h a l f b a k e r ySuperficial Intelligence
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
HD is great and all, but it's actually fairly low-resolution when you get down to it. Even 1080 video is only about
2 megapixels. There are video cameras that are capable of shooting at significantly higher resolutions; naturally
these tend to be rather expensive, both because of the more advanced
technology required and because there
exists at present only a niche market for such cameras.
Many consumer level digital cameras, however, are able to capture high resolution stills at a relatively low frame
rate. Your average DSLR may be able to capture RAW stills at 3fps continuously. If we gutted 8 such cameras,
extracting the CCD and memory pathway from each camera, and mounted the sensors in a ring on the inside of a
barrel with a lens on the front, we could then use a pentaprism attached to a stepper motor to achieve 24fps
RAW video by hitting each sensor sequentially for precisely 1/48th of a second (this being the standard film
exposure time given a 180 degree shutter), coming around again just in time for the CCD to be ready to take
another picture.
Perhaps surprisingly, the cost of 8 CCD sensors and associated memory circuitry, even from a relatively high-end
DSLR and assuming you paid retail for the entire camera rather than just the necessary parts, is still substantially
less than the cost of a video camera that shoots super-HD resolution video, making this idea cost-effective.
Rotating mirror cameras
http://en.wikipedia...phy#Rotating_mirror
[mitxela, Mar 27 2012]
Please log in.
If you're not logged in,
you can see what this page
looks like, but you will
not be able to add anything.
Destination URL.
E.g., https://www.coffee.com/
Description (displayed with the short name and URL.)
|
| |
There is a point beyond which increasing image quality, since the last image-sensing element in the system - the Mk.1 Human Eyeball - itself has a limited resolution. Unless you're planning to zoom images, improving resolution is a waste of time. Better to use the bandwidth for 3D. |
|
| |
That's true, but HD 1080 is nowhere close to approaching
that resolution. Besides, shooting at a higher resolution
and downrezing to HD yields better results than simply
shooting HD in the first place, and it provides additional
flexibility in post. That's why most feature films that are
shot digitally are shot at 4K or higher, even if the intention
is to finish at HD resolution. |
|
| |
Why do it sequentially? Just split the incoming light
5 ways and do all 5 exposures at the same time. |
|
| |
Are you tiling these sensors, or superimposing them
with sub-pixel offsets? If the former, the tiling will
have to be good to avoid discontinuities. If the
latter, you'll need some schnazzy post-processing to
recover HD, and even then I doubt you could do it. |
|
| |
This is actually the method used on early high-speed cameras, so the optics would not be a problem to construct. |
|
| |
The problem I foresee is the processing required to spit out the video at the end. Unless you want, like modern high-speed cameras have, a half hour wait before being able to play back only a few minutes of video, you will need an *awful* lot of processing power. |
|
| |
Also, don't forget that the DSLR circuitry will have a buffer of maybe 100 frames, probably less. |
|
| |
In reality the computing power for continuous video at this resolution will probably put the camera into the same price range as the commercial models we're trying to avoid. But [+] for the thought. |
|
| |
Someone should invent a single-frame sensor with
light-capturing elements that are as small as, or
smaller than, the wavelength of light. |
|
| |
If these sensors could be contrived to somehow
store the image internally, and if they could be
made cheap enough, then a succession of them
could be strung together and, with suitable
mechanical apparatus, could be arranged to capture
successive still images at high resolution. |
|
| |
There might be fringe benefits, though. |
|
| |
//Why do it sequentially? Just split the incoming
light 5 ways and do all 5 exposures at the same
time.// |
|
| |
This would drastically increase the amount of
noise, since you'd have to gain the signal by the
same factor that you split it to record the same
light intensity. |
|
| |
//Are you tiling these sensors, or superimposing
them with sub-pixel offsets?// |
|
| |
They are arranged in a ring, pointed inwards
towards a pentaprism that rotates into position
before each frame is captured. |
|
| |
//The problem I foresee is the processing required
to spit out the video at the end.// |
|
| |
Not really a problem for the intended market
(that is to say, low budget filmmakers,
documentarians, and so on). Most high-end video
cameras (along the lines of the RED Epic or Arri
Alexa) require significant post-processing of the
video before you can do anything useful with it. |
|
| |
//Also, don't forget that the DSLR circuitry will
have a buffer of maybe 100 frames, probably
less.// |
|
| |
My four-year-old Pentax K10D can shoot JPEG stills
at 2-3 fps until the SD card fills up, and RAW stills
at around 1 fps. It shouldn't be hard to tweak the
memory pathway of a newer DSLR to be able to
sustain a sufficient data rate of 3 fps for RAW
images. As long as you can sustain the data rate
to the recording medium, you shouldn't have a
problem getting shots of indefinite length. |
|
| |