Nirv
Carbonite Member
I definitely see the motion but I think there's an answer.
Give me a few days to sort some footage out, I just became really busy.
Sony cameras and the "video look".
- Thread starter Simon Shasha
- Start date
I definitely see the motion but I think there's an answer.
Give me a few days to sort some footage out, I just became really busy.
Seriously, it's like I don't even need to post anymore; I couldn't agree more. The main reason I pre-ordered the BMC was because of how well the motion rendering looked, and then for the reasons everyone else did, dr, raw etc.
Kyle McConaghy
Well-known member
Haha... I thought this post was written by me b/c I thought the same exact thing. I'm going to hang onto my FS100 for low-light, run & gun, etc. But I ordered the BMC mainly b/c of realism and film-like quality of the motion and dynamic range.
Nirv
Carbonite Member
It's almost as though I needn't have bothered posting.
For those that want the same look as the C300 video, a 90 degree shutter and no IS is the answer.
But for those that want to pretend it's something else then I'm sure there's a camera company out there willing to take your money.
For those that want the same look as the C300 video, a 90 degree shutter and no IS is the answer.
But for those that want to pretend it's something else then I'm sure there's a camera company out there willing to take your money.
philiplipetz
Veteran
The motion is nearly the ONLY reason we have ordered two of the cameras, and a FS for low light, for our next project. It looks creamy smooth even on a good showroom monitor. Been driving me crazy because there are so many vital pieces left out of this camera, some that can be replaced by kit and some that will be forever missing. Yet, The footage creates a flow that will make it so much easier to immerse the audience in the projected reality rather than having them watch it from a distance. There is a sense of history, instead of the video now there is past, present and future. We are shooting a mystical piece and this flow is exactly what we need.
Ran tests with C300 and it is not there so look forward to any test Nirv or anyone less can do to recreate the same flow.
Ran tests with C300 and it is not there so look forward to any test Nirv or anyone less can do to recreate the same flow.
Nirv
Carbonite Member
For those that are testing the 90 deg shutter and IS off also make sure you do a flat grade like the example video posted above. It makes all the difference.
It's not just one thing that gives the motion of that C300 video it's several things working together that gives the feel that people are after. I'd shoot with cinegamma 1, look after the highlights and flatten out the image further in post. The footage I have just shot looks every bit as good as the canon footage.
Don't take my word for it, try it yourself.
It's not just one thing that gives the motion of that C300 video it's several things working together that gives the feel that people are after. I'd shoot with cinegamma 1, look after the highlights and flatten out the image further in post. The footage I have just shot looks every bit as good as the canon footage.
Don't take my word for it, try it yourself.
Last edited:
paulcurtis
Veteran
So people are buying BMC based on motion from a preproduction model despite the full sized images being packed with aliasing. Therefore the camera isn't finished (as i'm sure this will be sorted out) and who knows what you'll get when it's out? Don't get me wrong i think it seems a wonderful camera (and if they'd put a m43 mount on it i would consider one)
Nirvs right, i really don't get it. At least with pointing out the motion on vimeo you eliminate some of the possibilities - it's not frame syncing or codecs.
Are we sure it's just not shutter angle and contrast? I did point out that excessive detail and contrast emphases stuttering.
Kholl, by the same token can you point to something on vimeo from that FS that you don't like?
thanks
Paul
Nirvs right, i really don't get it. At least with pointing out the motion on vimeo you eliminate some of the possibilities - it's not frame syncing or codecs.
Are we sure it's just not shutter angle and contrast? I did point out that excessive detail and contrast emphases stuttering.
Kholl, by the same token can you point to something on vimeo from that FS that you don't like?
thanks
Paul
ftfy
paulcurtis
Veteran
thanks (had to look up my internet urban dictionary for that one!)
