Non-Bayer arrays; how is Fuji the only one to have thought of this?

[ScottyP]

Is it me? How is it that most manufacturers are using the simple Bayer pattern on most of their sensors, and everyone has the same image issues as a result, and everyone band-aids it with the clarity-robbing AA filter.

There is something almost funny that Fuji is able to "stun the world" by going with a slightly more complex pattern. It just seems amazing that the others have not jumped on that a long time ago. It's not like Fuji, or anyone, could patent the whole idea of doing something, anything, other than Bayer.

Sony is mucking around with a 4th color in the RGB, I read. They and others will be innovating on all elements of the sensor and smaller manufacturing processes. Could something as simple as "non-Bayer" give Canon an easy lower-tech leg up against competition? Is that some low-hanging fruit Canon could pick?

 

[pharp]

Not exactly true - Sigma has the Foveon. The SD9, not only had no bayer filter, but no AA filter or MICROLENSES! In theory, this is better, in reality, not so much.

 

[ScottyP]

Not exactly true - Sigma has the Foveon. The SD9, not only had no bayer filter, but no AA filter or MICROLENSES! In theory, this is better, in reality, not so much.

Very true. I meant to reference Foveon, and then distinguish it by saying it is a lot more complicated an affair than just "randomizing" color receptors.

As I look at the title of my post, I realize I was overbroad and I overgeneralized. My point was just that the shuffling colors thing seems so easy. And it probably isn't, and if not, I was wondering if anyone knows why I am looking at this wrong.

 

[ChilledXpress]

Is it me? How is it that most manufacturers are using the simple Bayer pattern on most of their sensors, and everyone has the same image issues as a result, and everyone band-aids it with the clarity-robbing AA filter.

There is something almost funny that Fuji is able to "stun the world" by going with a slightly more complex pattern. It just seems amazing that the others have not jumped on that a long time ago. It's not like Fuji, or anyone, could patent the whole idea of doing something, anything, other than Bayer.

Sony is mucking around with a 4th color in the RGB, I read. They and others will be innovating on all elements of the sensor and smaller manufacturing processes. Could something as simple as "non-Bayer" give Canon an easy lower-tech leg up against competition? Is that some low-hanging fruit Canon could pick?

Wow... you should start your own company! I'll buy your first two cameras

 

[The_Arsonist]

I would still be shooting Fuji if they had had a cheaper DSLR model when I got my used 20D. I still am kicking around the idea of selling my standard focal length lenses and getting an EX1 with a 35mm

 

[EdB]

Not exactly true - Sigma has the Foveon. The SD9, not only had no bayer filter, but no AA filter or MICROLENSES! In theory, this is better, in reality, not so much.

In what reality doesn't the Foveon work? Have you seen the images the DP2M produces, the detail that camera produces is astonishing. Especially for the price.

 

[Nishi Drew]

Not exactly true - Sigma has the Foveon. The SD9, not only had no bayer filter, but no AA filter or MICROLENSES! In theory, this is better, in reality, not so much.

In what reality doesn't the Foveon work? Have you seen the images the DP2M produces, the detail that camera produces is astonishing. Especially for the price.

By what I've seen and understand the Sig sensors are phenomenal, while the cameras themselves are nothing to even bother considering. THOUGH, if they made an FF Foveon, and considering how they're glass is these days???

Also Fuji, with the rumor of an X200 coming out I'm excited to see what that's like. I suggested to and got a friend into buying a used X100, and man I get a little jealous whenever it gets pulled out, and straight from the camera the Jpegs have nice colors and the bokeh is nice, although I found F/2 to be quite soft when up close, but that's about it.
We're supposed to be talking about sensor tech here, yes, I guess the reason for the big boys Canikon haven't really experimented in public with new designs is because professionals want to stick with what works, like if Canon all of a sudden decided to finally add EVFs on all their DSLRs, that's too much of a change that pros won't feel comfortable adapting to.

 

[ScottyP]

Is it me? How is it that most manufacturers are using the simple Bayer pattern on most of their sensors, and everyone has the same image issues as a result, and everyone band-aids it with the clarity-robbing AA filter.

