PetaPixel

Canon Patents a Multi-Layer Image Sensor with Five Layers Instead of Three

70dsensor

Remember the advent of multi-blade shaving razors? Single blades gave way to two blades and then three, four and five. There are even a six+ blade razors out there in case you want a REALLY close shave. Well, a fresh Canon sensor patent has us wondering if the same thing isn’t happening to digital sensors.

Canon has patented a Foveon-like sensor before — a type of sensor, initially implemented by Sigma, that uses individual layers for detecting red, green and blue — but the company’s newest multi-layer sensor patent shows not three, but five layers.

Say hello to the Gillette Fusion of multi-layer sensors: a Foveon-like creation that sports individual layers for red, green, blue, UV and IR detection.

patent

This is said to enhance color reproduction, particularly for skin tones which, as we all know, is pretty important in photography.

Originally found/reported on by Northlight Images, the site points out that this type of design might cause more problems than it solves. The site maintains that the additional layers would “cause issues for current lens designs,” as well as “require a major re-write for support from RAW converter software.”

More trouble than it’s worth? It’s probably wise to refrain from too many predictions while this idea is still in patent form, but feel free to drop your opinions in the comments down below.

(via Northlight Images)


P.S. By the way, general wisdom dictates that a single blade razor still offers a better shave, and where replacements for Gillette’s 5-blade razor system cost $25+, a single razor blade costs less than a dollar. We’ll let the analogy sink in.


 
Get the hottest photo stories delivered to your inbox.
Get a daily digest of the latest headlines:
  • Felipe Manga

    No need to hack cameras for IR photography! o/

  • Renato Murakami

    Regarding the PS – poor analogy. Never make them if you are not absolutely shure about how closely both relate. Completely different functions, with completely different levels of complexity.

    In any case, as with several other “innovative” experiments on light sensor technology, we’ll only know when a working prototype is out there. Too many variable factors without a proper way to measure to judge.

  • Jonathan Maniago

    Sounds like they’re sacrificing dynamic range for higher resolution (i.e. no need for demosaicing). Perhaps this is Canon’s reply to Sony’s 36MP sensors?

  • Brian Calabrese

    7….minute.. ABS

  • DLCade

    It was meant more as a touch of humor than anything else @renatomurakami:disqus, obviously complex sensor technology can’t be equated with shaving razors : )

    For what it’s worth though, I feel like it applies on a basic level. Adding further complexity, as Northlight pointed out, can cause more problems than it solves. The three-layer Foveon sensors have had to deal with their fair share of issues; it seems to me that adding another two layers will address some while exacerbating others.

  • 322131

    this will not be a dslr sensor.
    all raw converter would need an update.
    todays lenses will no work well with such a sensor.
    this is for compact cameras or maybe surveilance.

  • 322131

    you are clueless. DR will not be scarificed.

  • lexplex

    This is the same basic principle as colour film photography, so why wouldn’t it work with a DSLR?

  • Henry Wang

    LOL is this going to be the new megapixel race? the sensor layer race?

  • PazinBoise

    On the flip side new ideas, while they can potentially lead to new issues, can also lead an industry as a whole to think differently and which may lead to better products. Of course form a business standpoint I wouldn’t expect to this happen all at once, at least from a Canon or Nikon.

  • arachnophilia

    raw converters need any updated ANY time a new camera is released.

    lenses will work, the question is how well.

  • arachnophilia

    depends on how thick the stack is. if it’s as thin as a sheet of film, cool. if it’s deeper, you get fun chromatic aberrations.

  • http://inthemistphoto.com/ InTheMist

    Hah! I’m patenting C, M, Y, K, R, G, B, IR, UV!

    Back off, I will sue!

  • Guest

    This was the first thing I thought of when I read the article

  • Danilo

    The K isn’t necessary. :)

  • Salvador Peña

    IR photography capable sensors?

  • Matt

    Nice. Love to see it come to market. This may make me a customer again.

  • Emily S.

    If you apply the UV and IR data in the correct manner, you can extrapolate a lot of detail about the image and reduce noise in INCREDIBLY low light situations. Also, it’s possible that canon would do all the converting in their firmware so that your raw files look the same.

  • Emily S.

    What you said is essentially the opposite of true. This technology uses 5 times more pixels, but it doesn’t generate an image with more pixels. If each sensor were 20MP then you’d be using 100MP worth of light and color data to make a 20MP file. This could dramatically increase DR and Color bit-depth.

  • Emily S.

    Not if canon converted the UV and IR data into a normal RAW file in their firmware.

  • Emily S.

    nice eye ;)

  • Adrian S

    So no more UV filters? I was under the impression that UV is not good for cameras. I’m also very curious if the RAW image will give out this data and how the world would look in UV.

  • Disqus is an NSA shill

    strange… UV and IR? If there’s an additional layer that is sorely needed in todays CMOS then its straight luma b/w one. I have a hunch that would do wonders for sharpness and definition.

  • Don

    Even if you don’t use the IR or UV directly, getting those spectral samples can, in theory, get you a better estimation of the full visible spectrum, if you interpolate a smooth curve through all five samples.