The image i was referring to was one of the only full sized samples from bondi beach. The aliasing is not JPEG compression but the kind of aliasing the Leica exhibits (as it has no AA filter). It's partly a debayering issue also. Regardless where it comes from i'm sure it won't be in the final production models. But all i'm saying is that buying a camera for a specific feature when it's not been released seems a little premature, especially based on so little real samples.
cheers
paul
philiplipetz
Veteran
This thread has spilled over to BMCuser.com and I cross post a note that offers an explanation of video motion, especially ifnthe FS100/700.
The Tessive filter works by reducing aliasing artifacts introduced by the video shutters opening and closing in a square wave fashion. Film shutters are rotating disks with curved edges that act as a sine wave, gradually opening and closing the shutter. Tessive filters are a secondary filter that approximates the way that film shutters open and close. This acts as a way to smear out shutter based aliasing introduced qui quickly scanning a moving object to create harmonic waves of amplified artifical intensity. The result of using the Tessive filter is that there is less aliasing associated with movements, and this this recreates the creamy look of film. The film look comes from the filter smearing temporal changes and thus reducing artificial harmonic aliasing created by the combination of a shutter and moving subject.
The Arri Alexa can use a mechanical mirror shutter to duplicate what the Tessive filter does, or it uses an electronic shutter modified to not have a square wave form (http://www.arri.de/?eID=registration&file_uid=8218). Similarly, the Sony F65 cine camera can be ordered with a mechanical shutter. So, this is something that is known to high end cine camera designers, but usually ignored video cameras. That is why motion can be so different between cine and video cameras.
The BMCC appears to use a Fairchild chip that reduces this subtle aliasing in other ways. While it is possible the BM altered with the shutter activity, or has modified a sensor to work in a new way, the antialiasing characteristics of the Fairchild chip could account for a lot of the BMCC's magic, both in motion between frames and image within each frame.
The Fairchild sensor has also has greater sensitivity per pixel and antiblooming to prevent signals from spilling over from one pixel to another. These combine to allow the sensor to achieve greater resolution per pixel at a greater signal to noise ratio than is possible with conventional back lit CMOS sensors. While this creates greater dynamic range it also means less shutter aliasing.
While the fairchild sensor has a global shutter mode, the BMCC uses the rolling shutter mode so the camera will have jello type spatial aliasing but will have reduced artifacts from shutter aliasing of resolution based information.
The Fairchild site says, "As long as the frame rate is such that the camera is temporally oversampling object dynamics, negligible spatial distortion will be observed. Such oversampling is good imaging practice, since it is undesirable to have an object travel a significant distance during a single exposure." If BMCC uses the high sampling rate of this chip to over sample for each exposure then that will act like a film shutter to smear out shutter artifacts. Of course all we know is that oversampling is possible, not if the BMCC uses it.
While I have been discussing primarily luma information up to now the same sensor characteristics combine to allow less smearing and aliasing of chromatic information, especially since there is less blooming and noise to create spill over spill over between pixels optimized for different colors.
Video cameras, especially Sony cameras such as the FS100/700 have noisier sensors and are optimized for low light capture, not high signal to noise ratio. Therefore, they have built in signal processing that amplifies some alaising artifacts, there is no getting around this. While not glaringly obvious this shows up most visibly in the amplification of shutter based alaising, hence the Sony look, or just the video look since the sensors on most video cameras create the same design constraints.
The Sony FS700 is capable of very high frame rates and the engineers could have opted to use that to over sample each frame as a way to reduce shutter aliasing.
Moving into the area of pure speculation, perhaps the refusal of BM to offer higher ISO settings on the camera, when the sensor has a base ISO OF 800 and most base 800 ISO sensors can deliver decent images at ISO 6400 or beyond, is due to BM not wanting to use sensor circuitry that will create these subtle aliasing artifacts. Hopefully the motion looks as good with compressed formats that require sensor processing.
People, including me, were upset that the BMCC offers only uncompressed RAW, while cameras such as the RED offer RAW with variable compressed ratios. While this could be simply a way to avoid the expense of compression it also could be a way to keep the signal as clean as possible. Again, what does this say for motion display within compressed formats on the BMCC?
