Corporate loyalty and my next video-camera
Call me a fan of Sony when you hear me enthusiastically talk about my Sony video-camera. Yet I am not.
My first camera was a Sony, in 1993 it cost 12000NOK and recorded on Hi8 -I chose it because you set the focus with a ring at the lens, like a traditional SLR.
I have never used auto-focus. Nor auto-shutter, or any other turnoffable automatic function. This ensures that my images look good.
In the meanwhile I went through some other cameras -JVC, Panasonic, Samsung...
Now I´m back at Sony. I bought it in 1999 for 12000NOK and it records on digital8. I chose it because you set the focus with a ring at the lens...
Now, it´s literally falling apart -yet it is still working.
Until I was in the states recently I had never used a semi-professional camera(known as prosumer) -I thought them expensive and offering no real advantages. Top of the line consumer videocameras are good quality and come with *all* of this years latest features and concepts.
The one I posess can shoot in the dark using infra-red, a life-saver many a time in less than ideal lightning conditions -it also has a look and useable for this as well. It can set the shutter-time to slower than 25fps, significantly increasing light while reducing the frame-rate, another life-saver and another look available to me(chungking express uses this extensively for example).
Most prosumer cameras emit these ´consumer´ features -fihuring that ´real´ videographers only shoot in ideal light-conditions, yet -given that the creation of such conditions virtually always will cost many times the cost of the camera...
Real videographers use 500000NOK cameras.
I once rented one to shoot a movie, it was big -heavy. I had to have a couple of friends along just to carry it(did not have access to a car -despite the fact that the camera cost as much as three average cars at the time -since I was below 25 no car-rental agency would rent me a car without a massive deposit which I could not afford, slightly avsurd)
Now, that camera had good-looking pictures, they blew me away -I could not see a single artifact(though I´m sure they were there) -it was HD(recording 700 lines on Betacam SP) for all practical purposes and intents -a real doubling of available resolution *and* colour-resolution.
A prosumer camera has 5% more rez than it´s consumer equivalent, there is zero increase in colour-rez. It uses a 3CCD system which makes it significantly less light-sensitive while offering no real advantage(those 5% up there and bragging rights).
Thus I conclude that a consumer camera is alot better than a prosumer camera.
You wanna talk about the sound? No dude, you don´t.
The built in mic´s in a camera aren´t really good for film-making, not because of any quality issue but because they´re in the wrong place. You don´t want your mic closer to your camera crew than your actors.
So you need a loose mic, you gotta buy it anyway.
But prosumer cameras comes with XLR mic-inputs and phantom power. Yes indeed, and I´m sure the mic-pres are great(ahem). But for the cost-difference between the prosumer and the consumer I´ll buy a fucking glorious sound-recording *system*.
Now we enter a new era in video-camera quality.
The first consumer HD camera.
I´ve seen it priced at 17000NOK.
It offers more than double the rez, is tiny, comes with a focus ring at the lens, and even has a real control for the shutter nearby.
Wonderful news.
And because it´s consumer, it has all those silly features and FX that are fun to play with and sometimes, a life-saver.
But in the HD era we have a new topic of discussion -bitrate.
The bitrate of a digital signal determines how many bits are processed or transmitted(to tape say) per second. More is better.
You see, our digital cameras uses several different modes of compression.
First in the actual image capture; consumer and prosumer uses the same color-sampling(4:1:1), essentially -it captures less rez for the chroma-signal(colour) than for the luma-signal(black/white). This is why colour often look terrible on both consumer and prosumer cameras(my opinion, go ahead -argue) and you can see it most evidently on shows like americas funniest home videos. Ofcourse, a professional videographer like myself knows about this and does his best to conceal it.
This process ofcourse, reducese bandwidth. The resulting signal is then digitalized, and the digital bitstream is compressed using another lossy technique -throwing away more rez and detail in the picture and the resulting bitstream is captured on tape.
Now I see people talking about how this camera is better than that camera because it has more bitrate. In theory this is correct(given everything else being the same).
I´ve been working with digital image compression some years now and have seen amazingly good quality pictures in amazingly small bit-streams, the digital bitstream seems amazingly resilient to compression. Notice the success of mp3 audio and divx video. Both typically reduce the bitstream by a factor of ten while most users observe no loss of quality. I have myself failed in blind-tests to identify what is what, though most of the time, my trained eye and ear(I´ve only got one of each -I´m mono :-)) will detect one or more artfifacts introduced by the process.