  • Matt

    I get the feeling that there will be a lot of new things we can get with the additional data. It will be fun to see the advancements.
    But, I don’t see a need for the raw files to be the same. It could be a while before Adobe cathces up, but in the long run so what… I’d like to have the actual raw data so that as we learn and develop new procedssing techniques they could be applied to existing photos.

  • JoeNoName

    well yes, Sony is releasing theirs at the end of this year or begining of next

  • RenderOne

    Who cares if all RAW converters need to be updated? That should be no reason not to progress technology. That would be like saying “Digital Cameras are going to be more trouble than they’re worth because now instead of darkrooms, we need to use computers, and who can deal with that kind of an update?”

  • Zos Xavius

    So they are making their own version of the Foveon. Interesting. I wonder how they will overcome the same problems and limitations the foveon sensors have shown in regards to sensitivity.

  • Alan Klughammer

    CMYK is a subtractive colour system, RGB is an additive colour system. they are completely different. You cannot have an RGB printer, just like you cannot have a CMYK backlit monitor.
    Each layer is b&w. it is just that only certain wavelengths penetrate the sensor to the depth they are measuring. The Raw conversion assigns a colour value…

  • D.G. Brown

    My thoughts exactly! Plus, I would love to get images with all five channels. Just think of what rainbows would look like!! :) (I have an EOS M that is converted to IR and I’ve been meaning to make composite rainbow photos, but it requires two cameras and a bit of work)

  • Robert Mark

    The images from the Foveon sensors are remarkable. I hope Canon has the guts to offer this technology in a mainstream body.

  • Mark Davidson

    Massive amounts of money often does the trick.

  • Jonathan Maniago

    My understanding of sensors using traditional color filtering (ex. Bayer arrangement) is that something like a 16MP sensors is consisted of 4MP of reds, 4MP of blues, and 8MP of greens. The RAW decoder then uses demosaicing algorithms to generates what it assumes a full 16MP image would look like.

    What Sigma’s Foveon sensor and this patent from Canon does is capture full RGB data (and then some) at every single photosite. Since demosaicing is not longer required, this could lead to more accurate colors, and a 16MP sensor could produce a true 16MP image.

    What I’m worried about is that stacking multiple color filters reduces the amount of light received particularly at the lower layers. Historically, Canon’s sensors don’t perform as well as Nikon’s/Sony’s in low light situations. Switching from traditional Bayer arrangement to something like Foveon’s could exacerbate this issue.

    http://en.wikipedia.org/wiki/Foveon_X3_sensor#Light_gathering_and_low-light_performance

  • Sheila Cain

    Asher . even though
    Leslie `s c0mment is really cool… on monday I bought a great new Aston
    Martin DB5 from earning $9990 this month and-also, 10/k this past month .
    no-doubt about it, this really is the most-rewarding Ive had . I actually
    started 6 months ago and immediately began to make minimum $69 per/hr . view
    wℴℛks77.ℂℴ­­­ℳ

  • cameraFan

    You’re not understanding at all. Each pixel in a Bayer arrangement has 2 green, one red and one blue. sub-pixel. That means a 16Mpix sensor actually has 16×4 sub-pixels, and 3/4 of blue and red and half of green is thrown away. The new arrangement is very differnet. There is no colour filter. It relies on different wavelengths of light penetratig to different depths, so all the light can be used. The challenge is cross-talk between colours. It requires a lot of processing, which results in errors when signals are noisy at low light. Hence the orignal Foveon idea had poor high ISO performance.

  • kenburkett202

    my
    roomate’s step-aunt makes $67 /hour on the internet . She has been fired for
    8 months but last month her pay check was $14347 just working on the internet
    for a few hours. Read More Here

    ☭☭☭ ☭�☭☭ ☭☭☭ ☭�☭☭ ☭☭☭

    w­w­w.ⓦⓞⓡⓚⓢ77.c­o­m

  • Zos Xavius

    Its not an easy problem to solve. As you introduce layers to a sensor you also reduce its sensitivity. You can’t get around the physics of light.

  • Zos Xavius

    I don’t think you understand. Those aren’t subpixels. Those are the actual pixels. A 16M chip only has 16M imaging sites. The color is interpolated through the bayer filter.

  • Zos Xavius

    This further illustrates what I’m talking about. This is what an image looks like off a bayer sensor before it is interpolated. Notice how green it is. That is because there are 2x as many green pixels as red and blue pixels.

  • ldfrmc

    “This is said to enhance color reproduction, particularly for skin tones which, as we all know, is pretty important in photography.”

    Spots and wrinkles on skin will be easier to remove. Not so with all the other things, besides faces, that are in photography.

  • jon

    It’s the biggest leap in photography since the inception of professional digital sensors. It’s a pretty big deal.

  • cameraFan

    Huh, you’re right. I always understood Bayer wrong! Good to learn something new.

    That being said, my second point is still right. I don;t believe anyone is talking about stacking colour filters. Bayer throws away half of the green light (in the red/blue filters) and 3/4 of each of blue and red (in the green/red and blue/green filters respectively). In theory, if the multi-layer cross-talk issue could be solved, the sensitivity should take a leap by 2-4x just by nature of not needing to cut anything out, and simply relying on the different depth penetration of different wavelengths.

  • Lefteris Heretakis

    Traditional film technology had 2 layers. Most Canon lenses are film lenses that already work on 3 layer technology. I dont see how that woud case problems.