Most significant clue so far about Blackmagic Cinema Camera performance

[jgharding]

[quote name='John Brawley' timestamp='1344378082' post='15084']
Hmm. Not a single L series lens was used for this clip. Maybe you should read the post that gives this context.

[url="http://johnbrawley.wordpress.com/2012/08/07/some-more-blackmagic-cinema-camera-footage/"]http://johnbrawley.w...camera-footage/[/url]

JB
[/quote][quote name='John Brawley' timestamp='1344378082' post='15084']
Hmm. Not a single L series lens was used for this clip. Maybe you should read the post that gives this context.

[url="http://johnbrawley.wordpress.com/2012/08/07/some-more-blackmagic-cinema-camera-footage/"]http://johnbrawley.w...camera-footage/[/url]

JB
[/quote]

Hmm I missed that link before, cheers, I've now read it in full! I'm into Zeiss, so it's something else then. Maybe I just don't like small digital sensors. I've never really been able to "get into" hacked GH2 even with huge bitrates, I just can't say I 'feel' it. I don't 'feel' the stuff from the RX100 either, but it's just a cheap practical and usefull B-cam for running about with. That's a 1-inch sensor.

Small film? Great, I love how 16mm looks. Perhaps a small sensor treated with many many layers of post until it feels as organic will be good. But photodiodes are fundamentally very different to film, so it's not fair to compare them straight up (as some in this thread have) and say "film this big looks awesome so a sensor this big will be awesome!"

Still I see very few final products that come from small sensors that don't feel slightly... flat... in a depth sense. Something I don't see as often with 135 full frame regardless of detail or resolution. Does anyone else get what I mean? Or am I off my swede? Is it actually just the lighting or the lack of subtle post in many productions shot with small-sensor cams?

I've yet to watch the Zacuto shootout, maybe that'll change my mind.

This is certainly the most fun I've had in ages chatting before a camera's release, but we'll all know soon enough... I know I'll be giving one a try either way.

[Germy1979]

[quote name='EOSHD' timestamp='1344407152' post='15109']
Forget the hack, you are no longer going to need it.

Cannot say any more.

Photokina... I will be there!!
[/quote]

You.......butthole!!!!!!!!

Lmao.

I almost pulled the plug on a 5d... But i took your advice and i'm gonna wait. I think Canon has its strong points i like. I actually just put the "Vision Color" picture style on my T2i with Magic Lantern and i really like it... But then the Canon issues are still there. I've come to realize, Canon's are like Stratocasters, lol.. You buy one, and the first thing you do is swap out the pick ups, tuners, bridge, strings, pots, and knobs. Then you have an awesome guitar.. On a Canon, you port Magic Lantern, buy a fast card, install a log picture style, buy a mosaic filter, and then you have a "decent" video camera... :).

I wanted to see the BMC's flaws. So your title's right. This is significant.. ..
Most people's issues on DOF should be settled now for sure... I think it has great range. I can see out of windows without crushing everything in the room, and the motion looks great. Granted i'm more for 24p than 25p... (such a minor difference, but you can kind of tell..) - I think Shian's grade on the frame grab from the pool scene was very great. I've seen 5d footage look wonderful when a seasoned colorist gets ahold of it.

This is the whole package honestly. A lot of people are just going to have to stop researching and start shooting! Which is liberating considering the Waiting we all did this year, lol. Nobody even saw this camera coming! We all had our hopes up for the 5d3... Oh the irony..

I think we're all kind of selling ourselves short with this camera. I don't think it will be a let down... Most of us know how to expose a scene correctly and how to stabilize, etc. John Brawley's sole purpose is not to market this camera also, i would assume. A lot of people know about this camera, yes... But the manufacturing plant for Canon is probably 5 to 10 times larger and faster than Blackmagic's. Just a hunch based on time in as a company...

Can you imagine how long we'd have to wait to get ours if they had handed this off to a bunch of celebrity DP's to market? It's kind of a trade off. I'd have liked to have seen some different styles on this camera, but I'd also like to have one sooner than Christmas 2013.

Blackmagic's a pretty damn good company. Davinci Resolve is a super powerful coloring program & the fact that you get the freedom $100,000 will get you from acquisition, to finished product for $3000 is awesome. It's a great time to be doing this.