And so I say the bitrate is irrelevant as long as it´s sufficient. And twice as much as sufficient will not deliver a twice as good picture, you get 5%.
The real problem is the colour-sampling I discussed above, professional cameras captures twice as much colour-rez(4:2:2) before the signal is digitized and compressed.
What we want is that. Or even better than that, we want *full* colour-rez(4:4:4), which is currently not available below ludicruos cost, even if we´re gonna compress the living daylights out of the bitstream.
There is now available superhighrez cameras that shoot up to 8 or 16 megapixels at full colour-rez, they don´t use tape, they use a truckload of raided hard-drives(quite literally) to capture this enormous uncrompressed bitstream.
Drool.
Now, let´s have a look at frame-rates.
Film shoots 24 frames per second(FPS), this is why fast motion on film is never smooth. Our eyes capture about 80 fps(subject to individual differences and levels of intoxication B-)).
PAl(european video) shoots 25 fps interlaced, NTSC shoots 30 fps(actually 29.97) interlaced.
Interlacing is process which splits the image in two, on a line by line basis.
First the odd lines are captured in a half-frame, then the even lines in the other half(or vica versa -makes no difference).
In every frame there thus is temporal distortion. The benfit of this process is that motion seems smoother(since the real framerate is twice that advertised). The disadvantage are the temporal artifacts introduced -which are ugly.
Now, there seems to a market for the ´film-look´ with more stuttering motion but with no temporal artifacts. We call this to shoot progressively, where the whole frame is captured in one go.
In this brave new world of HD, somehow, interlacing has stayed on. This is completely ununderstandable to me. Interlacing was introduced back in the 40s as a way to reduce bandwidth while keeping percieved image quality on par. 60 years later we don´t need it.
Fortunately HD comes with several alternative resolutions and framerates, they are:
1920x1080(2 megapixels) at 30 frames interlaced
1920x1080 at 24 frames progressive(what digital cinema is using)
1280x720(1 megapixel) at 30 frames interlaced
1280x720 at 60 frames progressive.
Compare this to what we´re currently using
720x576(0.4 megapixels) at 25 frames interlaced(PAL)
720x486(0.3 megapixels) at 30 frames interlaced(NTSC)
As you can see, the framerate sucks on all of these formats but one.
1280x720 is quite a lot of resolution and looks good on screens several meters wide.
60 frames per second is close to the limit of our eyes and produces very natural motion(as a film-maker, I am more concerned with motion than with rez).
Douglas Trumbull, the famous FX dude(look him in the IMDB if you´re cluelss) did experiments back in the 70s with cranking film up to 60 fps. He reported that film at 60 fps lost all its characteristics and the audience stopped thinking about as film and pulled out sentences like ´window to the world´. In essence, motion was now so natural that the film looked real.
And that is actually the reason why nobody touched it. There seemed to a belief that film should be artificial(so as not to confuse the audience?). If you have ever been to theme-parks with motion simulator rides you have most likely seen hi-rez 60 fps projection already -motion simulation is possible at anything less. Like, ask a computer-gamer if he can do anything but just die and die and die if he were playing counterstrike(or any other first person shooter game) at 24 fps. This is also the reason why there aren´t really a market for first person shooter on consoles like x-box or playstation2. They´re unplayable. The TV only shows 25 or 30 fps no matter.
(another reason is that the hand-controls that come with these consoles won´t let you make fast movements, but this can be allviated by plugging in a mouse and a keyboard).
And from this rant you should understand that the only HD format that carries any interest to me is 1280x720 at 60 fps. I can hardly wait. But I´m too broke now to go out and buy, soon, very soon.
Interlacing belongs in the steam-age, along with 30 fps video and, ehm, steam engines. 24 fps film is an anachronism, with a partial economic explanation -running film on 60 fps would more than double the actual amount of film required for a copy of a movie, and film is expensive.