It is possible that the film like BMCC image is due to a less aliased image but will this hold up in compressed formats? Most assurances that ProRes and DNxHD will be very close to RAW are based on close equivalence in chroma information, this form of shutter aliasing is luma based and so "could" be affected by the codec circuitry.
The Tessive filter works by reducing aliasing artifacts introduced by the video shutters opening and closing in a square wave fashion. Film shutters are rotating disks with curved edges that act as a sine wave, gradually opening and closing the shutter. Tessive filters are a secondary filter that approximates the way that film shutters open and close. This acts as a way to smear out shutter based aliasing introduced qui quickly scanning a moving object to create harmonic waves of amplified artifical intensity. The result of using the Tessive filter is that there is less aliasing associated with movements, and this this recreates the creamy look of film. The film look comes from the filter smearing temporal changes and thus reducing artificial harmonic aliasing created by the combination of a shutter and moving subject.
The Arri Alexa can use a mechanical mirror shutter to duplicate what the Tessive filter does, or it uses an electronic shutter modified to not have a square wave form (http://www.arri.de/?eID=registration&file_uid=8218). Similarly, the Sony F65 cine camera can be ordered with a mechanical shutter. So, this is something that is known to high end cine camera designers, but usually ignored video cameras. That is why motion can be so different between cine and video cameras.
The BMCC appears to use a Fairchild chip that reduces this subtle aliasing in other ways. While it is possible the BM altered with the shutter activity, or has modified a sensor to work in a new way, the antialiasing characteristics of the Fairchild chip could account for a lot of the BMCC's magic, both in motion between frames and image within each frame.
The Fairchild sensor has also has greater sensitivity per pixel and antiblooming to prevent signals from spilling over from one pixel to another. These combine to allow the sensor to achieve greater resolution per pixel at a greater signal to noise ratio than is possible with conventional back lit CMOS sensors. While this creates greater dynamic range it also means less shutter aliasing.
While the fairchild sensor has a global shutter mode, the BMCC uses the rolling shutter mode so the camera will have jello type spatial aliasing but will have reduced artifacts from shutter aliasing of resolution based information.
The Fairchild site says, "As long as the frame rate is such that the camera is temporally oversampling object dynamics, negligible spatial distortion will be observed. Such oversampling is good imaging practice, since it is undesirable to have an object travel a significant distance during a single exposure." If BMCC uses the high sampling rate of this chip to over sample for each exposure then that will act like a film shutter to smear out shutter artifacts. Of course all we know is that oversampling is possible, not if the BMCC uses it.
While I have been discussing primarily luma information up to now the same sensor characteristics combine to allow less smearing and aliasing of chromatic information, especially since there is less blooming and noise to create spill over spill over between pixels optimized for different colors.
Video cameras, especially Sony cameras such as the FS100/700 have noisier sensors and are optimized for low light capture, not high signal to noise ratio. Therefore, they have built in signal processing that amplifies some alaising artifacts, there is no getting around this. While not glaringly obvious this shows up most visibly in the amplification of shutter based alaising, hence the Sony look, or just the video look since the sensors on most video cameras create the same design constraints.
The Sony FS700 is capable of very high frame rates and the engineers could have opted to use that to over sample each frame as a way to reduce shutter aliasing.
Moving into the area of pure speculation, perhaps the refusal of BM to offer higher ISO settings on the camera, when the sensor has a base ISO OF 800 and most base 800 ISO sensors can deliver decent images at ISO 6400 or beyond, is due to BM not wanting to use sensor circuitry that will create these subtle aliasing artifacts. Hopefully the motion looks as good with compressed formats that require sensor processing.
People, including me, were upset that the BMCC offers only uncompressed RAW, while cameras such as the RED offer RAW with variable compressed ratios. While this could be simply a way to avoid the expense of compression it also could be a way to keep the signal as clean as possible. Again, what does this say for motion display within compressed formats on the BMCC?