[cpc]

[quote name='jgharding' timestamp='1344418324' post='15116']
Still I see very few final products that come from small sensors that don't feel slightly... flat... in a depth sense. Something I don't see as often with 135 full frame regardless of detail or resolution. Does anyone else get what I mean? Or am I off my swede? Is it actually just the lighting or the lack of subtle post in many productions shot with small-sensor cams?
[/quote]

This is normal and is related to how lens microcontrast interacts with sensor pixel density.
At the same image resolution subject "pop" will be more noticeable in an image made with a big sensor/film-size. This is usually immediately noticeable with still images but is generally masked out by video compression issues in moving pictures.

Without going into much detail, in order to achieve the same pop and brilliance with a smaller sensor, you would need a lens with the same MTF result as the lens used with the bigger sensor but at a [i]significantly higher[/i] spatial frequency.
For example, if you have a 95% MTF result at 10 lp/mm. You need a lens with 95% at 23 lp/mm for a 2.3x crop camera to achieve the same brilliance and pop.
And again, video compression artefacts generally screw the image enough so that this does not matter.

[jgharding]

[quote name='cpc' timestamp='1344421732' post='15122']
This is normal and is related to how lens microcontrast interacts with sensor pixel density.
At the same image resolution subject "pop" will be more noticeable in an image made with a big sensor/film-size. This is usually immediately noticeable with still images but is generally masked out by video compression issues in moving pictures.

....

This is normal and is related to how lens microcontrast interacts with sensor pixel density.
At the same image resolution subject "pop" will be more noticeable in an image made with a big sensor/film-size. This is usually immediately noticeable with still images but is generally masked out by video compression issues in moving pictures.

[/quote]

Wow! That's fascinating stuff! Is there somewhere I can read more about this, or would you mind going into more detail? I'd love to learn the science that's behind my natural reaction to said footage.

It really is noticeable with RAW stills: medium format beats full frame beats APS-C beats M4/3 and so on in the "3D" stakes in my experience, I've even double blind picked them from selections of portraits. I see what you're saying saying: that ordinarily the interf-rame compression and the like will smear this feel in video footage so that it's less pronounced, though I think it's getting noticeable with the likes of high-bitrate GH2 and BMD CineCam.

Just as with microphones, where the physical size of the diaphragm simply is a contributing factor to the sound no matter what other tech is strapped around it, it seems the same is true of the image-capture area: you can't fake the physical reality of the size of that part however hard you try.

Will there ever be a way to make photo-diodes behave more like film? Some kind of pseudo-lightfield microlens technology but used for traditional capture not gimmicky post DOF control? to broaden their angle of absorption and the spread of light in an organic way? Am I being intuitive here, or absurd?

[jgharding]

@CPC so effectively you need a lens that is 2.3 times sharper in terms of resolving power in order to achieve the same 'feel', as the sensor is simply less forgiving? That's a tall order.

[cpc]

The thing with resolving power and contrast is that you rarely need the very high spatial frequencies. For video, small-to-medium prints and web viewing, at least. Yet, most lens reviews obsess with pointless measurements of the extremes as if they are going to print posters...

Resolving power and microcontrast are different things. You need not simply resolving power, but good microcontrast at the respective frequencies affecting viewing. Pixel count actually can define the maximum frequency that needs to be taken under consideration(unlike film, that needs both print size and viewing distance, to give anything meaningful as quantity). Note that most decent lenses will resolve above 0% MTF up to very high frequencies, so technically they have lots of resolving power even if lacking in microcontrast

It is important to know that the less pixels you need, the more the [i]low[/i] spatial frequencies are critical. For example, a lens with [i]great [/i]microcontrast at low spatial frequencies and weak result at high frequencies will deliver a more brilliant image than a lens with just [i]good [/i]results at both low and high frequencies, if you actually only care about a 1280x720 image, and not tens of megapixels. This is why some Leica and Zeiss lenses produce exceptional images for web viewing/small print purposes even on APS-C sensors. They simply have exceptional results at 5, 10 and 20 lp/mm.

And yes, if you keep the number of pixels but decrese the image field size (or sensor size) you effectively increase the maximum spatial frequency influencing the image proportionally to the crop factor. As pretty much all lenses have a decreasing MTF chart in relation of spatial frequencies, it is easy to see why the smaller pixels will lose brilliance.