There is something almost funny that Fuji is able to "stun the world" by going with a slightly more complex pattern. It just seems amazing that the others have not jumped on that a long time ago. It's not like Fuji, or anyone, could patent the whole idea of doing something, anything, other than Bayer.

Sony is mucking around with a 4th color in the RGB, I read. They and others will be innovating on all elements of the sensor and smaller manufacturing processes. Could something as simple as "non-Bayer" give Canon an easy lower-tech leg up against competition? Is that some low-hanging fruit Canon could pick?

Wow... you should start your own company! I'll buy your first two cameras

@Chilly: Glad you were able to avoid the knee-jerk reaction on that one. After all, I did ask a question that could, with some effort, be viewed as obliquely questioning Canon, and I was therefore asking for it.

 

[well_dunno]

I, for one, am considering to get the XE-1 with 18-55 f/2.8-4. Have been looking at the pocketable and mirrorless options as a carry around camera. Even though not the pocket, the bag is big enough to hold it so no probs.

By the way, does anyone know whether the issue around third party raw conversion support has been resolved? To my knowledge, in cam jpegs come out best followed by silkypix that come with the cam. Any other raw converter is having trouble with the images though...

Also a 55-200 to appear in early 2013 if I am not mistaken...

Cheers!

 

[Mt Spokane Photography]

Your theory that others have not used different patterns is contradicted by the facts. Do a little research before making such claims.

There are many Bayer type patterns patented by Kodak. The alternate patterns have not worked out to be popular. http://en.wikipedia.org/wiki/Bayer_filter
Fujifilm is the new kid on the block to try alternate color filters, others who tried them did not find success.

There was the CYGM pattern used by Canon, Nikon, Sony, and Kodak in the late 1990's.
http://en.wikipedia.org/wiki/CYGM_filter
http://www.dpreview.com/news/1999/9/7/sony3megapixel
The CYGM filter is far less common than the Bayer filter. CCDs that use it include the 3 megapixel Sony ICX252AK and ICS252AKF (which sampled in October 1999[1]).
Cameras that use it include several Canon models of the 1999-2000 period, such as the Canon PowerShot S10,[2] the original Canon Digital IXUS (June 2000),[3] though subsequent IXUS models used the Bayer filter, and the Canon G1; the Kodak DCS 620x and DCS 720x DSLRs, and several Nikon Coolpix models.[4]

Sony introduced the Cybershot DSC-F828 with a RGBE pattern in 2004.
http://www.dpreview.com/news/2003/7/15/sonyrgbeccd
http://en.wikipedia.org/wiki/RGBE_filter

Fuji has at least two alternate patterns that they have patented, but how much is sales hype and how much is performance related is yet to be seen. There will be advantages and disadvantages.

 

[paul13walnut5]

I have often argued for a move to the 3sensor array as traditionally found on tv cameras.
This means a true colour value at every pixel, but it also would mean 3x the sensor cost, and the cost of a very large dichroic prism and assembly.

It would also make for a larger camera.

When you have mote than one sensor you can employ 'pixel shift' where the green sensir is offset by half a pixel, with massive gains in apparent resolution.

There is some precedent, Minolta had a multi sensor camera in the very early days (the rd-175, although the sensor wasn't split straight rgb) though very costly and more the size of a mf body.


As the Tv world has been quick to embrace single large sensor cameras with bayers, maybe the stills world could steal a trick from the tv camera world (sony and panasonic already manufacture 3ccd and 3cmos arrays)

I doubt any such camera would be intended for anything other than very high end professional or industrisl use, but like f1 cars, the lessons learned at the sharp end make better products for the consumer, over time.

 

[EvilTed]

Fuji have just bought a lot of patents from Kodak

ET

 

[SDsc0rch]

here's a bigger question... WHY even have these "bayer" filters - sensors made of an array that detect single colors that require "interpolation" (and thus a bunch-o-math that eats up cpu processing power)

imagine an array made of sensors that can detect "all" colors

problem solved!

(non-scientist here - but not afraid to ask "why not?!")

 

[pharp]

Not exactly true - Sigma has the Foveon. The SD9, not only had no bayer filter, but no AA filter or MICROLENSES! In theory, this is better, in reality, not so much.

