Canon still pursuing Foveon-style multi-layer sensor design

okay yabokkie-san

Kodak didn't fail because it missed the digital age. It actually invented the first digital camera in 1975. It brought to market the first professional camera, the DCS-100, in 1991. The QuickTake had an Apple label but were produced by Kodak in 1994. In 2001 Kodak held the No. 2 spot in U.S. digital camera sales, behind Sony. Over the years, Kodak put billions of dollars behind digital imaging. But ultimately hit an obstacle it couldn't overcome. At its core, it wasn't a camera or consumer electronics company. Companies can change what they do, but they rarely can change who the are.

"First" locally buffered CMOS sensors? seriously?
"First" DSLR with live view? The award goes to Olympus in E-10
"First" Practical stabilization system? The award goes to Nikon. On a lens with Nikon patent, yeah maybe to Canon
USM motor with tons of jargon tossed in- yeah maybe to canon

To continue with your "or"s
Or focusing on movie/video than still cameras in the last 2-3yrs
Or still using rebranded 5yr old sensors in cosmetic upgraded bodies
Or having ways for its users to justify new camera upgrades are worth while
I suppose.. those take talent.

(Note: I hope they seriously update this foveon tech and come up with a practical solution, but until then aptima or sony or samsung or panasonic or.. scores of others will continue to trounce canon' sensors)

@Joseph - I think they only did that for USM. And everything you mention is from the older era not the recent sit back and milk and wait.

Google CANON ION.
The first camera that made me dream of a world without film.

Canon LIDE
First slim scanners that worked on just a USB cable (great low-power tech).

Canon Bubblejet
Dominated the inkjet printer market for a while; probably still do; though I'd hate to give credit to printer companies for ripping us all off and adding to the monumental amount of e-waste as people opt for new printers rather than fixing up their dried-up printers or paying for overpriced ink...

Regardless, Canon innovates and does it well.

Odd, then, that several of the worlds most respected pros describe the new Foveon sensor results as "spectacular" and even compare it to medium format. I've personally seen similar responses and results. Maybe your experience is with the older designs.

@cinemascope
agreed. processing a nicely exposed Foveon RAW file, just takes an increase in saturation and I'm done. with conventional cameras though, i'll just never get what i want since I'm limited to Lightroom and I have no mastery with it.

Strip away the arrogance and this is like saying folks like Velvia but that's because they don't know how color is supposed to look.

The only time there is a "kaleidoscope" is when you are looking at an image with a lot of colors. The camera tracks colors better than most:

http://www.flickr.com/photos/kigiphoto/8729980521/sizes/o/in/set-72157633459763493/

I've also done a number of model shoots with other photographers, and the Sigma ends up being more accurate than most other cameras in the resulting shots.

yeah it would be baffling if canon does that but i dont think its stupid for leica to do it.

Monochrome sensors can provide a level of detail unheard of when using bayer layout sensor, because there is no guesswork involved when render the raw.
Phase One also makes monochrome backs. These are used for scientific applications as well as in some rental studios.

Conventional sensors are monochromatic. They just add the color filters on top of it.
Foveon/Merrill sensors can be monochromatic with the use of the latest SPP raw editing software. Search/learn about them if you're into BW.

It does because you lose luminance resolution otherwise. It takes a lot of processing power to drive since you need to collect 3x from each site instead of once. Now that MP counts are starting to get pretty high for APS-C sensors and processor speeds are getting up there it might make sense to think about going there in a few years though. They will be able to drive say 30*3 counts fast enough and that would make quite an APS-C body, 30MP foveon (90 Bayer MP worth of data to read).

I cannot agree with Larry that multi layered sensors don't need to have that many pixels, as the larger the number of pixels, the better the definition, regardless of how many colours in the pixels. Personally, I am hoping for the foveon type sensor with about 500 mpx at a bit bigger than 645 sensor size, and lenses capable of using all that resolution. Nikon and Canon already have lenses so fine at resolution, that you can use them to print circuit boards. what we want is that kind of detail, but not the £250,000 price ticket they have. I want to print 12 foot wide prints at fine detail, so I could view them at book-reading distance, and still marvel at the individual veins on leaves. It is possible to do, but it will probably be a new company, or one which is at the edge now, but willing to go quality instead of mass-production standards.

So cameras that are superior in every way outside of resolution are a third of the DPs? The DPs are good cameras, but they're the very definition of niche cameras.