Now, I would like to discuss how we should throw 24 bit colour depth into the toilet where it surely belongs and move to 48 bit, but that´s a whole ´nother issue. Later dudes -I´m heading for the beach.
b
My first camera was a Sony, in 1993 it cost 12000NOK and recorded on Hi8 -I chose it because you set the focus with a ring at the lens, like a traditional SLR.
I have never used auto-focus. Nor auto-shutter, or any other turnoffable automatic function. This ensures that my images look good.
In the meanwhile I went through some other cameras -JVC, Panasonic, Samsung...
Now I´m back at Sony. I bought it in 1999 for 12000NOK and it records on digital8. I chose it because you set the focus with a ring at the lens...
Now, it´s literally falling apart -yet it is still working.
Until I was in the states recently I had never used a semi-professional camera(known as prosumer) -I thought them expensive and offering no real advantages. Top of the line consumer videocameras are good quality and come with *all* of this years latest features and concepts.
The one I posess can shoot in the dark using infra-red, a life-saver many a time in less than ideal lightning conditions -it also has a look and useable for this as well. It can set the shutter-time to slower than 25fps, significantly increasing light while reducing the frame-rate, another life-saver and another look available to me(chungking express uses this extensively for example).
Most prosumer cameras emit these ´consumer´ features -fihuring that ´real´ videographers only shoot in ideal light-conditions, yet -given that the creation of such conditions virtually always will cost many times the cost of the camera...
Real videographers use 500000NOK cameras.
I once rented one to shoot a movie, it was big -heavy. I had to have a couple of friends along just to carry it(did not have access to a car -despite the fact that the camera cost as much as three average cars at the time -since I was below 25 no car-rental agency would rent me a car without a massive deposit which I could not afford, slightly avsurd)
Now, that camera had good-looking pictures, they blew me away -I could not see a single artifact(though I´m sure they were there) -it was HD(recording 700 lines on Betacam SP) for all practical purposes and intents -a real doubling of available resolution *and* colour-resolution.
A prosumer camera has 5% more rez than it´s consumer equivalent, there is zero increase in colour-rez. It uses a 3CCD system which makes it significantly less light-sensitive while offering no real advantage(those 5% up there and bragging rights).
Thus I conclude that a consumer camera is alot better than a prosumer camera.
You wanna talk about the sound? No dude, you don´t.
The built in mic´s in a camera aren´t really good for film-making, not because of any quality issue but because they´re in the wrong place. You don´t want your mic closer to your camera crew than your actors.
So you need a loose mic, you gotta buy it anyway.
But prosumer cameras comes with XLR mic-inputs and phantom power. Yes indeed, and I´m sure the mic-pres are great(ahem). But for the cost-difference between the prosumer and the consumer I´ll buy a fucking glorious sound-recording *system*.
Now we enter a new era in video-camera quality.
The first consumer HD camera.
I´ve seen it priced at 17000NOK.
It offers more than double the rez, is tiny, comes with a focus ring at the lens, and even has a real control for the shutter nearby.
Wonderful news.
And because it´s consumer, it has all those silly features and FX that are fun to play with and sometimes, a life-saver.
But in the HD era we have a new topic of discussion -bitrate.
The bitrate of a digital signal determines how many bits are processed or transmitted(to tape say) per second. More is better.
You see, our digital cameras uses several different modes of compression.
First in the actual image capture; consumer and prosumer uses the same color-sampling(4:1:1), essentially -it captures less rez for the chroma-signal(colour) than for the luma-signal(black/white). This is why colour often look terrible on both consumer and prosumer cameras(my opinion, go ahead -argue) and you can see it most evidently on shows like americas funniest home videos. Ofcourse, a professional videographer like myself knows about this and does his best to conceal it.
This process ofcourse, reducese bandwidth. The resulting signal is then digitalized, and the digital bitstream is compressed using another lossy technique -throwing away more rez and detail in the picture and the resulting bitstream is captured on tape.
Now I see people talking about how this camera is better than that camera because it has more bitrate. In theory this is correct(given everything else being the same).
I´ve been working with digital image compression some years now and have seen amazingly good quality pictures in amazingly small bit-streams, the digital bitstream seems amazingly resilient to compression. Notice the success of mp3 audio and divx video. Both typically reduce the bitstream by a factor of ten while most users observe no loss of quality. I have myself failed in blind-tests to identify what is what, though most of the time, my trained eye and ear(I´ve only got one of each -I´m mono :-)) will detect one or more artfifacts introduced by the process.
