or an Alexa, nor should any reasonable person expect it to be, since it's 1/20th the price.

laco
06-06-2012, 01:28 PM
At any rate, we'll just have to wait for some controlled tests to see how fast or slow the sensor readout is. If it's as bad as the 5D MKII, for example, that could be a deal breaker for me. If it's more like the C300 or AF100, then that's probably good enough for tracking, as long as the amount of movement in the scene is small. We'll see.

Check Brawleys test videos, mainly the one called "Dusk".
There you will see some cars passing by, and you won't see that much rolling shutter as you would see with a 5D.

Barry Green
06-06-2012, 01:38 PM
Long story short, if you are getting good enough results from 1080p 4:2:0, then the BMC's image should be significantly better, at least on paper.
Definitely agreed. Furthermore, that's as compared to real, quality, solid 1080p. If comparing to a 5D/7D generation DSLR, the reason I say it will be 700,000 times better is not only will you have dramatically sharper imagery, with much higher color sampling, but it'll also be "clean" footage -- no chroma aliasing should be present, so there shouldn't be any spurious purple and orange on your edges, conflicting with your key.

Again -- when making 1080p footage, will it key as well as a Red? Not a chance. Will it key as well as an Alexa? No way. Will it key as well as ... something like an HPX3700, with true 1920x1080 chips and true 4:2:2? Maybe. Will it key better than any FS100, FS700, AF100, or other large single-sensor camera? Definitely. Will it key as well as a C300? Probably on par, would be my guess. Will it blow everything away? Probably not -- but it's 1/5 the cost of a C300, 1/20th the cost of an Alexa, 1/12th the cost of an Epic. Will it blow away anything anywhere near its price range? Most definitely.

(at least on paper, obviously we have to see real-world results to know if it's delivering what the specs promise)

mhood
06-06-2012, 02:12 PM
(at least on paper, obviously we have to see real-world results to know if it's delivering what the specs promise)

As excited as I am about getting my BMC, I'm just as excited about you getting yours. ;-)

Any word about when the first batch goes out to you "prominent bloggers"?

Wesley Byram
06-06-2012, 02:51 PM
Check Brawleys test videos, mainly the one called "Dusk".
There you will see some cars passing by, and you won't see that much rolling shutter as you would see with a 5D.

Aye, the slanted cars jumped out at me immediately. They were moving pretty fast though, and the distortion wasn't *too* bad, but it was clearly visible and squashed any doubts about the BMC having a CCD chip.

There have been (anecdotal) reports that the distortion is as severe as that on an iPhone, while Adam Wilt has likened it to the EX1 and Red One, which are pretty good on that front. I'd still like to see a controlled test with the BMC set up against a common reference camera. I'm hoping that the BMC's readout is at least faster than that of my humble HMC40.

Barry Green
06-06-2012, 03:07 PM
Keying and compositing BMC footage should be a snap, but if the rolling shutter distortion is very strong, then it could make motion tracking problematic.
That is a valid concern, and one that is faced by any effects artist working on any rolling shutter camera (including the Red One). We will have to see how the rolling shutter in the BMC performs for such tasks. The Red is pretty good, but it's nowhere near as good as the Alexa, and the Red is extremely better than the 5D, so -- the amount of rolling shutter does indeed matter. Hopefully the BMC will be on par with the Red; if so, then that's a level of performance that the industry has pretty much had to accept, and established workflows should work with it. If it's a lot skewier than the Red, then that would be something to be concerned about.

Barry Green
06-06-2012, 03:18 PM
As excited as I am about getting my BMC, I'm just as excited about you getting yours. ;-)

Any word about when the first batch goes out to you "prominent bloggers"?
No word yet.

Barry Green
06-06-2012, 03:22 PM
Slanted cars is such an uncontrolled test though. We don't know how far away the cars were, what the lens magnification was, what speed the cars were going, etc...

The Red/EX1 are pretty much the best performers of common video CMOS chips; they and the HPX300/370, HPX250, AF100, F3, etc. all perform about the same. The DSLRs (7D/5D) are quite a bit worse, and the iPhone is really really really bad.

If the BMC ends up being on par with the Red/EX1 etc., I would think that's a best-case scenario that we should hope for. And John seems to think that it's on par with them, and the little bit of skewing I did with one at NAB seems to confirm that, as far as skewing, although jiggling the camera sure did make everything go mega-rubbery, so the jury's still out until we get some scientific testing. But I am expecting it to have good rolling shutter performance.

Wesley Byram
06-06-2012, 04:30 PM
Slanted cars is such an uncontrolled test though. We don't know how far away the cars were, what the lens magnification was, what speed the cars were going, etc...

Exactly.

