Canon 3 Layer Foveon Type Sensor

[Flake]

Now this is interesting, because a friend (reviewer) playing with the new Sigma SD1 is saying that the only thing which betters it is a 40MP+ medium format, and that nothing in 35mm comes close, in fact it makes them look 'Soft & two dimensional'.

Having looked at the patent his comment is that with the R & D resources of Canon, they could take this to a new generation and that Bayer matrix sensors could follow polaroid into history.

The politics of this Patent are very interesting, Canon would not announce this unless they were either close to releasing it in a real camera, or that they were doing it to wrong foot the competition, I can't imagine they would spend the R & D to reach a Patent stage without an intention to launch a product.

They only qualm I have about a camera of this type, and that is the cost. Sigmas latest caused a few sharp intakes of breath, similar to a 1Ds I wonder if Canons first offering will be any cheaper.

 

[Rocky]

The politics of this Patent are very interesting, Canon would not announce this unless they were either close to releasing it in a real camera, or that they were doing it to wrong foot the competition, I can't imagine they would spend the R & D to reach a Patent stage without an intention to launch a product.

Patent is a very interest tool. It has been used to stop someone else to make similar product, as saving in the bank, as trading card, etc. IBM was well known to hold (and even buy) a lot of patent that has got nothing to do with any of their own product line. Better yet you can patent an idea without making it actually work.
So I will believe it when the actual product is made.

 

[Loswr]

I can't imagine they would spend the R & D to reach a Patent stage without an intention to launch a product.

Patents are cheap, and even though the R&D to support them is not necessarily cheap, if it's part of the normal course of business, then it's written off. The 120MP APS-H sensor is case in point for that. I have no idea of the patentroduct ratio overall for Canon, but it's far greater than 10:1. In big pharma, that ratio is over 1000:1.

 

[Flake]

The foveon type sensor is a known technique, and the fact there's a model already available precludes this being an attempt to stop others entering this area of the market, but it does indicate that Canon is either or has been conducting research into it. Of course we don't know just how successful that research has been, nor the size of the sensor that they've been working on, as it's a single pixel size patent.

The Patent was filed 18 months ago (pre earthquake) and only just released, I do hope that this does reach the market, a camera with a similar pixel count to a 5D MkII would produce a 60MP count sensor without the associated noise/ dynamic range issues that plague Bayer types.

 

[Edwin Herdman]

I would love to play with a Canon attempt at a Foveon type sensor, however there are some particular setbacks to the technology, particularly as it has been implemented by Sigma. I imagine that Canon might be looking for a way to overcome the problem of fabrication plant limitations on chip size. They might not suffer as badly as Sigma from too-small photosites (at least I recall that being a characteristic of the Foveon) due to their good chip production ability.

If the approach is very similar to Foveon, though, there are still a whole bag of problems to be solved.

One thing I am not clear about is whether this design will allow them to record the three photosites as a single vector, versus whatever the Foveon does which seems to require them to call 15MP worth of pixels "45MP" and which makes the Foveon cameras and processing much slower than shooting from a Bayer-filtered sensor.

If there is no AA filter, that will have some effects as well, including false sharpness and the potential for artifacts. Observer metamerism is a problem with the Foveon type approach, as well, which leads to colors simply not looking right in unpredictable ways. Joseph S. Wisniewski has written about these issues elsewhere (I use his information as the basis of my posts on Foveon).

 

[Mt Spokane Photography]

Better yet you can patent an idea without making it actually work.
[/quote]

You cannot patent a idea, that would be rejected immediately.

Its a common fallacy that a person can patent a idea, not so. Its true that patents may not be practical to manufacture because there is no market, or too difficult to produce, but a patent has to be a real, not just a idea.

Its also true that a patent may be more of a protective patent than one you intend to start producing next week. Companies spend huge amounts on research, and if they make a patentable discovery or design, a patent will protect their design. They might even sell or trade rights to it if it is in their best interest.

 

[Rocky]

You cannot patent a idea, that would be rejected immediately.

May be I have used the word "Idea" too loosely. The following example illustrate what I mean by "idea". There is a company holding the patent on trimming the accuracy of a smeiconductor product through the I/O pin of teh packaged product . The same company have never make it work right. After 2 years, the product was killed and that "idea" was never used agin by the company. But the company have the patent.

Better yet. the following case is almost patenting an real "idea":
The Gilbert Hyatt Patent
A patent on the microcontroller, predating the only two Intel patents related to the MCS-4, was granted to Gilbert Hyatt in 1990. This patent described the architecture and logic design of a microcontroller, claiming that it could be integrated into a single chip. This patent was later invalidated in a patent interference case brought forth by Texas Instruments, on account that the device it described was never implemented and was not implementable with the technology available at the time of the invention. Actually Gilbert Hyatt recieved huge pay off before the patent is invalidated.