Is Canon lagging behind sensors Nikon designs, or which Nikon purchases elsewhere?

a 18M stacked pixels can resolve way more than 18M color filters.

Foveon is crap. That's why nobody uses it. It's not a secret sauce only Sigma has access to. It's a sauce nobody wants on their buns.

Doesn't mean no future multi-layer sensor design will ever be good. Just means Foveon is still crap. That's why it needs research.

@raincoat: I doubt you'd say that if you really know anything about sensors. Sigma's current Foveon sensor is pretty darn good. It's hard to top it if you're into portraiture and landscape photography.

I suggest you look at some images taken with SD-1 or one of the DP Merrill cameras. Open your eyes and you'll see better.

This may surprise you but the "fake filter innovation" you mention actually does have its benefits.

They didnt buy it cause they want different approach. Sigma clearly showed that it doesnt work as intended.

Unless you can live with base ISO camera. Which I can. Just 99% of photographers cant.

The current Sigma cameras are far from a base ISO camera, as this ISO 3200 shot shows:

http://www.flickr.com/photos/kigiphoto/8515712883/

And with current raw processing having a dedicated B&W pipeline, if you like B&W it can do even better as in this ISO 6400 shot:

http://www.flickr.com/photos/kigiphoto/8509990735/

So they tweaked it 4 times, show me how much of an improvement that has been i dare you. And don't try to pass off jpeg processing as sensor performance. Research before you type in the future huh... (ps. i wasn't picking a fight, the opposite in fact as this is GOOD news, but congratulations on turning it around.)

Yes the 4th gene 18MP sensor is pretty much the same as the 1st gen. There has not been any noticeable change or improvement in that sensor

Foveon stopped having National fab their sensors and switched to Dongbu Electronics way back in 2006. (There was all sorts of interesting talk at the time, including allegations of arson and organized crime involvement).

Intel, National, and Dongbu were just people Foveon had fab chips. They don't own the IP any more than print shop owns the rights to the books they print.

How do you know Sigma is not making any money? They have had quite a lot of success with the DP cameras so far (old and new), and the company is private so you have no way of knowing actual sales.

There was a lot of layered sensor art before Foveon. Sigma's patents are all about the diode structure they use in an effort to lower costs, and the use of the penetration depth vs. wavelength of photons in silicon to do the color separation.

Neither of these things turned out that well. A company with more resources, interested in layered sensors, would do something different.

"Neither of these things turned out that well"

http://www.flickr.com/photos/kigiphoto/8729980521/in/set-72157633459763493

http://www.flickr.com/photos/kigiphoto/8420816763/in/set-72157632628303006

Whatever man.

What's the attitude, Kendall?

I mentioned two things, lowering costs and color separation. Cost wasn't just a failure, it may have been "the" failure, the one that sank Foveon. They used to market their sensors very aggressively to the P&S and cell phone markets on purported cost savings that never materialized. That's the main reason they never got appreciable wins in those fields.

Every last Sigma model has launched at a significantly higher price than any color filter array camera in their class. It's not always been as insane as the launch of SD-1 at $10,000, but the Foveon sensors have never been cost competitive.

And the color you tout is achieved at the cost of an insane amount of math and guesswork. You get good results, now, because you've put a lot of effort into getting there. The average photographer wants good color because of the camera he uses, not "in spite of" it.

Foveon has its strengths, but neither cost nor efficient color separation are among them.

I don't think cost was what prevented Foveon from having greater success. From the stories we hear it was a lot more about other sensor makers (*cough* Sony *cough*) making sure Foveon could not sell to any major players. You can't win if the game is rigged. Lucky for all of us Sigma kept Foveon on the board.

Good point. The AVERAGE photographer wants good color out of the camera? All photographers want good colors out of the camera. So you are saying that there isn't an insane amount of heavy guesswork and math going into those bayer mosaic pattern cmos sensors? I can look at color charts at all the testing sites and see that the Sigma has decent color accuracy. So why is it that bayer sensors get a pass all the time and something is inherently wrong with the Foveon sensor? Have you even every shot with a Sigma SD1M DSLR or DPMerrill series compact camera? I'd like to know because you sure have some very strong opinions that it isn't color accurate enough and myself and others that shoot with it---I even use these to photograph works of art for museums and art catalogs and haven't found ANY problems with color accuracy--EVER.