Well, the camera's shipping next month, correct? I imagine that it won't be long before someone gives it a proper test.

legrevedotcom
06-24-2012, 01:17 PM
Again -- when making 1080p footage, will it key as well as a Red? Not a chance. Will it key as well as an Alexa? No way. Will it key as well as ... something like an HPX3700, with true 1920x1080 chips and true 4:2:2? Maybe. Will it key better than any FS100, FS700, AF100, or other large single-sensor camera? Definitely. Will it key as well as a C300? Probably on par, would be my guess. Will it blow everything away? Probably not -- but it's 1/5 the cost of a C300, 1/20th the cost of an Alexa, 1/12th the cost of an Epic. Will it blow away anything anywhere near its price range? Most definitely.

(at least on paper, obviously we have to see real-world results to know if it's delivering what the specs promise)

Barry... in regards to the statement on keying on BMC vs. FS100. Is that compared to footage taken from the SD card or of an external recorder in 4:2:2?

I currently own a FS100, and since the arrival of the BMC, my next move has been quite a dilemma. I'm caught up between getting the Samurai recorder or getting a BMC cam. The reason is that I want to be able to get that little extra out of the footage in the grading. I'm not necessarily talking heavy VFX but merely being able to push the footage without getting banding etc.
The AVCHD holds up pretty decently, but I can feel the structure just waiting to crumble ;)

Peter J. DeCrescenzo
06-24-2012, 02:20 PM
Barry... in regards to the statement on keying on BMC vs. FS100. Is that compared to footage taken from the SD card or of an external recorder in 4:2:2? ...

I'm not speaking for Barry, but my take on the statement of his that you quote above is that he's comparing the on-board (built-in) recording capabilities of all those cameras. He doesn't say anything about external recording.


... I currently own a FS100, and since the arrival of the BMC, my next move has been quite a dilemma. I'm caught up between getting the Samurai recorder or getting a BMC cam. The reason is that I want to be able to get that little extra out of the footage in the grading. I'm not necessarily talking heavy VFX but merely being able to push the footage without getting banding etc. The AVCHD holds up pretty decently, but I can feel the structure just waiting to crumble ;)

Theoreticals aside, a definitive answer to this part of your question can't be known until after the BMCC ships and would require a careful, side-by-side test of an FS100 and BMCC under identical conditions including lighting, lens, subject, computer keying hardware & software, and so forth. A valid test like this may eventually occur, but understand that until then, the road to heck is paved with intelligent (and not so intelligent) conjecture. :-)

mhood
06-24-2012, 02:29 PM
Theoreticals aside, a definitive answer to this part of your question can't be known until after the BMCC ships

You keep saying that as though it's a given. A few prototypes handed out ahead of the ship date to "prominent bloggers" (like Barry) is not only something BMD has committed to (IIRC) but is also a pretty common practice in the Industry. It's not expecting too much for answers to come from more than one source in a pre-shipment timeframe. It's sort of what the review community is all about.

Peter J. DeCrescenzo
06-24-2012, 02:42 PM
You keep saying that as though it's a given. A few prototypes handed out ahead of the ship date to "prominent bloggers" (like Barry) is not only something BMD has committed to (IIRC) but is also a pretty common practice in the Industry. It's not expecting too much for answers to come from more than one source in a pre-shipment timeframe. It's sort of what the review community is all about.

I don't entirely disagree with you. However, what's "normal" about the BMCC? Not much. So why should its roll-out, including its pre-release press/blog handling, be "normal"?

But anyway, BMD makes lots of products. I suspect they'll handle the BMCC's roll-out similar to their other products, plus or minus a few variations. We shall see.

Cheezweezl
06-25-2012, 03:32 AM
Barry... in regards to the statement on keying on BMC vs. FS100. Is that compared to footage taken from the SD card or of an external recorder in 4:2:2?

I currently own a FS100, and since the arrival of the BMC, my next move has been quite a dilemma. I'm caught up between getting the Samurai recorder or getting a BMC cam. The reason is that I want to be able to get that little extra out of the footage in the grading. I'm not necessarily talking heavy VFX but merely being able to push the footage without getting banding etc.
The AVCHD holds up pretty decently, but I can feel the structure just waiting to crumble ;)

I know this is off topic but...

I own an FS100 as well. To help you with your decision, using an external recorder such as the samurai comes with pros and cons. I'll skip the convenience factor of not having that extra chunk of gear to worry about as well as the advantage of a nice external monitor built into the samurai or whatever. I'll go straight to the two biggest points that made me decide to skip a recorder.

1. The image is definitely better, but only marginally. I grade for a living and yes, the avc holds up well for most things. The biggest advantage for me would be cleaner keys when shooting greenscreen, but I can bring a computer to that set and record over hdmi directly into fcp or whatever.

2. The avc is part of the low light magic on the FS100. The clean hdmi out is nowhere near as clean as the avc footage at higher ISOs. Being that this is my low light go to camera, that was a deal breaker.

Maybe your priorities are different but I thought I'd share since I was in the same boat.

I bought the bmc for the DR.