And so I say the bitrate is irrelevant as long as it´s sufficient. And twice as much as sufficient will not deliver a twice as good picture, you get 5%.
The real problem is the colour-sampling I discussed above, professional cameras captures twice as much colour-rez(4:2:2) before the signal is digitized and compressed.
What we want is that. Or even better than that, we want *full* colour-rez(4:4:4), which is currently not available below ludicruos cost, even if we´re gonna compress the living daylights out of the bitstream.
There is now available superhighrez cameras that shoot up to 8 or 16 megapixels at full colour-rez, they don´t use tape, they use a truckload of raided hard-drives(quite literally) to capture this enormous uncrompressed bitstream.
Drool.
Now, let´s have a look at frame-rates.
Film shoots 24 frames per second(FPS), this is why fast motion on film is never smooth. Our eyes capture about 80 fps(subject to individual differences and levels of intoxication B-)).
PAl(european video) shoots 25 fps interlaced, NTSC shoots 30 fps(actually 29.97) interlaced.
Interlacing is process which splits the image in two, on a line by line basis.
First the odd lines are captured in a half-frame, then the even lines in the other half(or vica versa -makes no difference).
In every frame there thus is temporal distortion. The benfit of this process is that motion seems smoother(since the real framerate is twice that advertised). The disadvantage are the temporal artifacts introduced -which are ugly.
Now, there seems to a market for the ´film-look´ with more stuttering motion but with no temporal artifacts. We call this to shoot progressively, where the whole frame is captured in one go.
In this brave new world of HD, somehow, interlacing has stayed on. This is completely ununderstandable to me. Interlacing was introduced back in the 40s as a way to reduce bandwidth while keeping percieved image quality on par. 60 years later we don´t need it.
Fortunately HD comes with several alternative resolutions and framerates, they are:
1920x1080(2 megapixels) at 30 frames interlaced
1920x1080 at 24 frames progressive(what digital cinema is using)
1280x720(1 megapixel) at 30 frames interlaced
1280x720 at 60 frames progressive.
Compare this to what we´re currently using
720x576(0.4 megapixels) at 25 frames interlaced(PAL)
720x486(0.3 megapixels) at 30 frames interlaced(NTSC)
As you can see, the framerate sucks on all of these formats but one.
1280x720 is quite a lot of resolution and looks good on screens several meters wide.
60 frames per second is close to the limit of our eyes and produces very natural motion(as a film-maker, I am more concerned with motion than with rez).
Douglas Trumbull, the famous FX dude(look him in the IMDB if you´re cluelss) did experiments back in the 70s with cranking film up to 60 fps. He reported that film at 60 fps lost all its characteristics and the audience stopped thinking about as film and pulled out sentences like ´window to the world´. In essence, motion was now so natural that the film looked real.
And that is actually the reason why nobody touched it. There seemed to a belief that film should be artificial(so as not to confuse the audience?). If you have ever been to theme-parks with motion simulator rides you have most likely seen hi-rez 60 fps projection already -motion simulation is possible at anything less. Like, ask a computer-gamer if he can do anything but just die and die and die if he were playing counterstrike(or any other first person shooter game) at 24 fps. This is also the reason why there aren´t really a market for first person shooter on consoles like x-box or playstation2. They´re unplayable. The TV only shows 25 or 30 fps no matter.
(another reason is that the hand-controls that come with these consoles won´t let you make fast movements, but this can be allviated by plugging in a mouse and a keyboard).
And from this rant you should understand that the only HD format that carries any interest to me is 1280x720 at 60 fps. I can hardly wait. But I´m too broke now to go out and buy, soon, very soon.
Interlacing belongs in the steam-age, along with 30 fps video and, ehm, steam engines. 24 fps film is an anachronism, with a partial economic explanation -running film on 60 fps would more than double the actual amount of film required for a copy of a movie, and film is expensive.
Now, I would like to discuss how we should throw 24 bit colour depth into the toilet where it surely belongs and move to 48 bit, but that´s a whole ´nother issue. Later dudes -I´m heading for the beach.
b
posted by bangskij at 12:51
0 Comments:
Legg inn en kommentar
<< Home