It is possible that the film like BMCC image is due to a less aliased image but will this hold up in compressed formats? Most assurances that ProRes and DNxHD will be very close to RAW are based on close equivalence in chroma information, this form of shutter aliasing is luma based and so "could" be affected by the codec circuitry.
cpc
Well-known member
There is nothing creamy about film. Film is artifact laden strobe fest.
None of the Tessive examples I've seen is even remotely mimicking film. Tessive video is overly fluid and reminiscent of high frame rate images.
philiplipetz
Veteran
Never seen a Tessive, but I have seen Alexa footage. there is a difference.
Nirv
Carbonite Member
The tessive filter makes film/video look smoother not more stuttery and not like the BMD (video posted above) camera/GH2. I include the GH2 there as Kholi says it produces the film like stutter that he looks for in a camera.
I am of the opinion that the look in the video above is due to higher shutter speed and a non stabilised lens (and grade).
If you shot with a lower shutter speed and stabilised lens the camera still records all of the motion blur of your micro movements and combined with the stabilised image it has a more video feel. This is in no doubt.
As a former scientist I'm a great advocate of testing and proving the theory, so I would love to see some side by side tests when the BMD camera becomes available. We are lacking vital information about how the video was shot so we're urinating in a blackout and the Fairchild BMCC info reads like marketing blurb.
Interesting subject though!
paulcurtis
Veteran
All sensors have anti-blooming, usually when the companies make a deal of it it's because it's a CCD not a CMOS as CCDs are read out in columns usually (the vertical streak effect) whereas the pixels on CMOS are quite separate.
I fail to see how a sensor can affect the way the photons are built up in a way that could mimic film response. Correct me if i'm wrong but a pixel on a sensor is a well of energy created by photons hitting it which is read out at the end of an exposure. Unlike film whereby the physical film is exposed and therefore that introduces frame by frame differences in the exposed medium itself. The speed at which you read out the wells dictates the rolling shutter effect. At the end of the process you get a frame with all the intensity levels in the wells as a single snapshot (or two snapshots if you're doing the HDRx thing!)
The tessive filter is a super clever idea (although isn't it for use on film as well - it deals with time aliasing?) but a lot of folks here are saying that only some digital cameras show this dodgy motion cadence, the BMCC doesn't and that doesn't have a time filter in front of it (i doubt!)
how exactly?
thanks
Paul
I fail to see how a sensor can affect the way the photons are built up in a way that could mimic film response. Correct me if i'm wrong but a pixel on a sensor is a well of energy created by photons hitting it which is read out at the end of an exposure. Unlike film whereby the physical film is exposed and therefore that introduces frame by frame differences in the exposed medium itself. The speed at which you read out the wells dictates the rolling shutter effect. At the end of the process you get a frame with all the intensity levels in the wells as a single snapshot (or two snapshots if you're doing the HDRx thing!)
The tessive filter is a super clever idea (although isn't it for use on film as well - it deals with time aliasing?) but a lot of folks here are saying that only some digital cameras show this dodgy motion cadence, the BMCC doesn't and that doesn't have a time filter in front of it (i doubt!)
how exactly?
thanks
Paul
Nirv
Carbonite Member
I wonder if the "look" that Kholi likes could be down to camera package size and way of holding?
The BMCC appeas to have a similar DSLR type size and shape and the micro-motions of the operator may be very similar.
When I went out for my morning walk I took my FS100 and shot with several shutter speeds and with IS on and off. All combinations gave slightly different feels but the motion is different to the BMCC clips. My FS100 was fully loaded and in its heaviest configuration so I'd be surprised if I could generate the same frequency content of motion as someone with a smaller lighter package.
Food for thought.
There's so much to like about the BMCC but there's unfortunately many things that have been compromised for the price that make it unworkable for me. Not bad for V1.0 though. If they take on the feedback of their customers future versions could be awesome.
There's loads of dead pixels in the Dusk Beach night footage video so I wonder if they're b grade sensors?
