Balazs Rozsa

In some of its cameras Fuji uses a special CCD which has photodiodes with different sensitivities. The camera combines the results from the two kind of photodiodes and produce a high dynamic range image (they claim only 2 extra stops). I think this is a much better way to increase dynamic range than trying to increase dynamic range of the individual photodiodes. At least after a certain point. Because now you probably want to use all your photodiodes to make the resolution of the chip as high as possible. But as the number of pixels increase on sensors, at some point you will choose to increase dynamic range rather than resolution.

Three 12MPixel 60x40mm CCDs if we talk about the same camera from 2001. Specificationwise it is still one of the best. They were considering it for for movie making as well. http://www.arri.com/infodown/news/0309_e.pdf

A few years ago I read the D20 can output the full data from its 6Mpixel sensor (internally downsampling to HD was the other option). I haven't heard about that since then. Maybe they gave up on that feature.

That is correct.

When you shoot 1080 and 720 you have a choice of using the full sensor area or only the middle section of the sensor. When you use the full sensor you have 35mm DOF but you are limited to a maximum of 60 fps. When you use the middle section you can have a maximum of 120 fps.

If it was only the chroma information, they could unite these pairs into smaller pixels. But that would halve the vertical luma resolution to 1080 which is only the native resolution of the output. Now without uniting they can use downsampling to generate the output that will give you a sharper luma image. But yes, the chroma resolution is about equivalent to what you could get from a 6 Mpixel Bayer imager. But chroma resolution doesn't need to be as high as luma (still it is quite high resolution considering it is downsampled from 6 Mpixels).

In the Kinetta one 1.8" 40MB HDD records 110Mb/sec sustained (50g iPod HDD).

Could you please tell at least the basic idea behind this statement? Look at the Fuji extended dynamic range sensors. If you are willing to dedicate half the number of your pixels to highlight aerias, you can extend the dynamic range. That is a possibility for the future, when increasing the resolution of the image will not be as important as increasing the dynamic range. Or look what happened to CMOSes. A few years ago they were very noisy, now the Canon digital SLRs have quite low noise. Every time you can add one more transistor to a CMOS pixel cell, you can build one better amplification circuit with less noise. And miniaturization allows that. Sure the fundamental workings of CCD and CMOS imaging devices did not change too much since their invention. But they are steadily getting better. Not big changes, but noise, dynamic range, resolution getting better steadily.

Jim Murdoch wrote: "It's going to take more than "another" generation too; CCD and CMOS image sensors have been around for 35 years now, and this is about as good as they're going to get. The laws of physics can be terrible curmudgeons sometimes." Do you mean they will not get better from now on? In the last 35 years they have been improving nicely, especially in the last few years. What laws of physics would stop them getting better now suddenly?

The reason ?10 years ago they were predicting that soon ? and look after ten years still nothing happened? sounds good. But, if somebody was predicting 10 years ago that movies would be shot digitally soon, he could only hope that big companies like Sony and the like will come up with a new big cheap tape mechanism that can record 4K or 8K images (which was unlikely). Or he could hope that movie makers will be happy with 2K resolution (which was obviously not the case). Now the situation is different. Computer technology is getting near the point where it will allow the recording of very high resolution images economically. Please! Theoretical limits are much higher than what is available now. And there are new technologies like laser interference volumetric storage and who knows what else. But of course you cannot base your prediction upon these vague things. Instead just look what the current hard drive manufacturers are doing and saying. The technology you will find in the shops in the next three or four years already exists in the laboratories. And even after that they have a fairly good idea how they will increase storage capacity. Capacity is so high already that you do not need to go that much far in the future to see how computer storage will allow high quality video capture cheaply. 10 years ago a 1GB hard drive was considered a really big hard drive, you could not even dream about recording any kind of video feasibly. Now we are very close. Ten years from now storing the video on hard drives will not be an issue. But this is the first time we can say that. And this time it is quite easy to see it is true.