There is of course more that goes into a poppy image: consistently good microcontrast accross the whole image field and well corrected aberrations are also important, because it is the exceptionally crisp edges above anything else that lead to an image with 3d pop. Subject in complete focus also helps a lot.


There is a great article on Zeiss's site about MTF charts and what they mean:
Part 1: [url="http://www.zeiss.de/C12567A8003B8B6F/EmbedTitelIntern/CLN_30_MTF_en/$File/CLN_MTF_Kurven_EN.pdf"]http://www.zeiss.de/C12567A8003B8B6F/EmbedTitelIntern/CLN_30_MTF_en/$File/CLN_MTF_Kurven_EN.pdf[/url]
PArt 2: [url="http://www.zeiss.de/C12567A8003B8B6F/EmbedTitelIntern/CLN_31_MTF_en/$File/CLN_MTF_Kurven_2_en.pdf"]http://www.zeiss.de/C12567A8003B8B6F/EmbedTitelIntern/CLN_31_MTF_en/$File/CLN_MTF_Kurven_2_en.pdf[/url]

[jgharding]

Oh dear, I spy a sinc graph! I've not read this many graphs since I read Dan Lavry's paper on AD/DA conversion and digital sampling in audio... I shall save this for the next train journey! ;) Thanks for that...

So to continue a little, why would a sensor with a 2.3x crop and only 2.5K sensor still appear so 'flat' in depth terms, despite using Zeiss glass (which can produce wonderfully 3D results with stills cameras cramming far more pixels into the same space on APS-C or full-frame) and using so little compression?

Surely the physical size of the sensor has an effect regardless, as a larger sensor is capturing light over a broader area, so the resulting image is bound to be a 'deeper' point of view?

[cpc]

Well, a few points:
1) The image we see is video compressed for web, which is quite heavily compressed. We can only judge compression once John is allowed to post DNG files.
2) The size of the light capturing area does affect SNR. Larger pixels will generally have crisper images due to lower noise.
3) Somewhat related: In terms of DOF, there is some weird notion on the web that shooting shallow focus gives more depth to the image. This is so full of bull. People should pay more attention to the semantics of the words "deep" and "shallow". A [i]slight [/i]defocusing can lead the eye to important image subjects and add some perceptual depth to an image. Very shallow focus on the other hand only produces flat images. There is plenty of "slight defocusing" (pun intended) in a sensor of this size.

[Bruno]

[quote name='jgharding' timestamp='1344433071' post='15140']
So to continue a little, why would a sensor with a 2.3x crop and only 2.5K sensor still appear so 'flat' in depth terms, despite using Zeiss glass (which can produce wonderfully 3D results with stills cameras cramming far more pixels into the same space on APS-C or full-frame) and using so little compression?
[/quote]

If you're talking about moving image, keep in mind that you rarely see 16mm or 35mm film footage that hasn't been lit at all or that was shot with cheap lenses, like most of the DSLR footage we see these days.
Even lower budget 16mm film productions rent $10k cine lenses and light the scenes properly, that's something you don't see much in sub $5-10k digital cameras footage, so yeah, it doesn't look as good...

[HurtinMinorKey]

cpc, don't you think that lighting(on the subject) is more important than DOF for making an image appear less flat?

[cpc]

[quote name='HurtinMinorKey' timestamp='1344440244' post='15147']
cpc, don't you think that lighting(on the subject) is more important than DOF for making an image appear less flat?
[/quote]

Surely light is important.
I am only discussing the characteristics of the image inherent to the camera (in this case, the sensor).

[tobyloc]

[quote name='EOSHD' timestamp='1344345988' post='15050']
Thanks for that John.

I don't mind film grain, the noise has a nice fine texture to it. I think some of the blacks would have been better off crushed a little bit in the grade to hide some of the more honest parts of the image :) It is good to see an honest image that gives a lot of options to work with in post and that is what we want as test footage. Thanks for uploading.

PS to those who can't download, you need to be a Vimeo Plus member.
[/quote]
I really like the test footage and am really happy John has released us some more, much appreciated. I am also looking forward to the camera arriving but I am a little worried now, that noise in the car's shadow does not have a film grain to it, it's got lots of banding in it and looks pretty hideous. Looking at the file in quicktime player it's a bit brighter (qt gamma issue?) and more obvious, this is a screengrab: [img]http://www.eoshd.com/comments/uploads/inline/14175/5022e071ecf8e_BM_banding.jpg[/img]
It's not in the other shadows so it is unusual, I'm not a techie but could that side of the sensor be overheating? This might be from the honest, sensor data with no noise reduction but noise reduction can't get rid of banding so I'd discount that. It's the kind of noise you might see on a 5D2 if you underexposed at high iso, and then raised in post, which worries me.