On another note, the Tessive people claim that it removes rolling shutter! My ickle brain can't work that out, anyone else?
The BMCC appeas to have a similar DSLR type size and shape and the micro-motions of the operator may be very similar.
When I went out for my morning walk I took my FS100 and shot with several shutter speeds and with IS on and off. All combinations gave slightly different feels but the motion is different to the BMCC clips. My FS100 was fully loaded and in its heaviest configuration so I'd be surprised if I could generate the same frequency content of motion as someone with a smaller lighter package.
Food for thought.
There's so much to like about the BMCC but there's unfortunately many things that have been compromised for the price that make it unworkable for me. Not bad for V1.0 though. If they take on the feedback of their customers future versions could be awesome.
There's loads of dead pixels in the Dusk Beach night footage video so I wonder if they're b grade sensors?
On another note, the Tessive people claim that it removes rolling shutter! My ickle brain can't work that out, anyone else?
For the record, I personally didn't refer to the Magic Cam examples to turn this into a Magic Cam thread, but I probably should've known it would've done that. Haha. There's no other way to really compare without doing so, because people will start to question the other camera or choices etc (like what's going on now.)
I linked to Brawley's videos because each and every one of them exhibit the motion that I think is desirable (to me). The video of the guys toying at the C300 was pointed out because Paul mentioned micro jitters, it was the perfect example of how stabilization doesn't matter. Every one of those videos does it for me. Not just that one.
No, I didn't say anything about a film like stutter, ever. Well, I mean, if I did then quote me so I can correct that. I can't say that it's stutter or blur, as both of those can change with shutter angle. Yes, depending on the camera changing the shutter angle can help, but it doesn't cure it for me.
To me, it has nothing to do with stabilization. Lock the camera off, you'll see the same thing (if you can see it.)
I personally won't test it, simply because I already know what I'm looking for. If a camera doesn't have it, it won't magically have it when I go to test it against a camera that does. But, if anyone else wants to test it, have at it. I also don't think it'll prove anything to anyone that doesn't already see it. If anything, it'll come down to too many variables and that'll give those people more reason to believe that everyone else is seeing something.
I'd rather leave it at the idea that I'm just seeing things (along with a lot of other people). Less time consuming that way.
And, only specific cameras actually get to where the minority wants it to be. So, that's another thing to consider. Why does one camera have "the goods" if the lot of them are CMOS, similar compression, etc.
No. It isn't. Has nothing to do with camera movement, has nothing to do with operator error, anything like that. Like I said, you could put it on sticks and it'd still be the same.
Feel free to post those clips.
This is why they aren't releasing footage. Too many times it's been said that the footage seen was from very early stages, dead pixels are going to be there because the sensor has not been balanced. But, if people don't take the time to read they A. Assume it's the final image and B. Start speculating, which causes a spread of speculation amongst the people they communicate with.
Like an infection. Then, when the camera's finally released, everyone with accurate information has to go back and combat that.
Do you know why you haven't seen any night time/lowlight FS700 footage yet? Just daytime or stuff where there's a lot of light?
I've had the camera, and I know exactly why.
Anyway, though, as far as the topic goes... I kinda think it's just not worth worrying about. It really may be similar to audio, where those that can tell the difference between certain brand-badged instruments will buy those because it's what they want, while others can't hear any difference.
Paul, you asked about pointing to some FS100 footage that lacks the motion... I'd say just look at every FS100 video out there. James Miller's stuff I feel is what I would use as examples of what the FS100 is capable of, and not a single video on his vimeo gets to the point that the few of us here are talking about, in regards to motion I mean.
I linked to Brawley's videos because each and every one of them exhibit the motion that I think is desirable (to me). The video of the guys toying at the C300 was pointed out because Paul mentioned micro jitters, it was the perfect example of how stabilization doesn't matter. Every one of those videos does it for me. Not just that one.