If you consider on camera recording from a purely storage capacity point of view, the Kinetta shows well what can be techonologically achieved. It is 6.5kg with battery and a hard drive magazine and it records 1.2Gb/sec into a fault tolerant raid array. At about 1.9Gb/sec you can record uncompressed 2K 10bit 24fps. That surely could be put into 10kg. And it uses the old 40GB drives, the new 60GB drives are already on the market. So I think uncompressed 2K on-board recording is very well within reach. When will be it 4K? Based on the increase of storage capacity in the last 15 years (which is about a steady 1.8 fold a year), the jump from 2K to 4K will take slightly more than 2 years. The jump to 8K another 2 years, and the jump to 16K another 2 years and so on. Of course it is possible that we are just at the point where the storage improvement will slow down and then we will need maybe 10 years for 16K capture.

Thanks for the info!

Where do you get this information? I could only find that the SRW1 records in both the 4:2:2 and the 4:4:4 mode at a rate of about 440 mbps and the 880 mbps mode is used for recording two streams of video. Balazs

Yes, the Genesis CCD has 12.4 Mpixels, but the camera output is only 2 Mpixels. The Dalsa gives you 8 Mpixels. Balazs

I think the difference comes from the fact that Dalsa specifies its ASA sensitivity for 8 megapixels while the Genesis is specified for 2 megapixels. When the sensor sizes are about the same. Balazs

I read a lot of discussions about the Panavision Genesis camera, but have not known they were working on another digital camera as well. I found out about the new camera at: http://www.tvtechnology.com/hd_notebook/one.php?id=8 On a NASA Research Partnership Center site they have a pdf file with some more detailed info about the camera: http://voyager.ee.fau.edu/Document/Panavis...%20pamphlet.pdf The camera has a Bayer color filtered 3840x2160 33mm diagonal CMOS sensor that was designed by Panavision affiliate, Panavision SVI. It can record progressive images with a variable frame rate up to 30 frames per second. It seems everybody is working on Ultra HD cameras these days.

There is some more information from Arri News: "Together with Lockheed-Martin, an experimental CCD camera was developed and tested under demanding conditions. Featuring three 12 megapixel sensors, the camera was designed to show the upper limits of image quality as would be required to completely supersede the 35 mm film format. The set-up was able to demonstrate superior resolution and color reproduction, however, it also showed that the path to a compact and cost-effective production camera with appropriate storage media was not a short-term venture." I was thinking that for the appropriate storage media they could use two SRW-1 VTRs (the one on top of the Genesis) for recording. The SRW-1 can record two video streams simultenaously (900Mbits/s!). They would only need to divide an 4K image into four 2K parts. The 12 megapixel sensor would give a sharp downsampled 4K image. This is something you cannot easily do with the Dalsa camera. The Dalsa people say they need to store the original data because their de-Bayer processing is slower then realtime. The result is a big recorder. But with 3 CCDs there is no need for de-Bayer, the SRW-1 can compress and store the video on tapes.

This is the Lockheed-Martin camera from 2001. A joint project with Arri using three 12MPixel 60x40mm CCDs. Information is from the 2003 September Arri News. Balazs

The light is not split up into 3 ways first and then filtered for each of the CCDs, but the splitting and the filtering happens at the same time. For the red CCD the red portion of the light is extracted from the light beam and directed towards the red CCD, for the blue CCD the blue light is extracted and so on. This may not be a big difference, but in my opinion this is the nice thing about a prism video camera: all the incoming red light reaches the CCD pixels. If the light was split into 3 ways and then filtered, only about one third of the red light would reach the red CCD. The same is true for the bayern filtered one CCD cameras: only about one third of the light gets to the pixels, the rest is blocked in the filters. This is the reason 3 CCD cameras are more sensitive to light than 1 CCD cameras. At http://www.perrybits.co.uk under Articles in "Article 4: How does a video camera capture color?" you can read a precise desription how the prism and the filters work. Balazs

Hello Patrick, I'm glad to see a person here from DALSA. Can I ask why it is that much more problematic to design a 4K sensor than a smaller 2K one. At first look I would think that if you put together 4 pieces of 2K sensors, you will have a 4K sensor. You need to increase the number of signal outputs four times as well to make the read-out speed large enough because of the increased number of pixels. Is the production yield problematic? I'm very curious. Thanks, Balazs