[HurtinMinorKey]

[quote name='EOSHD' timestamp='1344407152' post='15109']
Forget the hack, you are no longer going to need it.

Cannot say any more.

Photokina... I will be there!!
[/quote]

AVC Intra 100?

[Axel]

[quote name='HurtinMinorKey' timestamp='1344485744' post='15168']
AVC Intra 100?
[/quote]

If the GH3 were a GH2 with AVC Intra 100 and the internal processor supporting the then possible 10-bit 422, it would not beat off the BMCC, it would beat any competitors.

[Andrew Reid]

[quote name='jgharding' timestamp='1344425319' post='15132']Just as with microphones, where the physical size of the diaphragm simply is a contributing factor to the sound no matter what other tech is strapped around it, it seems the same is true of the image-capture area: you can't fake the physical reality of the size of that part however hard you try.

Will there ever be a way to make photo-diodes behave more like film? Some kind of pseudo-lightfield microlens technology but used for traditional capture not gimmicky post DOF control? to broaden their angle of absorption and the spread of light in an organic way? Am I being intuitive here, or absurd?
[/quote]

I think a lot of the feel of video is the fault of the display technology rather than the actual camera side. I think the sensors we have today are all very closely matched and there's no reason, lens aside, that a smaller micro four thirds or BMD sensor cannot look as cinematic as a one in a camera that is considered 'soft, organic and film-like' like the 5D or C300.

Certainly upping the bitrate doesn't make for a less cinematic image or a more electronic video feel.

What I believe is happening here is that sharp and detailed images are done a disservice by laptop screens, LCD monitors, even some high end TVs and projectors because they are too clinical, too punchy and too vibrant. The fact that the modern lenses are also this way, conspires against the sensor and yet we too often blame the camera for giving us a 'video-like' image!

Certainly there are some cameras that are more cinematic than others but the bigger varying factors are almost certainly in the display side of things, and the lensing.

Remember the film guys even accuse the Alexa and F65 of looking 'too digital', and we know that is not the case. Lots of beautiful organic and soft cinema out there shot on the 5D and GH2 let alone the Alexa. It is all about hiding the digital nature of an image with the right projection format and the right lenses.

As a DP at least you can fix one of those, the other you unfortunately don't have as much control of!

[jgharding]

I hear what you're saying about glass. On a day to day basis I deal with editing a lot more C300 than anything else, and so much of it looks really brittle and electronic, even though it grades really nicely. It's not that it's bad, just that it's a bit lifeless. Nothing that wrong with it, but not much very right either. It's usually shot with those overpriced dull-looking L lenses too, though sometimes stuff comes in from CP2s and looks a bit nicer..

But then recently I had some 7D footage from a Hot Rodded model, using some older hired PL-mount cine lenses. F**k me it looked amazing (aside from the moire). The sense of 3D depth to the movement was something else! Plus it was all nicely lit. But even the rather flatly lit shots were just lovely... the glass really makes a huge difference. I can't wait til you can get some of your Iscos on the BMD!

Speaking of monitors and compression doing justice, I watched Drive recently on standard def DVD on the Dell U2711, so compressed to hell and on a sharp monitor and it STILL looked amazing and nice and 3D and lively (sat back a little way ;) ). So I can't get into the theory that it's web compression or monitors that make footage feel flat, because if it has a real feel in its raw form it'll be there when it's compressed. If you shoot something looking dead, it's not gonna liven up because you add a few MBPS at encoding, but if you shoot it with real life, it'll maintain the life even when it's crushed down for web! Little 360p movie trailers really prove that.

So perhaps if someone whacked some really nice glass on a GH2 and used high bitrate and lit exceptionally well and used it to tell a great story I'd change my mind... but I dunno... I think there's just something about that sensor I don't feel. I suppose it's similar with old film stocks, where some would prefer one type and one another, or the way some people love the Nuemann signature mic sound and some hate it! Maybe I should put my money where my mouth is, and buy one and see what I can do with it!