In what reality doesn't the Foveon work? Have you seen the images the DP2M produces, the detail that camera produces is astonishing. Especially for the price.

Please read posts first - no one said it didn't work. Yes I actually have seen images from DP2M. I tried a friends SD9 - it worked fine, but whether you agree or not - it just wasn't better [different] in my view, in some ways worse. It may or may not be the future, it just isn't there yet. If it was a big improvement - somebody, other than Sigma would have used - not aware of any. As noted, the camera was awful - no in-camera JPEGs, two kinds of batteries, proprietary mount [should have licensed from someone like Nikon, Pentax [close], etc) - ugh. Especially for the price - in what reality is that true!?

 

[pharp]

here's a bigger question... WHY even have these "bayer" filters - sensors made of an array that detect single colors that require "interpolation" (and thus a bunch-o-math that eats up cpu processing power)

imagine an array made of sensors that can detect "all" colors

problem solved!

(non-scientist here - but not afraid to ask "why not?!")

?? Google Foveon

 

[SDsc0rch]

?? Google Foveon

ah

interesting..

 

[well_dunno]

As per the rumors, Capture One has a beta version that supports Fuji's X-trans sensor. That Capture One was working on it I had heard but not that there was a beta version being tested - so says a fujirumors reader:

http://www.fujirumors.com/feedback-capture-one-beta-x-trans-potential-unlocked-2/

Cheers!

 

[Mt Spokane Photography]

here's a bigger question... WHY even have these "bayer" filters - sensors made of an array that detect single colors that require "interpolation" (and thus a bunch-o-math that eats up cpu processing power)

imagine an array made of sensors that can detect "all" colors

problem solved!

(non-scientist here - but not afraid to ask "why not?!")

Invent one and wealth will be yours. For now, sensors only detect luminance (brightness) Thats why the Bayer type filter is used, it allows a color image to be constructed from a B&W image.

 

[ScottyP]

Your theory that others have not used different patterns is contradicted by the facts. Do a little research before making such claims.

There are many Bayer type patterns patented by Kodak. The alternate patterns have not worked out to be popular. http://en.wikipedia.org/wiki/Bayer_filter
Fujifilm is the new kid on the block to try alternate color filters, others who tried them did not find success.

There was the CYGM pattern used by Canon, Nikon, Sony, and Kodak in the late 1990's.
http://en.wikipedia.org/wiki/CYGM_filter
http://www.dpreview.com/news/1999/9/7/sony3megapixel
The CYGM filter is far less common than the Bayer filter. CCDs that use it include the 3 megapixel Sony ICX252AK and ICS252AKF (which sampled in October 1999[1]).
Cameras that use it include several Canon models of the 1999-2000 period, such as the Canon PowerShot S10,[2] the original Canon Digital IXUS (June 2000),[3] though subsequent IXUS models used the Bayer filter, and the Canon G1; the Kodak DCS 620x and DCS 720x DSLRs, and several Nikon Coolpix models.[4]

Sony introduced the Cybershot DSC-F828 with a RGBE pattern in 2004.
http://www.dpreview.com/news/2003/7/15/sonyrgbeccd
http://en.wikipedia.org/wiki/RGBE_filter

Fuji has at least two alternate patterns that they have patented, but how much is sales hype and how much is performance related is yet to be seen. There will be advantages and disadvantages.

Maybe I was too sweeping in my generality, but note I asked why only Fuji is currently pushing a non-bayer pattern as "the next great thing" and the camera magazines seem willing to trumpet it as a new thing also.

The history of pixel designs you describe is all factually correct, but you name old/discontinued/minor/failed cameras, and also patents that have been filed, but not new products with fresh R&D dollars and fresh advertizing dollars behind it. Other than Fuji (and Foveon in a radically different sense) at this particular moment. What I thought would make for an interesting discussion is whether Canon could take the idea of a simple non-bayer pixel pattern and eliminating the AA filter for a major camera sensor line, (not just for a special limited edition version of an existing camera where they leave the AA filter off).

Is there enough potential to a more complex/more "random" pixel pattern for Canon to make something of it?

 

[PVS]

Sony recently released pattern for sensor with photosites aligned in honeycomb grid.
So, besides Foveon and Xtrans there's always more to come.