What I tried to say really in the beginning is that with a bigger sensor you can shoot sharper images. Not because the sensor can have more pixels, but because it is easier to construct a sharp lens for a bigger sensor. When you have an 8K sensor for example, you need extremely sharp lenses. Let's say you have a 35mm sensor and you want to shot with a 35mm lens. Because the 8K sensor requires extremely high resolution, you need at least 10mm diameter of aperture to prevent diffraction soften the image (I didn't exacly calculated this 10mm). Then you need to use your lens at f/3.5. But it is very hard if at all possible to construct a wide lens that gives you 8K resolution from edge to edge, no chromatic and other abberation, etc. But if you increase the size of the 8K sensor to 100mm for example, you can use a 100mm lens at f/10 to get 10mm diameter of aperture. A lens that can give you a sharp 8K picture at f/10 is easier to construct.

I'm sorry, I meant f stop. You do not need four times as much light: If you use a 4K 35mm chip with a 35mm lens at f/3.5, and a 4K 65mm chip with a 65mm lens at f/6.5 you will get the same picture. The diameter of aperture will be 10 mm in both cases so the depth of field will be the same and so will be the amount of light getting onto one pixel on the CCDs. The only difference between the two situations will be that in the first case the lens will work at f/3.5 while in the second case at f/6.5. If you want extremely good sharpness, you should choose the lens working at f/6.5. Completely independently of the size of your sensor, if you want to shoot a scene at f6.5 rather than f3.5 you need MORE LIGHT. Thta's just physics. You cannot just close your lens down 2 stops. If you are shooting outside in the sun, getting an f6.5 will not be a problem, but if you have to light a set to f6.5 rather than f3.5 that will be a big difference. You will need four times as much light to win 2 stops. If you are talking about film this is true. You need to get a certain amount of light per a given area on the film for proper exposure. For digital this is not the case. If you have a pixel that has an area that is four times as big as the area of another pixel, the bigger pixel will be able to give you a better signal if they get the same amount of light apiece. In the digital still photography world you can see how using a bigger imaging device is better for your picture. The digital SLRs have much better light sensitivity and noise than the compact digital cameras with smaller sensors. You can close your lens down several stops more on your SLR camera to get the same amount of noise in the picture that you get from your compact camera. As pixel counts will increase, so will the sizes of the imaging sensors. There is no point to make a 10 MPixel sensor that is the size of the sensor of your current 2MPixel camera. No lens can give you a sharp picture onto such a small area.

I'm sorry, I meant f stop. You do not need four times as much light: If you use a 4K 35mm chip with a 35mm lens at f/3.5, and a 4K 65mm chip with a 65mm lens at f/6.5 you will get the same picture. The diameter of aperture will be 10 mm in both cases so the depth of field will be the same and so will be the amount of light getting onto one pixel on the CCDs. The only difference between the two situations will be that in the first case the lens will work at f/3.5 while in the second case at f/6.5. If you want extremely good sharpness, you should choose the lens working at f/6.5.

I don't think that's going to happen, because you will need completely new lenses (not even 65mm lenses would work on this) and your depth of field would drop so much that it would become very impractical to shoot. Your depth of field would not drop because you would use these lenses at a higher f ratio. Actually the higher f ratio is exactly the reason why the lenses for a bigger sized CCD give you a sharper picture. Pixel count will get really high, recording capacity for the data will be cheap. The only resolution limiting factor will be the lens. I think movie makers will not be able to resist building larger CCDs with sharper lenses.

The 65/70mm films have more resolution of course, but I think this is exactly why digital will be much better in a few years than film. Digital resolution and storage will increase steadily and at some point the lenses will not be good enough to take advantage of all the resolution a 35mm CCD will give you. Then the CCD (or some other imaging device) size will start increase and quite soon 100mm will be reached for example. It is true such a chip will be expensive, but not very expensive. As opposed to how expensive it would be to shoot and work with 100mm moving film. With digital you only need to manufacture this chip only once. The output of these large CCDs will be incredible. Maybe the only limiting factor will be that you will not need more resolution after a point.