Having watched that Zacuto shootout I picked the Alexa, F65 and GH2 blind, and the GH2 only fell down a step or two cos I thought it looked kinda.... flat, especially in the deeper focus toward the start :/ So I suppose it's all taste!

But the scene was just ugly and looked like something of Tommy Wiseau's The Room, so it was kinda hard to get into the feel of the shots either way.

I guess we just end back at: if the film is awesome, you won't care too much which one it was shot on. :wub:

I'm shooting for 3 days with FS700 soon so I'll be giving that a tough time to see what it can do...

[John Brawley]

[quote name='tobyloc' timestamp='1344463525' post='15164']
I really like the test footage and am really happy John has released us some more, much appreciated. I am also looking forward to the camera arriving but I am a little worried now, that noise in the car's shadow does not have a film grain to it, it's got lots of banding in it and looks pretty hideous. Looking at the file in quicktime player it's a bit brighter (qt gamma issue?) and more obvious, this is a screengrab:
It's not in the other shadows so it is unusual, I'm not a techie but could that side of the sensor be overheating? This might be from the honest, sensor data with no noise reduction but noise reduction can't get rid of banding so I'd discount that. It's the kind of noise you might see on a 5D2 if you underexposed at high iso, and then raised in post, which worries me.
[/quote]

It's not the sensor overheating.

You can usually see this kind of fixed pattern noise when you lift the blacks of the camera beyond normal. Now the camera has a wide dynamic range. You're seeing the point where the dynamic range is bottom out and intend of black you're getting the black calibration of the sensor.

Now this CAN be exacerbated by very hot cameras and hot environments. I've seen it on Canon dSLRs when they've been on and operating contioniously.

IN this case I DONT think this is what's happening.

I think what you're seeing is where the blacks run out because the sensor has run out or DR and you're left with that kind of noise. It's no longer useful shadow information. In the grade I probably should have paid a bit more attention and crushed the blacks by half a percent till it disappeared.

This shot is a very difficult shot. It's backlit and I'm trying to hold (largely succeeding) specular highlights, in an image that also has a lot of shadow. That's a huge ask of any camera.

In a camera that was doing more processing in camera, you'd just have those values down there in the blacker than what the sensor can capture values turn to solid black.

In a camera like this, you need to pay attention and remember to do it...or choose to leave it in for the marginal shadow information that's there in near black.

That's as near as my LAY understanding of how it works.

You get everything, including the marginal stuff normally truncated by other cameras. This is really a grading choice by me to leave more information IN than I probably should from a camera that gives you everything.

Post is CRUCIAL to getting the most from this camera.

jb

[tobyloc]

[quote name='John Brawley' timestamp='1344522915' post='15204']
It's not the sensor overheating.

You can usually see this kind of fixed pattern noise when you lift the blacks of the camera beyond normal. Now the camera has a wide dynamic range. You're seeing the point where the dynamic range is bottom out and intend of black you're getting the black calibration of the sensor.

Now this CAN be exacerbated by very hot cameras and hot environments. I've seen it on Canon dSLRs when they've been on and operating contioniously.

IN this case I DONT think this is what's happening.

I think what you're seeing is where the blacks run out because the sensor has run out or DR and you're left with that kind of noise. It's no longer useful shadow information. In the grade I probably should have paid a bit more attention and crushed the blacks by half a percent till it disappeared.

This shot is a very difficult shot. It's backlit and I'm trying to hold (largely succeeding) specular highlights, in an image that also has a lot of shadow. That's a huge ask of any camera.

In a camera that was doing more processing in camera, you'd just have those values down there in the blacker than what the sensor can capture values turn to solid black.

In a camera like this, you need to pay attention and remember to do it...or choose to leave it in for the marginal shadow information that's there in near black.

That's as near as my LAY understanding of how it works.

You get everything, including the marginal stuff normally truncated by other cameras. This is really a grading choice by me to leave more information IN than I probably should from a camera that gives you everything.

Post is CRUCIAL to getting the most from this camera.

jb
[/quote]

Nice one John, appreciate the response and tips on how to deal with the lows. Just not used to having a warts and all image I think, looking forward to how flexible the footage will be in post, I'm sure my editing times will quadruple just because I'll be having too much fun!

[amband]

it needs to be taken to a place where there is lots of colour contrast. We don't need an acting demo. It's not relevent.

After August, hard question will need to be asked not only of black magic, but Digital Bolex too