No, I didn't say anything about a film like stutter, ever. Well, I mean, if I did then quote me so I can correct that. I can't say that it's stutter or blur, as both of those can change with shutter angle. Yes, depending on the camera changing the shutter angle can help, but it doesn't cure it for me.
To me, it has nothing to do with stabilization. Lock the camera off, you'll see the same thing (if you can see it.)
I personally won't test it, simply because I already know what I'm looking for. If a camera doesn't have it, it won't magically have it when I go to test it against a camera that does. But, if anyone else wants to test it, have at it. I also don't think it'll prove anything to anyone that doesn't already see it. If anything, it'll come down to too many variables and that'll give those people more reason to believe that everyone else is seeing something.
I'd rather leave it at the idea that I'm just seeing things (along with a lot of other people). Less time consuming that way.
And, only specific cameras actually get to where the minority wants it to be. So, that's another thing to consider. Why does one camera have "the goods" if the lot of them are CMOS, similar compression, etc.
No. It isn't. Has nothing to do with camera movement, has nothing to do with operator error, anything like that. Like I said, you could put it on sticks and it'd still be the same.
Feel free to post those clips.
This is why they aren't releasing footage. Too many times it's been said that the footage seen was from very early stages, dead pixels are going to be there because the sensor has not been balanced. But, if people don't take the time to read they A. Assume it's the final image and B. Start speculating, which causes a spread of speculation amongst the people they communicate with.
Like an infection. Then, when the camera's finally released, everyone with accurate information has to go back and combat that.
Do you know why you haven't seen any night time/lowlight FS700 footage yet? Just daytime or stuff where there's a lot of light?
I've had the camera, and I know exactly why.
Anyway, though, as far as the topic goes... I kinda think it's just not worth worrying about. It really may be similar to audio, where those that can tell the difference between certain brand-badged instruments will buy those because it's what they want, while others can't hear any difference.
Paul, you asked about pointing to some FS100 footage that lacks the motion... I'd say just look at every FS100 video out there. James Miller's stuff I feel is what I would use as examples of what the FS100 is capable of, and not a single video on his vimeo gets to the point that the few of us here are talking about, in regards to motion I mean.
Nirv
Carbonite Member
Kholi
For someone with such ninja vision you don't seem interested in trying to understand what is exactly causing the motion you so desire.
If BMD didn't want footage e posting they wouldn't be giving the camera to people. One thing that is very obvious is the rolling shutter in the BMCC, it's as bad as the worst DSLR. Which is something I can see in abundance. 1 step for 2 back IMHO.
From a personal point of view I think the motion here is far superior since you quote miller, https://vimeo.com/31896063.
From what we understand of how the tessive filter works the motion that you desire would appear to be undesirable by most other observers.
For someone with such ninja vision you don't seem interested in trying to understand what is exactly causing the motion you so desire.
If BMD didn't want footage e posting they wouldn't be giving the camera to people. One thing that is very obvious is the rolling shutter in the BMCC, it's as bad as the worst DSLR. Which is something I can see in abundance. 1 step for 2 back IMHO.
From a personal point of view I think the motion here is far superior since you quote miller, https://vimeo.com/31896063.
From what we understand of how the tessive filter works the motion that you desire would appear to be undesirable by most other observers.
One person. Not "people", and no raw footage. None after the week of NAB. It's not very hard to find this information on the internet, they've been very vocal about their choices.
Really? Because, I panned the hell out of the camera with my own hands and I am pretty sure it's nowhere near as bad as the worse DSLR.
Guess this is another case of two sets of eyes seeing something different. All well and good.
Far superior, but there was no difference before hand? I'm confused.
I don't know what you understand, but I never said anything about the Tessive filter's motion being desirable. Ever. I never even commented on it... which leads me to your first comment...
I'm not interested when people are putting words in my mouth, thoughts in my head.
At that point, it's not worth anyone's time. And, the conversation's already shifting the way they always do: I don't think I'd be surprised if someone came out of the shadows to tell me I was bashing the camera and to move on. xDDD